ProgrammableMarketing/godela.ai (opens in a new tab)
PDF

Views

  • Plan
  • Roadmap
  • Research

On this page

  • The brief
  • Start here
  • Idea portfolio
  • Effort vs impact
  • This week · 7
  • This quarter · 11
  • Big bets · 3
  • Wildcards · 2
  • How we can help
Generated Nov 10, 2026
  1. Programmable
  2. Playbooks
  3. Godela
Godela logo

growth playbook · Nov 10, 2026

godela.ai (opens in a new tab)

Godela growth playbook

Diagnosis

Godela is a roughly ten-person YC X25 startup with strong technical claims: a car aerodynamics run cut from 2.5 hours to 1.8 seconds, a first contract to replace an incumbent simulation tool, and over 200 inquiries after launch. It is hiring three founding engineers. The website has not kept up. godela.ai is a single page whose "Contact us" and two "Request a Pilot" buttons all open a founder's personal calendar.

The site has no page on how customers' CAD and simulation data is handled, no privacy policy or terms, no accuracy figure beside its "solver-level accuracy" claim, no application pages, no sitemap and no customer proof. Competitors show named case studies and deployment partners. This playbook first makes the pilot path trustworthy and specific: a pilot page that answers the data question, a qualifying intake, a published validation number and small fixes on the page.

It then builds application pages, a public benchmark, an anonymised first case study and better pilot packaging. The big bets are a live public demo, a packaged data center thermal product and working inside engineers' existing simulation tools.

Growth Score

6/10

Biggest Opportunity

Turn interest into pilots that pass a security review. Engineers already reach out; what stops a pilot is an unanswered question about where their CAD and simulation data goes and how accurate the model is.

Biggest Risk

Well-funded rivals (Neural Concept, PhysicsX, Monolith) already show named enterprise customers and deployment partners. If Godela's proof stays in investor posts rather than on its own site, buyers will default to the vendor that looks safer.

01

Start here

The three moves to make first. Each one is written up in full below.

  1. 1

    Give "Request a Pilot" its own page that answers where a team's CAD and simulation data goes

    Write plain answers to five questions: where is uploaded data stored, who can access it, is it used for anything beyond the customer's model, how is it deleted, and where can the model run.

    High impact First result in 2-4 weeks
  2. 2

    Replace the personal calendar link with a pilot intake that asks for application, solver and data volume

    Write the six questions and the rule for what counts as qualified.

    High impact First result in 2 weeks
  3. 3

    Put a number next to "solver-level accuracy": publish the validation for the data center demo

    Pull the held-out error numbers for the data center model.

    High impact First result in 2 weeks

02

Your idea portfolio

23 ideas, grouped by when to run them. Open any idea for how to run it: the steps, who to approach, copy you can use, and how to measure it.

  • 8 ideas are marked Unexpected: plays you are unlikely to have tried.

Every idea placed by the effort it takes against the impact it could have. Quick wins sit top left; select a number to open that idea. The filters below narrow the list by status, timing, category and what Programmable can build.

Effort vs impactEach number is an idea. Select one to open it.
Impact →
Quick wins · 5Big bets · 6Fill-ins · 7Reconsider · 5
Effort →

Showing 23 of 23 ideas

This week 7

Quick wins with a clear first action.

  • 01

    Godela's whole product starts with "upload simulation data", and the homepage pushes one action: "Request a Pilot", twice. Both buttons, and "Contact us", open a founder's personal calendar. Nothing on godela.ai says what a pilot includes, what data Godela needs, where that data is stored, where the models run, or who can see them. There is also no privacy policy, terms or security page (/privacy returns 404). For a hardware team, CAD and simulation results are core intellectual property. The engineer who wants the pilot has to get it past IT security and legal, and right now has nothing to forward to them.

    High impactLow effort First result in 2-4 weeks CEO and Co founder

    First action

    Write plain answers to five questions: where is uploaded data stored, who can access it, is it used for anything beyond the customer's model, how is it deleted, and where can the model run.

    How to measure it

    Pilot request to signed pilot rate

    Now
    –
    Target
    +30% versus the last quarter
    By
    90 days

    Steps

    1. 1

      Answer the data questions

      Storage, access, training use, deletion and deployment options, as true today.

      CTO and Co-founder 2 days

    2. 2

      Define the pilot

      Scope, data needed, deliverables (validated model plus report), length and what happens after.

      CEO and Co-founder 1 day

    3. 3

      Write the page

      A /pilot page with the scope, data list and data-handling answers.

      CEO and Co-founder 2 days

    4. 4

      Add the legal basics

      A privacy policy and terms linked from the footer, reviewed by counsel.

      CEO and Co-founder 1 week

    5. 5

      Make the reviewer PDF

      A one-page security and data summary to download.

      CTO and Co-founder 1 day

    6. 6

      Repoint the buttons

      Both "Request a Pilot" buttons go to /pilot.

      Founding Engineer (SWE) 1 hour

    7. 7

      Track

      Requests, security questions raised and time to signed pilot.

      CEO and Co-founder monthly

    First result
    2-4 weeks
    Cost
    Founder time plus a few hours of legal review
    Owner
    CEO and Co founder

    Smallest test

    Send the one-page data summary to the next three pilot prospects and record how many security questions come back.

    Why it works

    Publish a /pilot page and point every "Request a Pilot" button at it. It covers: (1) what a pilot is: one application, the team's existing runs (the homepage demo uses 2,400 simulations and 12 parameters), a validation report against held-out solver results, and a typical length; (2) what to send: file formats and how much data is enough; (3) data handling, written as plain answers: where data is stored, who can access it, whether it is used to train anything beyond the customer's own model, deletion on request, and options such as running in the customer's own cloud account if offered; (4) a one-page PDF version for security reviewers; (5) the pilot intake (idea 2). Link a short privacy policy and terms in the footer. Only state what is true today; where something is on the roadmap, say so.

    The bottleneck in enterprise engineering pilots is rarely the engineer's interest; it is the internal review that follows. A page that answers the data question up front lets the champion forward one link instead of waiting on a questionnaire, and it reads as maturity next to rivals that show deployment partners. It also turns a generic calendar booking into a scoped request.

    Ready to use

    ChecklistPilot page checklist

    [ ] What a pilot includes (one application, validation report, length)
    [ ] What to send: formats and minimum data
    [ ] Where data is stored and who can access it
    [ ] Whether data is used beyond the customer's own model
    [ ] Deletion on request
    [ ] Where the model can run
    [ ] Downloadable one-page summary for security reviewers
    [ ] Privacy policy and terms in the footer
    [ ] Both "Request a Pilot" buttons point here

    Risks and how to handle them

    • Promising data controls that are not built yet

      List only what is true today and label anything planned.

    • Too much detail scares off early-stage interest

      Keep the page short and put detail in the PDF.

    Evidence

    • Your websiteHomepage (checked November 9, 2026): headline "Godela is the frontier of AI for the physical world."; a Train, Explore, Optimize flow (upload simulation data and Godela trains a physics-constrained model); applications: Data Center Thermals, Electronics Cooling, Aerodynamics, Electromagnetics, Structures and Build Your Own. The "Contact us" button and both "Request a Pilot" buttons open the same personal Calendly page of a founder; the only other links are [email protected], LinkedIn and the YC launch post. There is no privacy policy, terms, security or data-handling page, careers, docs, blog, case studies or pricing, and nothing on how uploaded CAD or simulation data is stored or where the models run.Source
    • Your websiteSite structure (checked November 9, 2026): godela.ai is one page; /sitemap.xml, /robots.txt, /privacy and /careers return 404. Meta description: "Godela is bringing AI-powered physics intelligence to the physical world." The headline changed from "Godela" to "Godela is the frontier of AI for the physical world." after October 2025. No blog, newsletter, case studies, testimonials or video channel; LinkedIn is the only social profile. Domain registered February 2025.Source
    • News coverageY Combinator launch post: Godela turns engineering questions into physics-informed models that are faster and cheaper than traditional simulation, positioned as a layer on top of existing simulations rather than a replacement; engineers can ask in natural language or upload CAD files, experimental results or simulation data. The problem it names: weeks or months of simulation and parameter tuning to answer one question, so many questions never get asked. Its asks: introductions to engineers with simulation bottlenecks and partnerships with engineering teams.Source
    • Competitor researchRivals' own homepages: Neural Concept, an "AI-first engineering platform" with four named customer stories across automotive and aerospace and AWS, Microsoft and NVIDIA as deployment partners; PhysicsX, "Engineering. Reengineered.", serving five industries with a sailing-team case study and a "Talk to an engineer" button; Monolith, "Test less. Learn More.", claiming test plans cut by up to 70%, with testimonials from carmakers and aerospace firms, "Request a demo", and now part of CoreWeave.Source
  • 02

    "Contact us" and "Request a Pilot" lead to the same personal calendar, so a student, a recruiter and an aerospace thermal lead all land in the same founder slot. Godela reported over 200 inquiries after launch. At that volume, founder calls with unqualified leads cost time the founders need for pilots and hiring.

    High impactLow effort First result in 2 weeks CEO and Co founder

    First action

    Write the six questions and the rule for what counts as qualified.

    How to measure it

    Call-to-pilot rate

    Now
    –
    Target
    1 in 3 calls
    By
    60 days

    Steps

    1. 1

      Write the questions

      Application, solver, runs and parameters, formats, timeline, email.

      CEO and Co-founder 1 hour

    2. 2

      Set the rule

      Which answers get a call and which get a reply.

