ProgrammableMarketing/hud.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 9, 2026 1 view
  1. Programmable
  2. Playbooks
  3. HUD
HUD logo

growth playbook · Nov 9, 2026

hud.ai (opens in a new tab)

HUD growth playbook

Diagnosis

HUD (Human Union Data, YC W25, $16M Series A) has grown fast since June: environments built went from 2,500+ to 13,764 and task runs from 1.3M to 4.8M. It is hiring 38 people, with a whole GTM team (outbound AEs and SDRs in San Francisco and Singapore) whose job is to get businesses to list data on the marketplace, and a CMO to position HUD for both lab buyers and data vendors.

The website hasn't caught up with this two-sided business. Seven guides written for people who want to sell data or code end with "Start building for free", a call to action for environment builders, so sellers who arrive from search get no way to list. "Free" signups start with zero credits.

There is no public pricing page, both customer stories say "Coming soon", and the homepage fails Core Web Vitals. This playbook first fixes the paths each audience takes to act (sellers, builders, buyers), then builds proof, comparison and directory pages, then makes three bigger bets: a public quality index, a packaged enterprise benchmark and expert-network supply partnerships.

Growth Score

7/10

Biggest Opportunity

Turn the inbound seller traffic HUD already earns from its guides into marketplace listings, so the new outbound team can work warm leads as well as cold ones, and make the builder path free to the first trace.

Biggest Risk

Two audiences share one site and one message. If vendors, builders and lab buyers all get the same "Start building" path, each converts worse, while Prime Intellect's 2,500+ environment hub and open-source tools compete for the same builders.

01

Start here

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

  1. 1

    Give the seven "sell your data" guides a seller call to action instead of "Start building for free"

    Swap the closing block on "How to Sell Source Code to AI Companies" for a seller CTA and a four-field form.

    High impact First result in 2 weeks
  2. 2

    Give new teams enough starter credits to finish the ten-minute quickstart, so "Start building for free" is true

    Price out one quickstart run and QA review, then set the starter balance at three times that.

    High impact First result in 2 weeks
  3. 3

    Publish a public pricing page: hud.ai/pricing currently opens the sign-in screen

    Turn the billing docs into a page draft with three worked examples.

    Medium 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.

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

    1 of these are things Programmable can build and run for you.

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 · 2Big bets · 9Fill-ins · 5Reconsider · 7
Effort →

Showing 23 of 23 ideas

This week 7

Quick wins with a clear first action.

  • 01

    Seven of HUD's 20 guides are written for people who want to sell data or code to AI labs, for example "How to Sell Source Code to AI Companies", which quotes about $30,000 per license and deals above $10 million. A reader who gets to the end of that guide ready to sell is shown "Build environments that frontier labs trust" and "Start building for free", the call to action for environment builders. The one action link in the Acquire.com alternatives guide is a plain text link to datavendor.ai. Meanwhile HUD is hiring outbound AEs and SDRs to cold-call businesses and get them to "list useful data on HUD". The warmest sellers HUD has, the ones who searched for how to sell and found HUD, get the wrong next step.

    High impactLow effort First result in 2 weeks Product Manager

    First action

    Swap the closing block on "How to Sell Source Code to AI Companies" for a seller CTA and a four-field form.

    How to measure it

    Seller-guide leads that submit a listing

    Now
    –
    Target
    40 a month
    By
    60 days

    Steps

    1. 1

      Tag the seven guides

      Mark them as seller content in the CMS so they get a different closing component.

      Product Engineer, Frontend & Design 1 day

    2. 2

      Write the seller CTA

      Headline, one line on licensing without selling the company, and a button. No credit card mention.

      Chief Marketing Officer 1 day

    3. 3

      Build the short form

      Asset type, rough size, domain or language, rights confirmation, email.

      Product Engineer 2 days

    4. 4

      Route to outbound

      New leads go to the SDR queue with the guide name and answers attached; follow up within one working day.

      GTM Engineer 2 days

    5. 5

      Add a mid-article box

      Place it after the pricing section, where intent peaks.

      Product Engineer, Frontend & Design 1 day

    6. 6

      Measure by guide

      Views, form starts, leads and submitted listings per guide.

      GTM Engineer weekly

    First result
    2 weeks
    Cost
    A few days of frontend and GTM engineering time
    Owner
    Product Manager, Marketplace

    Smallest test

    Change the closing block on one guide for two weeks and count the form submissions.

    Why it works

    Replace the closing block on the seven seller guides (and add a mid-article box) with a seller call to action: "Find out what your codebase or dataset is worth to AI labs". It leads to a short form (asset type, size, language or domain, ownership, contact) that creates a lead for the outbound team, plus a direct link to start a DataVendor listing. Keep "Start building for free" on the builder guides. Tag the guide each lead came from so the GTM Engineer's CRM shows which content creates listings.

    Search intent is the strongest lead signal there is. Someone who read a 3,000-word guide on licensing their code is far warmer than a cold call, and the outbound team already has the scripts to take that seller from intent to a submitted listing.

    Ready to use

    CopySeller call to action for the seller guides

    Heading: Find out what your code or data is worth to AI labs
    Line: License it without selling your company. We grade it, list it on DataVendor and bring buyers.
    Button: Get my asset reviewed
    Form: What is it? (codebase / dataset / environment / other) · Rough size · Language or domain · I own the rights (yes/no) · Work email
    Small link: Building RL environments instead? Start building for free

    Risks and how to handle them

    • Low-quality submissions waste SDR time

      Ask about rights and size up front and auto-reply to assets that won't sell.

    • Builders landing on seller guides lose their path

      Keep a small "Building environments instead?" link under the seller CTA.

    Evidence

    • Your websiteGuides (20, checked November 9, 2026): seven are for people selling data or code, not building environments: "How to Sell Data to Model Labs", "How to Sell Source Code to AI Companies", "How to Sell Startup Assets to AI Labs", "How AI Data Licensing Works", "How AI Data Marketplaces Work", "Acquire.com Alternatives for Founders Who Can't (or Won't) Sell the Whole Company" and "Best Platforms for Selling Your Codebase in 2026". The guides index and every guide checked end with the same block: "Build environments that frontier labs trust", "Start building for free", "Book a call".Source
    • Your website"How to Sell Source Code to AI Companies" (September 25, 2026): a good codebase starts at about "$30,000 per license" on DataVendor and some single-codebase deals have passed $10 million; readers are told to "add it to DataVendor Inventory"; the page then closes with "Build environments that frontier labs trust" and "Start building for free".Source
    • Your website"Acquire.com Alternatives for Founders Who Can't (or Won't) Sell the Whole Company" (July 7, 2026): HUD is Option 1 for selling code and IP; the only action link is "See how HUD grades and sells codebases at" datavendor.ai; the page closes with "Start building for free" and "Book a call".Source
    • HiringOutbound Account Executive (remote, Pacific hours; also Singapore and SGT versions): persuade businesses and individuals to "list useful data on HUD", following buyer demand "usually from frontier AI labs", with daily cold calls, emails and LinkedIn messages and weekly listing targets; USD 60,000-90,000 on-target pay. Three Sales Development Representative roles sit alongside.Source
  • 02

    The homepage says "Start building for free" and "No credit card required". The quickstart then says runs and QA reviews are paid from the team's credit balance "and a new team starts with none", so the first step after signup is to add credits or upgrade to Pro. The promise breaks before the new user sees a single trace.

    High impactLow effort First result in 2 weeks Product Manager

    First action

    Price out one quickstart run and QA review, then set the starter balance at three times that.

    How to measure it

    Signup-to-first-trace rate

    Now
    –
    Target
    +50%
    By
    45 days

    Steps

    1. 1

      Cost the quickstart

      Runtime, model calls and QA for one run of the template task.

      Product Manager, Platform 1 day

    2. 2

      Set the grant

      One per team, expiring in 30 days, limited to cheaper models.

      Product Manager, Platform 1 day

    3. 3

      Build it

      Credit the balance on team creation and show it in the app header.

      Product Engineer 3 days

    4. 4

      Guard against abuse

      One grant per verified email domain or GitHub account.

      Senior Software Engineer, Infrastructure 2 days

    5. 5

      Update the docs

      Change the "starts with none" line in the quickstart.

      Technical Writer 1 hour

    6. 6

      Nudge to paid

      An in-app message at 20% left, offering Pro or Pay as you go.

      Product Manager, Platform 2 days

    7. 7

      Measure

      First trace, second session and paid conversion by cohort.

      GTM Engineer weekly

    First result
    2 weeks
    Cost
    A few dollars of compute per signup
    Owner
    Product Manager, Platform

    Smallest test

    Grant starter credit to half of new teams for two weeks and compare first-run rates.