      CEO and Co-founder 1 hour

    3. 3

      Create a branded event

      A Godela scoping-call event on a company booking page.

      CEO and Co-founder 1 hour

    4. 4

      Build the form

      On /pilot, routing qualified answers to the event.

      Founding Engineer (SWE) 1 day

    5. 5

      Write the reply

      A helpful email for leads that are not ready yet.

      CEO and Co-founder 1 hour

    6. 6

      Log everything

      Every request in one sheet or CRM with its answers.

      CEO and Co-founder ongoing

    7. 7

      Review

      Which applications and solvers convert.

      CEO and Co-founder monthly

    First result
    2 weeks
    Cost
    $0-$20 a month
    Owner
    CEO and Co founder

    Smallest test

    Add the six questions to the existing booking event this week.

    Why it works

    Make a short pilot form (on /pilot from idea 1) with six questions: application (from the six on the homepage), current solver, number of existing runs and parameters, file formats, decision timeline and company email. Qualified answers go to a Godela-branded booking event for a 30-minute scoping call; everyone else gets a reply with the pilot page and an example. Keep "Contact us" as email for general questions.

    The questions are the ones the first scoping call asks anyway, so founders arrive prepared. A branded event also looks more established to enterprise buyers than a personal link.

    Risks and how to handle them

    • A form adds friction for strong leads

      Keep it to six questions and show the calendar straight after.

    Builds on

    Evidence

    • Your websiteHomepage (checked November 9, 2026): headline "Godela is the frontier of AI for the physical world."; a Train, Explore, Optimize flow (upload simulation data and Godela trains a physics-constrained model); applications: Data Center Thermals, Electronics Cooling, Aerodynamics, Electromagnetics, Structures and Build Your Own. The "Contact us" button and both "Request a Pilot" buttons open the same personal Calendly page of a founder; the only other links are [email protected], LinkedIn and the YC launch post. There is no privacy policy, terms, security or data-handling page, careers, docs, blog, case studies or pricing, and nothing on how uploaded CAD or simulation data is stored or where the models run.Source
    • News coverageCLAI Ventures investment post (July 11, 2025): a car drag and pressure simulation that "takes 2.5 hours on Simscale runs in just 1.8 seconds on Godela"; Godela predicted the Navier-Stokes equation "with 99.99% accuracy from 72mb of fluid flow data"; a first contract signed to replace an incumbent simulation tool; over 200 inquiries after launch; target markets include drop simulation, thermal reentry and soft materials.Source
  • 03

    The homepage promises Godela "predicts system behavior with solver-level accuracy in seconds instead of days" but gives no error figure, test set or method. Engineers are trained to distrust unquantified accuracy. Godela already has the numbers: the open Founding Simulation Engineer role owns "validation of model outputs against simulation and experimental ground truth", and an investor post cites 99.99% on a fluid flow case.

    High impactLow effort First result in 2 weeks Founding Simulation Engineer

    First action

    Pull the held-out error numbers for the data center model.

    How to measure it

    Accuracy objections per call

    Now
    –
    Target
    Halved
    By
    60 days

    Steps

    1. 1

      Pick the test set

      Held-out solver runs for the data center case.

      Founding Simulation Engineer 1 day

    2. 2

      Compute errors

      Mean and worst-case error on the outputs shown.

      Founding Simulation Engineer 1 day

    3. 3

      Write the method note

      Data, split, metric and limits, in one page.

      CTO and Co-founder 1 day

    4. 4

      Add the box

      Validation box under the demo, linked to the note.

      Founding Engineer (SWE) 2 hours

    5. 5

      Repeat for aerodynamics

      The car drag case if shareable.

      Founding Simulation Engineer 1 week

    6. 6

      Use on calls

      Open scoping calls with the validation.

      CEO and Co-founder ongoing

    First result
    2 weeks
    Cost
    Engineer time
    Owner
    Founding Simulation Engineer

    Smallest test

    Add one line under the demo: test runs, mean error, worst error.

    Why it works

    Under the data center demo, add a short validation box: how many held-out solver runs were tested, error on peak temperature (for example mean and worst-case in °C), the time per answer versus the solver, and a link to a one-page method note. Do the same for the car drag case in the investor post if it can be shared.

    A quantified, honest error bar is more persuasive to engineers than "solver-level", and it pre-answers the first question on every scoping call.

    Risks and how to handle them

    • Worst-case errors look bad

      Show them with context; engineers trust candour more than perfection.

    Evidence

    • Your websiteHomepage demo panel, "Data center cooling optimization": "2,400 simulations · 12 parameters", "inference: 0.8s", "4,200 configs evaluated · 18 min · peak temp ↓ 10.1%", peak temperature cut from 72.1°C to 64.8°C; the prompt asks for "the top 5 configurations ranked by cooling efficiency" but the answer lists three. The page says Godela "predicts system behavior with solver-level accuracy in seconds instead of days" but gives no accuracy figure and names no customers. Team: "Stanford and MIT trained physicists, ML researchers, and engineers"; job seekers are asked to email [email protected].Source
    • HiringFounding Simulation Engineer (YC job board, San Francisco, $120K–$180K, visa sponsorship): owns validation of model outputs against simulation and experimental ground truth; generates simulation data across CFD, FEA and multiphysics; builds pipelines that turn simulation and experimental data into training-ready datasets, with solvers such as Star-CCM+, FEniCS and OpenFOAM. Godela says it is building "the first Physics Foundation Model"; founded 2025, about 10 people.Source
    • News coverageCLAI Ventures investment post (July 11, 2025): a car drag and pressure simulation that "takes 2.5 hours on Simscale runs in just 1.8 seconds on Godela"; Godela predicted the Navier-Stokes equation "with 99.99% accuracy from 72mb of fluid flow data"; a first contract signed to replace an incumbent simulation tool; over 200 inquiries after launch; target markets include drop simulation, thermal reentry and soft materials.Source
  • 04

    In the homepage demo, the user asks "Show me the top 5 configurations ranked by cooling efficiency" and the answer lists three. The audience is engineers who check numbers for a living, and the product's promise is precision. A visible mismatch in the one demo on the site is a small but avoidable credibility cost.

    Medium impactLow effort First result in Same day Founding Engineer

    First action

    Change "top 5" to "top 3" or add two rows.

    How to measure it

    Visitors reaching the pilot button

    Now
    –
    Target
    +5%
    By
    30 days

    Steps

    1. 1

      Fix the count

      Five rows or a "top 3" prompt.

      Founding Engineer (SWE) 30 minutes

    2. 2

      Check the numbers

      Reductions, peaks and zone limits agree.

      Founding Simulation Engineer 30 minutes

    3. 3

      Check units

      W/°C, metres and seconds consistent.

      Founding Simulation Engineer 15 minutes

    4. 4

      Add a caption

      Say whether the panel is real output or illustrative.

      CEO and Co-founder 15 minutes

    5. 5

      Deploy

      Publish and check on mobile.

      Founding Engineer (SWE) 15 minutes

    6. 6

      Review quarterly

      Recheck whenever the demo changes.

      Founding Engineer (SWE) quarterly

    First result
    Same day
    Cost
    $0
    Owner
    Founding Engineer (SWE)

    Smallest test

    Change the prompt to "top 3" today.

    Why it works

    Either show five ranked configurations or change the prompt to "top 3". While there, check that every number in the panel agrees (the 10.1% reduction, 72.1°C to 64.8°C, the zone changes and "No zone exceeds 66°C" against the listed peaks).

    The demo stands in for the product. Making it internally consistent costs minutes and removes a reason for a sceptical engineer to stop reading.

    Risks and how to handle them

    • Fixing one number breaks another

      Regenerate the panel from a real model run.

    Evidence

    • Your websiteHomepage demo panel, "Data center cooling optimization": "2,400 simulations · 12 parameters", "inference: 0.8s", "4,200 configs evaluated · 18 min · peak temp ↓ 10.1%", peak temperature cut from 72.1°C to 64.8°C; the prompt asks for "the top 5 configurations ranked by cooling efficiency" but the answer lists three. The page says Godela "predicts system behavior with solver-level accuracy in seconds instead of days" but gives no accuracy figure and names no customers. Team: "Stanford and MIT trained physicists, ML researchers, and engineers"; job seekers are asked to email [email protected].Source
  • 05

    Godela is hiring a Founding Engineer (SWE), a Founding Simulation Engineer and a Founding Engineer (ML Systems) on its YC job page, with salary, equity and visa sponsorship. godela.ai does not link to any of them; it asks candidates to email [email protected], and /careers returns 404. Simulation engineers who find Godela through its site never see the roles, and every application lands in the same inbox as customer requests.

    Medium impactLow effort First result in 2 weeks CEO and Co founder

    First action

    Add a footer link to the YC job page today.

    How to measure it

    Applicants from godela.ai

    Now
    –
    Target
    +50%
    By
    60 days

    Steps

    1. 1

      Link today

      Footer link to the YC job page.

      Founding Engineer (SWE) 15 minutes

    2. 2

      Write the page

      Three roles, pitch, first 90 days.

      CEO and Co-founder half a day

    3. 3

      Show the team

      Research roots and problems being solved.

      CTO and Co-founder 2 hours

    4. 4

      Publish /careers

      Header and footer links.

      Founding Engineer (SWE) 1 hour

    5. 5

      Share

      Post the page on LinkedIn.

      CEO and Co-founder 1 hour

    6. 6

      Track

      Applicants by source.

      CEO and Co-founder monthly

    First result
    2 weeks
    Cost
    $0
    Owner
    CEO and Co founder

    Smallest test

    Add the footer link and count applications for a month.

    Why it works

    Add a "Careers" link in the header and footer pointing to a /careers page that lists the three roles with a one-line pitch each and links to the YC postings. Add a sentence on what a new hire will work on in the first 90 days (for example the validation work in idea 3).

    Founding engineers decide in seconds whether a startup is serious. A real careers page brings more candidates and keeps hiring mail out of the sales inbox. Buyers also read open roles as a sign that the company is growing.

    Risks and how to handle them

    • Page goes stale when roles close

      Link to the YC job page as the live source.

    Evidence

    • HiringFounding Engineer (SWE): builds user-facing systems that expose physics-informed models as applications and APIs connecting training pipelines, inference and customer interfaces; CAD and interactive engineering apps are a plus. Founding Engineer (ML Systems): multi-GPU, multi-node training of large physics-informed models on AWS. The three open roles are on Godela's YC job page; godela.ai does not link to them.Source
    • Your websiteSite structure (checked November 9, 2026): godela.ai is one page; /sitemap.xml, /robots.txt, /privacy and /careers return 404. Meta description: "Godela is bringing AI-powered physics intelligence to the physical world." The headline changed from "Godela" to "Godela is the frontier of AI for the physical world." after October 2025. No blog, newsletter, case studies, testimonials or video channel; LinkedIn is the only social profile. Domain registered February 2025.Source
    • Your websiteHomepage demo panel, "Data center cooling optimization": "2,400 simulations · 12 parameters", "inference: 0.8s", "4,200 configs evaluated · 18 min · peak temp ↓ 10.1%", peak temperature cut from 72.1°C to 64.8°C; the prompt asks for "the top 5 configurations ranked by cooling efficiency" but the answer lists three. The page says Godela "predicts system behavior with solver-level accuracy in seconds instead of days" but gives no accuracy figure and names no customers. Team: "Stanford and MIT trained physicists, ML researchers, and engineers"; job seekers are asked to email [email protected].Source
  • 06

    The headline changed to "Godela is the frontier of AI for the physical world.", but the description in search results still reads "bringing AI-powered physics intelligence" and names no use case. /sitemap.xml and /robots.txt return 404. Engineers searching for faster CFD or thermal simulation get no signal that Godela does that.

    Medium impactLow effort First result in 1 month CEO and Co founder

    First action

    Rewrite the meta description.

    How to measure it

    Non-brand search impressions

    Now
    –
    Target
    From near zero to a measurable baseline
    By
    90 days

    Steps

    1. 1

      Rewrite description

      What it does, for whom, in seconds.

      CEO and Co-founder 30 minutes

    2. 2

      Fix the title

      Add "AI physics engine for engineers" to the page title.

      CEO and Co-founder 15 minutes

    3. 3

      Add robots and sitemap

      Static files on the host.

      Founding Engineer (SWE) 1 hour

    4. 4

      Verify Search Console

      Submit the sitemap.

      Founding Engineer (SWE) 30 minutes

    5. 5

      Check previews

      LinkedIn and messaging previews.

      CEO and Co-founder 15 minutes

    6. 6

      Measure

      Impressions and queries.

      CEO and Co-founder monthly

    First result
    1 month
    Cost
    $0
    Owner
    CEO and Co founder

    Smallest test

    Change the description today and check the LinkedIn preview.

    Why it works

    Write a description that says what Godela does in engineers' words: physics-informed AI models trained on your simulation data, answering CFD, thermal and structural what-ifs in seconds. Add a robots.txt and a sitemap.xml and submit them in Search Console. Update both as application pages (idea 8) go live.

    The search snippet and link previews are how most people first meet the company, and concrete terms like CFD and thermal match what buyers type.

    Risks and how to handle them

    • One page can only rank for a few terms

      Pair with application pages (idea 8).

    Evidence

    • Your websiteSite structure (checked November 9, 2026): godela.ai is one page; /sitemap.xml, /robots.txt, /privacy and /careers return 404. Meta description: "Godela is bringing AI-powered physics intelligence to the physical world." The headline changed from "Godela" to "Godela is the frontier of AI for the physical world." after October 2025. No blog, newsletter, case studies, testimonials or video channel; LinkedIn is the only social profile. Domain registered February 2025.Source
  • 07

    The site is fast: PageSpeed scores 100 and real visitors pass Core Web Vitals. But 101 KB of the 121 KB page is font files from Google Fonts, render-blocking requests cost about 0.7s, and field LCP has slowed 12%. Accessibility scores 94 because of low-contrast text and out-of-order headings.

    Low impactLow effort First result in 1 month Founding Engineer

    First action

    List the font weights actually used.

    How to measure it

    Real-user LCP

    Now
    –
    Target
    Under 1.5s
    By
    60 days

    Steps

    1. 1

      Audit fonts

      Weights and characters used.

      Founding Engineer (SWE) 30 minutes

    2. 2

      Self-host

      Subset files with swap and preload.

      Founding Engineer (SWE) 2 hours

    3. 3

      Fix contrast

      Grey text to at least 4.5:1.

      Founding Engineer (SWE) 1 hour

    4. 4

      Fix headings

      One H1, ordered H2s and H3s.

      Founding Engineer (SWE) 1 hour

    5. 5

      Retest

      PageSpeed mobile and desktop.

      Founding Engineer (SWE) 15 minutes

    6. 6

      Watch field data

      Field LCP trend.

      Founding Engineer (SWE) monthly

    First result
    1 month
    Cost
    $0
    Owner
    Founding Engineer (SWE)

    Smallest test

    Fix the contrast today and rerun PageSpeed.

    Why it works

    Self-host one subset font file with font-display: swap and preload it, then drop the extra font requests. Raise the contrast on the grey text and fix the heading order. This keeps the site fast as pages are added.

    It is cheap insurance: the site stays fast while it grows, and engineers reading on bright screens or with low vision can read the claims.

    Risks and how to handle them

    • Font licence terms

      Use an open-licence font, as Google Fonts are.

    Evidence

    • Google PageSpeedMobile and desktop performance 100, SEO 100, accessibility 94 (text contrast too low; headings out of order). Real visitors pass Core Web Vitals (LCP p75 1.67s, INP 64ms, CLS 0), though field LCP has slowed 12%. The page is 121 KB, of which 101 KB are Google Fonts files over 6 requests; render-blocking requests cost about 0.7s; the largest element is the "Backed by Y Combinator (YC X25)" badge. Checked November 9, 2026.Source

This quarter 11

The core plays for the next three months.

  • 08

    The homepage lists six applications (Data Center Thermals, Electronics Cooling, Aerodynamics, Electromagnetics, Structures, Build Your Own), but none has its own page. Engineers search by problem ("rack airflow simulation", "heat sink CFD", "drag prediction"), and a single page cannot rank for six different problems or speak to six different buyers.

    High impactMedium effort First result in 2 months CEO and Co founder

    First action

    Outline the Data Center Thermals page from the existing demo.

    How to measure it

    Application-page pilot requests

    Now
    –
    Target
    10 a month
    By
    2 quarters

    Steps

    1. 1

      Pick the order

      Rank applications by inbound interest and pilots.

      CEO and Co-founder 1 day

    2. 2

      Gather the example

      One validated what-if per application.

      Founding Simulation Engineer 1 week each

    3. 3

      Write

      Problem, inputs, example, validation, call to action.

      CEO and Co-founder 2 days each

    4. 4

      Build the template

      A reusable page layout.

      Founding Engineer (SWE) 3 days

    5. 5

      Publish

      Add to the sitemap and the homepage cards.

      Founding Engineer (SWE) 1 day each

    6. 6

      Promote

      Share each page on LinkedIn with the example.

      CEO and Co-founder per page

    7. 7

      Measure

      Searches and pilot requests by page.

      CEO and Co-founder monthly

    First result
    2 months
    Cost
    Team time
    Owner
    CEO and Co founder

    Smallest test

    Publish the Data Center Thermals page and track requests for a month.

    Why it works

    Build one page per application: the problem and the usual solver time; what Godela needs (formats, typical number of runs); an example what-if with real output and its validation; who it is for; and a link to /pilot with the application filled in. Start with Data Center Thermals, where the homepage demo already lives, then Electronics Cooling and Aerodynamics.

    Each page matches a real search and a real buyer, and the pilot request arrives already scoped.

    Risks and how to handle them

    • Thin pages

      Publish only with a real, validated example.

    Builds on

    Evidence

    • Your websiteHomepage (checked November 9, 2026): headline "Godela is the frontier of AI for the physical world."; a Train, Explore, Optimize flow (upload simulation data and Godela trains a physics-constrained model); applications: Data Center Thermals, Electronics Cooling, Aerodynamics, Electromagnetics, Structures and Build Your Own. The "Contact us" button and both "Request a Pilot" buttons open the same personal Calendly page of a founder; the only other links are [email protected], LinkedIn and the YC launch post. There is no privacy policy, terms, security or data-handling page, careers, docs, blog, case studies or pricing, and nothing on how uploaded CAD or simulation data is stored or where the models run.Source
    • Your websiteSite structure (checked November 9, 2026): godela.ai is one page; /sitemap.xml, /robots.txt, /privacy and /careers return 404. Meta description: "Godela is bringing AI-powered physics intelligence to the physical world." The headline changed from "Godela" to "Godela is the frontier of AI for the physical world." after October 2025. No blog, newsletter, case studies, testimonials or video channel; LinkedIn is the only social profile. Domain registered February 2025.Source
  • 09

    Godela's strongest numbers (2.5 hours to 1.8 seconds, 99.99% accuracy) appear in an investor's post, not on godela.ai, and cannot be checked. Rivals also lean on customer logos rather than open numbers. In a category where every vendor claims solver-level accuracy, a public benchmark on open datasets is proof that buyers and researchers can check for themselves.

    High impactMedium effort First result in 2 months Founding Simulation Engineer

    First action

    Shortlist open datasets whose licences allow publishing results.

    How to measure it

    Benchmark-sourced pilot requests

    Now
    –
    Target
    20% of requests
    By
    2 quarters

    Steps

    1. 1

      Pick datasets

      Openly licensed, widely used, matching the applications.

      Founding Simulation Engineer 1 week

    2. 2

      Fix the protocol

      Splits, metrics, hardware and timing method.

      CTO and Co-founder 3 days

    3. 3

      Run

      Train and evaluate.

      Founding Engineer (ML Systems) 2 weeks

    4. 4

      Write up

      Results, method and failure cases.

      Founding Simulation Engineer 1 week

    5. 5

      Publish /benchmark

      Page plus downloadable method.

      Founding Engineer (SWE) 2 days

    6. 6

      Share

      LinkedIn and engineering communities.

      CEO and Co-founder 1 day

    7. 7

      Refresh

      With each model release.

      Founding Simulation Engineer per release

    First result
    2 months
    Cost
    Compute plus about four weeks of engineer time
    Owner
    Founding Simulation Engineer

    Smallest test

    Publish one dataset result with the method.

    Why it works

    Pick two or three openly licensed CFD or structural datasets that the research community already uses. Train Godela on a fixed split and publish error, speed and data needs with the method, and the test inputs where licences allow. Show what Godela gets wrong as well as right. Host it at /benchmark and refresh it with each model release.

    Engineers trust what they can reproduce. A benchmark builds credibility with buyers, gets cited by researchers and attracts the kind of simulation engineers Godela is hiring.

    Risks and how to handle them

    • Results lag a rival's

      Pick tasks that match Godela's applications and report honestly.

    Builds on

    Evidence

    • News coverageCLAI Ventures investment post (July 11, 2025): a car drag and pressure simulation that "takes 2.5 hours on Simscale runs in just 1.8 seconds on Godela"; Godela predicted the Navier-Stokes equation "with 99.99% accuracy from 72mb of fluid flow data"; a first contract signed to replace an incumbent simulation tool; over 200 inquiries after launch; target markets include drop simulation, thermal reentry and soft materials.Source
    • HiringFounding Simulation Engineer (YC job board, San Francisco, $120K–$180K, visa sponsorship): owns validation of model outputs against simulation and experimental ground truth; generates simulation data across CFD, FEA and multiphysics; builds pipelines that turn simulation and experimental data into training-ready datasets, with solvers such as Star-CCM+, FEniCS and OpenFOAM. Godela says it is building "the first Physics Foundation Model"; founded 2025, about 10 people.Source
    • Competitor researchRivals' own homepages: Neural Concept, an "AI-first engineering platform" with four named customer stories across automotive and aerospace and AWS, Microsoft and NVIDIA as deployment partners; PhysicsX, "Engineering. Reengineered.", serving five industries with a sailing-team case study and a "Talk to an engineer" button; Monolith, "Test less. Learn More.", claiming test plans cut by up to 70%, with testimonials from carmakers and aerospace firms, "Request a demo", and now part of CoreWeave.Source
  • 10

    Step one of Godela's flow is "upload simulation data", and the Founding Simulation Engineer role names Star-CCM+, FEniCS and OpenFOAM. The site does not say which formats are accepted, so a prospect cannot tell whether their data will work without a call. Godela positions itself as a layer on top of existing simulations, so reading those outputs is the product's front door.

    Medium impactMedium effort First result in 1 month Founding Engineer

    First action

    List the formats imported in pilots so far.

    How to measure it

    Time to first model in pilots

    Now
    –
    Target
    Under 3 days
    By
    1 quarter

    Steps

    1. 1

      Inventory formats

      From all pilots and inquiries.

      Founding Simulation Engineer 2 days

    2. 2

      Publish the page

      Supported and planned formats.

      Founding Engineer (SWE) 2 days

    3. 3

      Build importers

      Top three formats.

      Founding Engineer (SWE) 6 weeks

    4. 4

      Add a data checker

      Runs and parameters found; is it enough?

      Founding Engineer (SWE) 2 weeks

    5. 5

      Link from /pilot

      So prospects check before booking.

      Founding Engineer (SWE) 1 hour

    6. 6

      Measure

      Time to first model.

      CTO and Co-founder per pilot

    First result
    1 month for the page; 1 quarter for importers
    Cost
    Engineer time
    Owner
    Founding Engineer (SWE)

    Smallest test

    Publish the format list this month.

    Why it works

    Publish a "Works with" page listing the solver output and CAD formats supported today, with planned ones marked. Build one-click importers for the top three formats seen in pilots, with a checker that says how many runs and parameters it found and whether that is enough to train.

    "Will it read our data?" is the first technical question. Answering it publicly brings in self-qualified teams and cuts time to first model in pilots.

    Risks and how to handle them

    • Formats vary across solver versions

      List tested versions.

    Builds on

    Evidence

    • HiringFounding Simulation Engineer (YC job board, San Francisco, $120K–$180K, visa sponsorship): owns validation of model outputs against simulation and experimental ground truth; generates simulation data across CFD, FEA and multiphysics; builds pipelines that turn simulation and experimental data into training-ready datasets, with solvers such as Star-CCM+, FEniCS and OpenFOAM. Godela says it is building "the first Physics Foundation Model"; founded 2025, about 10 people.Source
    • News coverageY Combinator launch post: Godela turns engineering questions into physics-informed models that are faster and cheaper than traditional simulation, positioned as a layer on top of existing simulations rather than a replacement; engineers can ask in natural language or upload CAD files, experimental results or simulation data. The problem it names: weeks or months of simulation and parameter tuning to answer one question, so many questions never get asked. Its asks: introductions to engineers with simulation bottlenecks and partnerships with engineering teams.Source
  • 11

    Godela has signed a first contract and reported 200+ inquiries, so the step that matters now is pilot to contract. Without agreed criteria, technical pilots drift: the model works, but nobody has said what "good enough to buy" means, or who signs.

    High impactLow effort First result in Next pilot CEO and Co founder

    First action

    Draft the scorecard template from the first contract.

    How to measure it

    Pilots converting to paid contracts

    Now
    –
    Target
    50%+
    By
    2 quarters

    Steps

    1. 1

      Draft template

      Questions, threshold, time saved, decision-maker, date.

      CEO and Co-founder 1 day

    2. 2

      Set thresholds

      Accuracy targets by application.

      Founding Simulation Engineer 2 days

    3. 3

      Use at kickoff

      Signed by the champion.

      CEO and Co-founder per pilot

    4. 4

      Weekly note

      Progress against the scorecard.

      Founding Simulation Engineer weekly

    5. 5

      Readout

      Results and a business case.

      CEO and Co-founder per pilot

    6. 6

      Review

      Why pilots converted or did not.

      CEO and Co-founder quarterly

    First result
    Next pilot
    Cost
    $0
    Owner
    CEO and Co founder

    Smallest test

    Use the scorecard on the next pilot.

    Why it works

    Before any data moves, agree a one-page scorecard with the champion: the questions to answer, the accuracy threshold against held-out solver runs, the time-saved target, the decision-maker and the date of the decision. Send a short weekly progress note against it, and end with a readout that becomes the business case.

    Agreed criteria turn a technical success into a purchase decision and give the champion the document they need internally.

    Risks and how to handle them

    • Champions resist commitments

      Keep it to one page with no legal language.

    Evidence

    • News coverageCLAI Ventures investment post (July 11, 2025): a car drag and pressure simulation that "takes 2.5 hours on Simscale runs in just 1.8 seconds on Godela"; Godela predicted the Navier-Stokes equation "with 99.99% accuracy from 72mb of fluid flow data"; a first contract signed to replace an incumbent simulation tool; over 200 inquiries after launch; target markets include drop simulation, thermal reentry and soft materials.Source
    • HiringFounding Simulation Engineer (YC job board, San Francisco, $120K–$180K, visa sponsorship): owns validation of model outputs against simulation and experimental ground truth; generates simulation data across CFD, FEA and multiphysics; builds pipelines that turn simulation and experimental data into training-ready datasets, with solvers such as Star-CCM+, FEniCS and OpenFOAM. Godela says it is building "the first Physics Foundation Model"; founded 2025, about 10 people.Source
  • 12

    Godela publishes no pricing or pilot terms; the only option is to book a call. Engineering managers can often approve a small, fixed spend themselves, but an open-ended evaluation usually goes up the chain to procurement.

    High impactLow effort First result in 1-2 months CEO and Co founder

    First action

    Cost the last pilot in engineer days and compute.

    How to measure it

    Paid pilots per quarter

    Now
    –
    Target
    4
    By
    2 quarters

    Steps

    1. 1

      Cost a pilot

      Engineer days and compute.

      CTO and Co-founder 1 day

    2. 2

      Set scope and fee

      One application, fixed weeks, fixed fee.

      CEO and Co-founder 2 days

    3. 3

      Write terms

      Short pilot agreement with the credit.

      CEO and Co-founder 1 week

    4. 4

      Add to /pilot

      What you get and how long it takes.

      Founding Engineer (SWE) 1 day

    5. 5

      Offer on calls

      As the default next step.

      CEO and Co-founder ongoing

    6. 6

      Review

      Uptake and conversion to annual.

      CEO and Co-founder quarterly

    First result
    1-2 months
    Cost
    Legal review of the pilot agreement
    Owner
    CEO and Co founder

    Smallest test

    Offer the paid pilot to the next three qualified prospects.

    Why it works

    Define a standard paid pilot: one application, the customer's existing runs, a validated model and report in a fixed number of weeks, for a fixed fee credited against a first-year contract. Describe it on /pilot (with or without the price) and keep a free "data check" as the step before.

    A small paid pilot filters for serious buyers, funds the pilot work, and the credit makes the move to a contract the obvious next step.

    Risks and how to handle them

    • A fee deters early adopters

      Waive it for design partners who agree to a case study.

    Builds on

    Evidence

    • Your websiteHomepage (checked November 9, 2026): headline "Godela is the frontier of AI for the physical world."; a Train, Explore, Optimize flow (upload simulation data and Godela trains a physics-constrained model); applications: Data Center Thermals, Electronics Cooling, Aerodynamics, Electromagnetics, Structures and Build Your Own. The "Contact us" button and both "Request a Pilot" buttons open the same personal Calendly page of a founder; the only other links are [email protected], LinkedIn and the YC launch post. There is no privacy policy, terms, security or data-handling page, careers, docs, blog, case studies or pricing, and nothing on how uploaded CAD or simulation data is stored or where the models run.Source
  • 13

    Godela has signed a contract to replace an incumbent simulation tool, but godela.ai names no customers and shows no testimonials or case studies. Neural Concept and Monolith lead with customer stories and testimonials. Many hardware companies will not allow their name to be used, but they often allow an anonymised story with numbers.

    High impactMedium effort First result in 2 months CEO and Co founder

    First action

    Ask the customer what they would be comfortable sharing.

    How to measure it

    Pilot requests after case study view

    Now
    –
    Target
    25% of requests
    By
    2 quarters

    Steps

    1. 1

      Ask

      Permission and limits.

      CEO and Co-founder 1 week

    2. 2

      Gather numbers

      Before and after time, accuracy, iterations.

      Founding Simulation Engineer 1 week

    3. 3

      Interview

      One engineer, 30 minutes.

      CEO and Co-founder 1 day

    4. 4

      Write

      Problem, approach, result, quote by role.

      CEO and Co-founder 3 days

    5. 5

      Approve

      Customer sign-off.

      CEO and Co-founder 2 weeks

    6. 6

      Publish

      Page, PDF and links from /pilot.

      Founding Engineer (SWE) 1 day

    First result
    2 months
    Cost
    Founder time
    Owner
    CEO and Co founder

    Smallest test

    Ask for one approved quote by role this month.

    Why it works

    Ask the first customer for an anonymised story ("a hardware team in X industry"): the question they could not afford to ask, solver hours before, time per answer after, validation and what they did with the extra design iterations. Add one quote from an engineer, attributed by role only. Publish on the site, link it from /pilot and the matching application page, and make a one-page PDF.

    One credible story from a paying customer makes the claims real for the next buyer, and anonymity makes it easier for customers to say yes.

    Risks and how to handle them

    • Customer declines even anonymised use

      Offer a version with ranges instead of exact figures.

    Evidence

    • News coverageCLAI Ventures investment post (July 11, 2025): a car drag and pressure simulation that "takes 2.5 hours on Simscale runs in just 1.8 seconds on Godela"; Godela predicted the Navier-Stokes equation "with 99.99% accuracy from 72mb of fluid flow data"; a first contract signed to replace an incumbent simulation tool; over 200 inquiries after launch; target markets include drop simulation, thermal reentry and soft materials.Source
    • Your websiteSite structure (checked November 9, 2026): godela.ai is one page; /sitemap.xml, /robots.txt, /privacy and /careers return 404. Meta description: "Godela is bringing AI-powered physics intelligence to the physical world." The headline changed from "Godela" to "Godela is the frontier of AI for the physical world." after October 2025. No blog, newsletter, case studies, testimonials or video channel; LinkedIn is the only social profile. Domain registered February 2025.Source
    • Competitor researchRivals' own homepages: Neural Concept, an "AI-first engineering platform" with four named customer stories across automotive and aerospace and AWS, Microsoft and NVIDIA as deployment partners; PhysicsX, "Engineering. Reengineered.", serving five industries with a sailing-team case study and a "Talk to an engineer" button; Monolith, "Test less. Learn More.", claiming test plans cut by up to 70%, with testimonials from carmakers and aerospace firms, "Request a demo", and now part of CoreWeave.Source
  • 14

    Godela's team is described as "Stanford and MIT trained physicists, ML researchers, and engineers", and the founders are commercialising Stanford research. Yet there is no blog, newsletter or video, and LinkedIn is the only social channel. The technical depth that persuades engineers stays invisible.

    Medium impactMedium effort First result in 1 quarter CTO and Co founder

    First action

    List six topics from the questions asked on scoping calls.

    How to measure it

    Engineering subscribers

    Now
    –
    Target
    500
    By
    2 quarters

    Steps

    1. 1

      Pick topics

      Six from call questions.

      CTO and Co-founder 1 day

    2. 2

      Set up /notes

      Simple pages plus email sign-up.

      Founding Engineer (SWE) 3 days

    3. 3

      Write the first

      How much data a surrogate needs.

      CTO and Co-founder 1 week

    4. 4

      Review

      Technical accuracy check.

      Founding Simulation Engineer 1 day

    5. 5

      Publish and share

      Site plus LinkedIn.

      CEO and Co-founder monthly

    6. 6

      Measure

      Subscribers and replies.

      CEO and Co-founder monthly

    First result
    1 quarter
    Cost
    About 1 day a month of founder time
    Owner
    CTO and Co founder

    Smallest test

    Publish one note on LinkedIn and count engineer replies.

    Why it works

    Publish one technical note a month at /notes, cross-posted to LinkedIn: how physics constraints are applied, how much data a surrogate needs, where surrogates fail, how validation is done. Write for practising simulation engineers, not investors. Add an email sign-up for new notes.

    Engineers buy from people they believe understand physics. Notes build that belief before a call, give sales material to forward, and help with hiring.

    Risks and how to handle them

    • Sharing too much method

      Explain principles and limits, not internals.

    Evidence

    • Your websiteHomepage demo panel, "Data center cooling optimization": "2,400 simulations · 12 parameters", "inference: 0.8s", "4,200 configs evaluated · 18 min · peak temp ↓ 10.1%", peak temperature cut from 72.1°C to 64.8°C; the prompt asks for "the top 5 configurations ranked by cooling efficiency" but the answer lists three. The page says Godela "predicts system behavior with solver-level accuracy in seconds instead of days" but gives no accuracy figure and names no customers. Team: "Stanford and MIT trained physicists, ML researchers, and engineers"; job seekers are asked to email [email protected].Source
    • Your websiteSite structure (checked November 9, 2026): godela.ai is one page; /sitemap.xml, /robots.txt, /privacy and /careers return 404. Meta description: "Godela is bringing AI-powered physics intelligence to the physical world." The headline changed from "Godela" to "Godela is the frontier of AI for the physical world." after October 2025. No blog, newsletter, case studies, testimonials or video channel; LinkedIn is the only social profile. Domain registered February 2025.Source
  • 15

    Godela's pitch is that simulation takes weeks or months per question. Companies with several open CFD, thermal or FEA analyst roles are showing, in public, that demand for simulation exceeds their capacity, which is exactly the bottleneck Godela removes. The same skills Godela hires for (Star-CCM+, OpenFOAM, CFD, FEA) are the search terms for finding them.

    Medium impactLow effort First result in 1 month CEO and Co founder

    First action

    Run the search for data center thermal roles and list ten companies.

    How to measure it

    Calls booked from outreach

    Now
    –
    Target
    4 a month
    By
    1 quarter

    Steps

    1. 1

      Define searches

      Role titles and tools per application.

      Founding Simulation Engineer 1 day

    2. 2

      Build the list

      Companies with two or more open roles.

      CEO and Co-founder weekly

    3. 3

      Write notes

      Short, specific to the role posted.

      CEO and Co-founder 1 day

    4. 4

      Send

      Ten a week.

      CEO and Co-founder weekly

    5. 5

      Follow up

      With the application page or benchmark.

      CEO and Co-founder weekly

    6. 6

      Measure

      Reply and call rates by application.

      CEO and Co-founder monthly

    First result
    1 month
    Cost
    Founder time
    Owner
    CEO and Co founder

    Smallest test

    Send ten notes in one application this month.

    Why it works

    Each week, search public job boards for simulation roles in Godela's applications (data center thermal, electronics cooling, aerodynamics, structures). List companies with two or more open roles, find the hiring manager's team lead, and send a short note offering to show how a surrogate could answer more questions with the team they already have. Track replies by application.

    A public signal of a capacity problem makes outreach timely and relevant, rather than cold.

    Risks and how to handle them

    • Teams read it as replacing their hires

      Frame it as more answers from the same team.

    Builds on

    Evidence

    • News coverageY Combinator launch post: Godela turns engineering questions into physics-informed models that are faster and cheaper than traditional simulation, positioned as a layer on top of existing simulations rather than a replacement; engineers can ask in natural language or upload CAD files, experimental results or simulation data. The problem it names: weeks or months of simulation and parameter tuning to answer one question, so many questions never get asked. Its asks: introductions to engineers with simulation bottlenecks and partnerships with engineering teams.Source
    • HiringFounding Simulation Engineer (YC job board, San Francisco, $120K–$180K, visa sponsorship): owns validation of model outputs against simulation and experimental ground truth; generates simulation data across CFD, FEA and multiphysics; builds pipelines that turn simulation and experimental data into training-ready datasets, with solvers such as Star-CCM+, FEniCS and OpenFOAM. Godela says it is building "the first Physics Foundation Model"; founded 2025, about 10 people.Source
  • 16

    Godela's only homepage demo is data center cooling: 2,400 simulations, 4,200 configurations evaluated in 18 minutes, peak temperature down 10.1%. Consultancies that design and commission data center cooling run these studies for many clients. One partner reaches dozens of projects, and the consultancy keeps the client relationship.

    Medium impactMedium effort First result in 1-2 quarters CEO and Co founder

    First action

    List ten consultancies that publish data center CFD or thermal services.

    How to measure it

    Pilots from partners

    Now
    –
    Target
    2
    By
    2 quarters

    Steps

    1. 1

      List partners

      Ten consultancies offering data center CFD.

      CEO and Co-founder 2 days

    2. 2

      Write the offer

      Partner pilot plus co-marketing.

      CEO and Co-founder 1 day

    3. 3

      Reach out

      Engineering directors.

      CEO and Co-founder 2 weeks

    4. 4

      Run one pilot

      On a past study.

      Founding Simulation Engineer 6 weeks

    5. 5

      Co-publish

      Case study with the partner.

      CEO and Co-founder 2 weeks

    6. 6

      Set referral terms

      Simple fee or discount.

      CEO and Co-founder 1 week

    First result
    1-2 quarters
    Cost
    Pilot engineering time
    Owner
    CEO and Co founder

    Smallest test

    One partner conversation this month.

    Why it works

    Offer two or three cooling and data center engineering consultancies a partner pilot: Godela trains on their past CFD studies (with permission) so they can answer client what-ifs in a meeting instead of in a follow-up report. In return: a co-authored case study and referrals to their clients' in-house teams.

    Consultancies feel the simulation bottleneck on every bid and project, and faster answers become their own selling point.

    Risks and how to handle them

    • Client data permissions

      Use the partner's own reference studies first.

    Builds on

    Evidence

    • Your websiteHomepage demo panel, "Data center cooling optimization": "2,400 simulations · 12 parameters", "inference: 0.8s", "4,200 configs evaluated · 18 min · peak temp ↓ 10.1%", peak temperature cut from 72.1°C to 64.8°C; the prompt asks for "the top 5 configurations ranked by cooling efficiency" but the answer lists three. The page says Godela "predicts system behavior with solver-level accuracy in seconds instead of days" but gives no accuracy figure and names no customers. Team: "Stanford and MIT trained physicists, ML researchers, and engineers"; job seekers are asked to email [email protected].Source
    • News coverageY Combinator launch post: Godela turns engineering questions into physics-informed models that are faster and cheaper than traditional simulation, positioned as a layer on top of existing simulations rather than a replacement; engineers can ask in natural language or upload CAD files, experimental results or simulation data. The problem it names: weeks or months of simulation and parameter tuning to answer one question, so many questions never get asked. Its asks: introductions to engineers with simulation bottlenecks and partnerships with engineering teams.Source
  • 17

    The YC launch asked for introductions to engineers with simulation bottlenecks, and Godela has several investors. But there is nothing short and specific to forward: the homepage is general, and the best numbers are in an investor's post. Generic asks get generic intros.

    Medium impactLow effort First result in 1 month CEO and Co founder

    First action

    Write the data center thermal blurb.

    How to measure it

    Qualified intros

    Now
    –
    Target
    6 a month
    By
    1 quarter

    Steps

    1. 1

      Write blurbs

      Three sentences per application.

      CEO and Co-founder 1 day

    2. 2

      Make PDFs

      One page each.

      CEO and Co-founder 2 days

    3. 3

      Pick the monthly ask

      One application, one team type.

      CEO and Co-founder monthly

    4. 4

      Send the note

      Investors and YC hardware founders.

      CEO and Co-founder monthly

    5. 5

      Thank and report

      Tell introducers what happened.

      CEO and Co-founder per intro

    6. 6

      Track

      Intros to calls to pilots.

      CEO and Co-founder monthly

    First result
    1 month
    Cost
    $0
    Owner
    CEO and Co founder

    Smallest test

    Send one specific ask to investors this month.

    Why it works

    Write a three-sentence blurb and a one-page PDF per application (who it is for, the before and after, the pilot offer), plus a monthly note to investors and YC hardware founders naming the one application and type of team Godela wants to meet that month. Track every intro to outcome.

    Specific asks get specific intros, and warm intros convert to pilots far better than cold inbound.

    Risks and how to handle them

    • Intro fatigue

      One focused ask a month, not a list.

    Evidence

    • News coverageY Combinator launch post: Godela turns engineering questions into physics-informed models that are faster and cheaper than traditional simulation, positioned as a layer on top of existing simulations rather than a replacement; engineers can ask in natural language or upload CAD files, experimental results or simulation data. The problem it names: weeks or months of simulation and parameter tuning to answer one question, so many questions never get asked. Its asks: introductions to engineers with simulation bottlenecks and partnerships with engineering teams.Source
    • News coverageCLAI Ventures investment post (July 11, 2025): a car drag and pressure simulation that "takes 2.5 hours on Simscale runs in just 1.8 seconds on Godela"; Godela predicted the Navier-Stokes equation "with 99.99% accuracy from 72mb of fluid flow data"; a first contract signed to replace an incumbent simulation tool; over 200 inquiries after launch; target markets include drop simulation, thermal reentry and soft materials.Source
  • 18

    Godela trains on AWS (the ML Systems role requires hands-on AWS), and Neural Concept already lists AWS as a deployment partner. For a ten-person vendor, the slowest step after a successful pilot is often getting set up as a new supplier. Buying through a marketplace uses a relationship the customer already has.

    Medium impactMedium effort First result in 1-2 quarters CTO and Co founder

    First action

    Check the marketplace seller requirements and the effort to list.

    How to measure it

    Procurement cycle time

    Now
    –
    Target
    -30%
    By
    2 quarters

    Steps

    1. 1

      Check requirements

      Seller registration and listing type.

      CTO and Co-founder 2 days

    2. 2

      Define offers

      Paid pilot and annual licence.

      CEO and Co-founder 2 days

    3. 3

      Create the listing

      Private offers first.

      Founding Engineer (SWE) 2 weeks

    4. 4

      Add to /pilot

      Show the buying option.

      Founding Engineer (SWE) 1 hour

    5. 5

      Offer on calls

      To prospects already on AWS.

      CEO and Co-founder ongoing

    6. 6

      Measure

      Cycle time by buying route.

      CEO and Co-founder quarterly

    First result
    1-2 quarters
    Cost
    Marketplace fees on transactions plus setup time
    Owner
    CTO and Co founder

    Smallest test

    Ask the next three prospects whether marketplace buying would help.

    Why it works

    Publish a private-offer listing for the paid pilot and annual licence, so customers can buy through their existing AWS billing. Mention it on /pilot ("available through AWS Marketplace"). Later, add a deployment option in the customer's own cloud account if the data-handling work (idea 1) supports it.

    It removes a procurement step at the moment a champion has momentum, and the marketplace badge adds credibility next to rivals' partner logos.

    Risks and how to handle them

    • Customer is on another cloud

      Keep direct contracts; add other marketplaces only when demand shows.

    Builds on

    Evidence

    • HiringFounding Engineer (SWE): builds user-facing systems that expose physics-informed models as applications and APIs connecting training pipelines, inference and customer interfaces; CAD and interactive engineering apps are a plus. Founding Engineer (ML Systems): multi-GPU, multi-node training of large physics-informed models on AWS. The three open roles are on Godela's YC job page; godela.ai does not link to them.Source
    • Competitor researchRivals' own homepages: Neural Concept, an "AI-first engineering platform" with four named customer stories across automotive and aerospace and AWS, Microsoft and NVIDIA as deployment partners; PhysicsX, "Engineering. Reengineered.", serving five industries with a sailing-team case study and a "Talk to an engineer" button; Monolith, "Test less. Learn More.", claiming test plans cut by up to 70%, with testimonials from carmakers and aerospace firms, "Request a demo", and now part of CoreWeave.Source

Big bets 3

Larger plays that can change the trajectory.

  • 19

    The homepage demo looks like a working product, with an input box reading "Ask a question about your system...", but it is a fixed illustration. Answers in under a second are the product's whole promise, and the one place a visitor could feel that is not interactive. The Founding SWE role is hired to build "user-facing systems that expose physics-informed models as applications".

    High impactHigh effort First result in 2 quarters Founding Engineer

    First action

    Choose the pre-trained model and the three controls to expose.

    How to measure it

    Sandbox-to-pilot rate

    Now
    –
    Target
    3%
    By
    2 quarters after launch

    Steps

    1. 1

      Pick the model

      Data center case with validation.

      Founding Simulation Engineer 1 week

    2. 2

      Design controls

      Three sliders plus a question box.

      Founding Engineer (SWE) 1 week

    3. 3

      Build inference API

      Rate-limited public endpoint.

      Founding Engineer (ML Systems) 4 weeks

    4. 4

      Build the UI

      Live temperature map.

      Founding Engineer (SWE) 4 weeks

    5. 5

      Add the handoff

      "Try this on your data" to /pilot.

      Founding Engineer (SWE) 2 days

    6. 6

      Launch

      LinkedIn, notes and outreach.

      CEO and Co-founder 1 week

    7. 7

      Measure

      Questions asked and pilot requests.

      CEO and Co-founder monthly

    First result
    2 quarters
    Cost
    About 2 engineer-months plus inference compute
    Owner
    Founding Engineer (SWE)

    Smallest test

    Make the existing three-step demo clickable with precomputed answers.

    Why it works

    Ship a public sandbox on a pre-trained data center model: visitors adjust airflow per zone or ask a what-if in plain language and see the temperature map update in under a second, with the validation figure beside it. Cap usage, require no sign-up for the first questions, then offer "Try this on your data" leading to /pilot.

    Seeing a solver-quality answer in a second persuades engineers better than any claim, and the sandbox becomes the asset every outreach and intro links to.

    Risks and how to handle them

    • Visitors push the model outside its valid range

      Limit inputs to the trained range and say so.

    • Compute costs

      Rate-limit and cache common queries.

    Builds on

    Evidence

    • Your websiteHomepage demo panel, "Data center cooling optimization": "2,400 simulations · 12 parameters", "inference: 0.8s", "4,200 configs evaluated · 18 min · peak temp ↓ 10.1%", peak temperature cut from 72.1°C to 64.8°C; the prompt asks for "the top 5 configurations ranked by cooling efficiency" but the answer lists three. The page says Godela "predicts system behavior with solver-level accuracy in seconds instead of days" but gives no accuracy figure and names no customers. Team: "Stanford and MIT trained physicists, ML researchers, and engineers"; job seekers are asked to email [email protected].Source
    • HiringFounding Engineer (SWE): builds user-facing systems that expose physics-informed models as applications and APIs connecting training pipelines, inference and customer interfaces; CAD and interactive engineering apps are a plus. Founding Engineer (ML Systems): multi-GPU, multi-node training of large physics-informed models on AWS. The three open roles are on Godela's YC job page; godela.ai does not link to them.Source
  • 20

    Godela lists six applications, plus "Build Your Own". It leads with data center cooling, a field where every new hall and rack layout needs thermal studies. A horizontal "physics engine" is hard for a buyer to budget for; a named product for one recurring job is easier to buy.

    High impactHigh effort First result in 2-3 quarters CEO and Co founder

    First action

    Interview five data center thermal engineers about which studies they repeat most.

    How to measure it

    Data center contracts

    Now
    –
    Target
    5
    By
    3 quarters

    Steps

    1. 1

      Interview buyers

      Five thermal engineers.

      CEO and Co-founder 3 weeks

    2. 2

      Define the product

      Standard what-ifs and deliverables.

      CEO and Co-founder 2 weeks

    3. 3

      Build the base model

      Common layouts.

      Founding Simulation Engineer 2 months

    4. 4

      Fine-tune flow

      Customer runs to a tuned model.

      Founding Engineer (ML Systems) 6 weeks

    5. 5

      Price it

      Per site or per team, annual.

      CEO and Co-founder 1 week

    6. 6

      Launch

      Page, sandbox, partners.

      CEO and Co-founder 1 month

    7. 7

      Review

      Win rate versus general pilots.

      CEO and Co-founder quarterly

    First result
    2-3 quarters
    Cost
    Several engineer-months
    Owner
    CEO and Co founder

    Smallest test

    Sell three data center pilots with the same scope before building the base model.

    Why it works

    Turn the data center work into a named offering: a pre-trained base model for common hall and rack layouts, fine-tuned on the customer's own CFD runs, with a standard set of what-ifs (airflow, rack density, failure scenarios) and an annual price per site or per team. Support it with the application page, the sandbox and consultancy partners.

    Focus on one repeatable product shortens sales cycles, builds reference customers in one community and gives the team a clear first market before broadening.

    Risks and how to handle them

    • Narrowing too early

      Keep "Build Your Own" pilots running in parallel.

    Builds on

    Evidence

    • Your websiteHomepage (checked November 9, 2026): headline "Godela is the frontier of AI for the physical world."; a Train, Explore, Optimize flow (upload simulation data and Godela trains a physics-constrained model); applications: Data Center Thermals, Electronics Cooling, Aerodynamics, Electromagnetics, Structures and Build Your Own. The "Contact us" button and both "Request a Pilot" buttons open the same personal Calendly page of a founder; the only other links are [email protected], LinkedIn and the YC launch post. There is no privacy policy, terms, security or data-handling page, careers, docs, blog, case studies or pricing, and nothing on how uploaded CAD or simulation data is stored or where the models run.Source
    • Your websiteHomepage demo panel, "Data center cooling optimization": "2,400 simulations · 12 parameters", "inference: 0.8s", "4,200 configs evaluated · 18 min · peak temp ↓ 10.1%", peak temperature cut from 72.1°C to 64.8°C; the prompt asks for "the top 5 configurations ranked by cooling efficiency" but the answer lists three. The page says Godela "predicts system behavior with solver-level accuracy in seconds instead of days" but gives no accuracy figure and names no customers. Team: "Stanford and MIT trained physicists, ML researchers, and engineers"; job seekers are asked to email [email protected].Source
  • 21

    Godela positions itself as a layer on top of existing simulations, not a replacement, and its simulation hire works in Star-CCM+, FEniCS and OpenFOAM. But today the flow is to upload data to Godela's own interface. Engineers spend their day in their solver and post-processing tools, and switching tools is the hardest habit to change.

    High impactHigh effort First result in 2-3 quarters CTO and Co founder

    First action

    Map how pilot users move data in and out today.

    How to measure it

    Weekly active engineers

    Now
    –
    Target
    2x
    By
    3 quarters

    Steps

    1. 1

      Map workflows

      From pilot users.

      Founding Simulation Engineer 2 weeks

    2. 2

      Ship a Python package

      Train and predict from scripts.

      Founding Engineer (SWE) 6 weeks

    3. 3

      Build the OpenFOAM path

      Send runs, return predictions.

      Founding Simulation Engineer 2 months

    4. 4

      Document

      Quick-start guides.

      Founding Engineer (SWE) 2 weeks

    5. 5

      Release

      Open-source the connector.

      CTO and Co-founder 1 week

    6. 6

      Measure

      Usage per customer.

      CTO and Co-founder monthly

    First result
    2-3 quarters
    Cost
    Several engineer-months
    Owner
    CTO and Co founder

    Smallest test

    A Python package that trains from a folder of runs.

    Why it works

    Start with an open-source path: a script or plugin for an OpenFOAM-based workflow that sends finished runs to Godela for training and pulls fast predictions back into the familiar format. Then add an API and Python package for teams with their own pipelines. Publish both on the "Works with" page.

    Meeting engineers in their own tools turns Godela from a separate destination into a step in the existing workflow, which raises daily use and renewals.

    Risks and how to handle them

    • Licensing limits on commercial tool plugins

      Start with open-source solvers and file-based exchange.

    Builds on

    Evidence

    • News coverageY Combinator launch post: Godela turns engineering questions into physics-informed models that are faster and cheaper than traditional simulation, positioned as a layer on top of existing simulations rather than a replacement; engineers can ask in natural language or upload CAD files, experimental results or simulation data. The problem it names: weeks or months of simulation and parameter tuning to answer one question, so many questions never get asked. Its asks: introductions to engineers with simulation bottlenecks and partnerships with engineering teams.Source
    • HiringFounding Simulation Engineer (YC job board, San Francisco, $120K–$180K, visa sponsorship): owns validation of model outputs against simulation and experimental ground truth; generates simulation data across CFD, FEA and multiphysics; builds pipelines that turn simulation and experimental data into training-ready datasets, with solvers such as Star-CCM+, FEniCS and OpenFOAM. Godela says it is building "the first Physics Foundation Model"; founded 2025, about 10 people.Source

Wildcards 2

Unexpected experiments, each with a cheap way to test it.

  • 22

    Godela comes out of Stanford research, and its team is "Stanford and MIT trained". It is hiring simulation and ML engineers who come from those labs. Graduate students who use a tool in their thesis bring it into their first job, and labs produce exactly the varied simulation data and published results that build a technical reputation.

    Medium impactMedium effort First result in 2 quarters CTO and Co founder

    First action

    Pick three labs the founders already know.

    How to measure it

    Academic users

    Now
    –
    Target
    5 labs
    By
    1 year

    Steps

    1. 1

      Pick labs

      Three known labs.

      CTO and Co-founder 1 week

    2. 2

      Write terms

      Free, capped, acknowledgement, publishing rights.

      CEO and Co-founder 2 weeks

    3. 3

      Onboard

      One engineer as point of contact.

      Founding Simulation Engineer 1 month

    4. 4

      Run a challenge

      One open dataset per year.

      Founding Simulation Engineer 1 quarter

    5. 5

      Feature results

      On /notes and LinkedIn.

      CEO and Co-founder ongoing

    6. 6

      Recruit

      Invite strong students to apply.

      CTO and Co-founder ongoing

    First result
    2 quarters
    Cost
    Compute credits plus engineer support time
    Owner
    CTO and Co founder

    Smallest test

    One lab, one semester.

    Why it works

    Offer a capped, free academic licence to a handful of university labs working in thermal, fluids or structures, in exchange for acknowledgement in papers and permission to publish anonymised benchmark results. Run a yearly student challenge on one open dataset. Use the programme as a hiring pipeline too.

    Academic adoption compounds: citations, trained graduates who carry the tool into industry and a steady flow of candidates.

    Risks and how to handle them

    • Support load distracts from customers

      Cap labs and limit support to office hours.

    Builds on

    Evidence

    • Your websiteHomepage demo panel, "Data center cooling optimization": "2,400 simulations · 12 parameters", "inference: 0.8s", "4,200 configs evaluated · 18 min · peak temp ↓ 10.1%", peak temperature cut from 72.1°C to 64.8°C; the prompt asks for "the top 5 configurations ranked by cooling efficiency" but the answer lists three. The page says Godela "predicts system behavior with solver-level accuracy in seconds instead of days" but gives no accuracy figure and names no customers. Team: "Stanford and MIT trained physicists, ML researchers, and engineers"; job seekers are asked to email [email protected].Source
    • HiringFounding Simulation Engineer (YC job board, San Francisco, $120K–$180K, visa sponsorship): owns validation of model outputs against simulation and experimental ground truth; generates simulation data across CFD, FEA and multiphysics; builds pipelines that turn simulation and experimental data into training-ready datasets, with solvers such as Star-CCM+, FEniCS and OpenFOAM. Godela says it is building "the first Physics Foundation Model"; founded 2025, about 10 people.Source
    • News coverageThe American Bazaar (June 3, 2025): Godela came out of stealth from YC's 2025 batch; engineers ask questions in natural language without coding and can add CAD, experimental and legacy simulation data; its surrogate models are grounded in physical laws rather than black-box predictions; target areas include clean energy, aerospace and advanced materials.Source
  • 23

    Godela's launch framed the problem as questions that never get asked because each one costs weeks of simulation. That is a story format, not just a pitch. LinkedIn is Godela's only social channel, and engineers share surprising physics.

    Medium impactLow effort First result in 1 month CEO and Co founder

    First action

    Pick the first four questions from the six applications.

    How to measure it

    Inbound "ask us yours" questions

    Now
    –
    Target
    10 a month
    By
    1 quarter

    Steps

    1. 1

      Pick questions

      Four to start, one per application.

      CEO and Co-founder 1 day

    2. 2

      Run them

      Validated answers.

      Founding Simulation Engineer weekly

    3. 3

      Record

      30-second screen capture.

      Founding Engineer (SWE) weekly

    4. 4

      Post

      LinkedIn, with the solver-time comparison.

      CEO and Co-founder weekly

    5. 5

      Invite questions

      "Ask us yours" form.

      CEO and Co-founder ongoing

    6. 6

      Archive

      Series page on the site.

      Founding Engineer (SWE) monthly

    First result
    1 month
    Cost
    A few hours a week
    Owner
    CEO and Co founder

    Smallest test

    Post one question this week and count replies.

    Why it works

    Every week, pick one engineering question people wonder about but never simulate (what happens to rack temperatures if one cooling unit fails, how much drag a roof box adds, how a heat sink changes with fin spacing). Answer it with a Godela model, post a 30-second screen recording and the result, and say how long the traditional run would take. Collect the series on the site.

    It shows the product working rather than describing it, gives a reason to follow every week, and each post ends with an easy step: "ask us yours".

    Risks and how to handle them

    • An answer is wrong in public

      Validate each against a solver run before posting.

    Evidence

    • News coverageY Combinator launch post: Godela turns engineering questions into physics-informed models that are faster and cheaper than traditional simulation, positioned as a layer on top of existing simulations rather than a replacement; engineers can ask in natural language or upload CAD files, experimental results or simulation data. The problem it names: weeks or months of simulation and parameter tuning to answer one question, so many questions never get asked. Its asks: introductions to engineers with simulation bottlenecks and partnerships with engineering teams.Source
    • Your websiteSite structure (checked November 9, 2026): godela.ai is one page; /sitemap.xml, /robots.txt, /privacy and /careers return 404. Meta description: "Godela is bringing AI-powered physics intelligence to the physical world." The headline changed from "Godela" to "Godela is the frontier of AI for the physical world." after October 2025. No blog, newsletter, case studies, testimonials or video channel; LinkedIn is the only social profile. Domain registered February 2025.Source

03

How Programmable can help

Where an idea in this playbook is something Programmable builds or runs, it's listed here with what we'd do. Everything else is yours to run.

ProgrammableInfo

Turn the research into a list you can act on

These ideas start with finding the right companies. Info finds them and the right person at each, researches every one and drafts the outreach, then asks before it sends.

  • Build this list in Info for idea #15 (opens programmable.info in a new tab)

New accounts: 100,000 free credits with code BETA100K

Find the companies (opens programmable.info in a new tab)
More example playbooks and reviews

More growth playbooks

Other companies we have researched, each read from what they are hiring for.

  • Lance growth strategyCollaborative AI for hotels: an AI front desk that answers guest calls 24/7, guest messaging, an AI sales agent for group and event…23 growth ideas
  • Launch Potato growth strategyA performance marketing company that owns consumer brands (FinanceBuzz, SmarterAuto, All About Cookies, Only In Your State, PrizeGrab and…21 growth ideas
  • Primitive growth strategyEmail infrastructure for AI agents: managed *.primitive.email addresses, a one-call send-and-wait-for-reply verb (primitive chat), hosted…23 growth ideas
  • Crystal Art Gallery growth strategyDesigns, makes, sources and distributes mass-market home décor (framed and canvas art, mirrors, frames, rugs, storage and accents) and…22 growth ideas
  • Aravolta growth strategyOne platform for data center infrastructure, building and power: BMS, EPMS, SCADA, NOC, DCIM and CMMS. An on-site node auto-discovers…23 growth ideas
  • Pylon growth strategyAn agentic support platform for B2B companies: one inbox for Slack, Teams, email, chat, Discord, WhatsApp and phone, with AI agents that…21 growth ideas
  • Gecko Robotics growth strategyWall-climbing inspection robots, sensors and drones that measure the condition of industrial and defense assets, plus Cantilever, a…21 growth ideas
  • bunch growth strategyAI-native fund operations for private markets: investor portal, transfer agency, fund accounting and reporting, treasury, compliance, fund…21 growth ideas

See every playbook and website review

Independent research. Programmable wrote this playbook from public sources, as they stood on November 10, 2026, and cites them throughout. Godela did not commission, review or endorse it, and Programmable is not affiliated with Godela. Company and product names are trademarks of their owners, used only to identify them. The ideas are suggestions and opinion, not professional, legal or financial advice, and not statements about how Godela runs its business.

Something wrong, or do you represent Godela? Email [email protected] and we will correct it or take it down.

30/60/90 roadmap

What to focus on in each month, and the ideas that carry it.

  1. Days 1 to 307 ideas

    Make the pilot path trustworthy: a pilot page with data handling, a qualifying intake, an accuracy number, the demo fix, a careers link, search basics and font/contrast fixes.

  2. Days 31 to 6011 ideas

    Build proof and reach: application pages, a public benchmark, supported formats, pilot scorecards and packaging, the first anonymised case study, engineering notes, hiring-signal outbound, data center partners, investor intros and a cloud marketplace listing.

  3. Days 61 to 905 ideas

    Commit to the big bets and wildcards: a live public demo, a packaged data center thermal product, an add-on to existing simulation workflows, a university program and a weekly what-if series.

The research behind the plan

3 research areas, each summarised in a few lines. Open any of them for the full detail.

Customers

What customers say

Customers value answers in seconds instead of days of solver time. The most cited problem: weeks or months of simulation and parameter tuning per question.

  • Asking what-if questions in plain language
  • Questions that never get asked because each one is too expensive
  • In their words: “simulation bottleneck”

Market

Competitive landscape

4 competitors compared: Neural Concept, PhysicsX, Monolith and Incumbent simulation tools.

  • Gap to exploit: Published, reproducible accuracy figures for AI surrogates
  • Neural Concept: Enterprise-heavy sales motion that is slow for smaller teams
  • PhysicsX: Broad scope with less focus on a single quick answer

Hiring signals

Godela (about 10 people) is hiring three founding engineers in San Francisco, all with visa sponsorship: a simulation engineer who owns validation against ground truth and builds CFD/FEA data pipelines (ideas 3, 9, 10…

  • Founding Simulation Engineer: Validation against simulation and experimental ground truth; CFD, FEA and multiphysics data. Ideas 3, 9, 10.
  • Founding Engineer (SWE): User-facing applications and APIs on top of the models. Ideas 5, 19.
  • Founding Engineer (ML Systems): Multi-GPU training of large physics-informed models on AWS. Ideas 18, 19.

From the team that wrote this playbook

All products
  • ProgrammableAgencyWorking AI systems in days, not quarters (opens programmable.agency in a new tab)
  • ProgrammableInfoAn agent that finds, researches and asks firstNew accounts get 100,000 free enrichment credits with code BETA100K (opens programmable.info in a new tab)
  • ProgrammableDigitalScreenshots in, an animated product demo out (opens programmable.digital in a new tab)
  • ProgrammableWatchKnow the day a supplier falls out of compliance (opens programmable.watch in a new tab)
  • ProgrammableTalkGet registrants for your events, and know who’s coming (opens programmable.talk in a new tab)
  • ProgrammableMediaAn AI note taker, interpreter and webinar host for your calls (opens programmable.media in a new tab)
  • ProgrammablePageText what you need, your customer gets a page that works (opens programmable.page in a new tab)
  • ProgrammableCommunityForums with labelled AI members, never empty (opens programmable.community in a new tab)
ProgrammableMarketing
ExamplesLeaderboardsProductsTermsPrivacyCookiesDPALegalContact

© 2026 AI-Powered Growth Intelligence