    Why it works

    Give every new team a small, capped starter balance that covers the quickstart (one task, one run, one QA review) plus a few more runs. Use cheap default models and short environment time limits. Show the balance in the app and nudge the team to Pay as you go or Pro when it runs low.

    Developer tools win when the first result comes before the first payment. A trace and a QA verdict on your own task is HUD's "aha" moment, and it costs cents to provide.

    Risks and how to handle them

    • Free-credit farming

      Cap per account, limit models and expire unused credit.

    • Grant cost grows with signups

      Review the cost per paid conversion monthly.

    Evidence

    • Your websiteHomepage (checked November 9, 2026): H1 "The platform for building RL environments"; counters of 4,831,034 task runs, 13,764 environments built and 22,297,355 inference calls; "Start building for free" and "Create your first environment in minutes. No credit card required."; a "Turn tasks into revenue" section for listing on DataVendor, "where AI labs buy training and evaluation data"; two customer stories (Sharpe and UiPath), both "Coming soon"; FAQ includes "Can I monetize environments I build?"; contact paths are "Book a call" and a founders email.Source
    • Your websiteDocs quickstart: creates, runs and reviews a task "in about ten minutes", but "Runs and QA reviews are paid from your team's credit balance, and a new team starts with none"; the next step offered is "Add credits" or "Upgrade to Pro for 50 credits each month".Source
    • Your websiteBilling docs: Pay as you go ("No subscription"), Pro at $50 a month adding $50 of credits, and Enterprise at custom pricing; credits pay for environment runtime by the hour, model calls at the model's price plus a 5% platform fee (5% of the equivalent cost with your own provider key) and QA runs; no free tier is described. hud.ai/pricing opens the sign-in page.Source
  • 03

    HUD's pricing is simple and already written down in the billing docs: Pay as you go, Pro at $50 a month with $50 of credits, Enterprise custom, and a 5% platform fee on model calls. But hud.ai/pricing opens the sign-in page and the sitemap has no pricing page, so evaluators and procurement teams can't find it.

    Medium impactLow effort First result in 2 weeks Product Manager

    First action

    Turn the billing docs into a page draft with three worked examples.

    How to measure it

    Visitors to /pricing who add credits or upgrade within 14 days

    Now
    –
    Target
    5%
    By
    60 days

    Steps

    1. 1

      Draft from the docs

      Plans, credit uses and the 5% fee.

      Product Manager, Platform 1 day

    2. 2

      Add worked examples

      Three common jobs with estimated cost ranges.

      Research Engineer, Benchmarks 1 day

    3. 3

      Design the page

      Three columns plus an FAQ.

      Design Engineer 2 days

    4. 4

      Release /pricing

      Replace the sign-in route for logged-out visitors.

      Product Engineer, Frontend & Design 1 day

    5. 5

      Link it

      Top navigation, footer and the docs billing page.

      Product Engineer, Frontend & Design 1 hour

    6. 6

      Track

      Pricing views to upgrades and Enterprise enquiries.

      GTM Engineer weekly

    First result
    2 weeks
    Cost
    Staff time only
    Owner
    Product Manager, Platform

    Smallest test

    Publish the plan table alone and watch Enterprise enquiries.

    Why it works

    Publish /pricing with the three plans, what uses credits (runtime, model calls, QA), the 5% fee with and without your own key, three worked examples (a 100-task eval, a weekly regression run, a QA pass) and an Enterprise contact form. Link it from the top navigation and the footer.

    Technical buyers check pricing before they book a call. A transparent page with the 5% fee spelled out removes the fear of hidden costs and gives AEs a link to send.

    Risks and how to handle them

    • Example costs go stale as model prices change

      Show ranges and a date, and review them monthly.

    Evidence

    • Your websiteBilling docs: Pay as you go ("No subscription"), Pro at $50 a month adding $50 of credits, and Enterprise at custom pricing; credits pay for environment runtime by the hour, model calls at the model's price plus a 5% platform fee (5% of the equivalent cost with your own provider key) and QA runs; no free tier is described. hud.ai/pricing opens the sign-in page.Source
    • Your websiteSitemap (37 URLs, checked November 9, 2026): homepage, careers, 2 blog posts, 7 research posts, 20 guides and legal pages; no pricing, customer story, comparison or environment pages.Source
  • 04

    HUD serves three audiences (lab and enterprise buyers, data vendors and environment builders), but the site gives all of them one "Book a call" button and a founders email. The Customer Operations Lead is hired to onboard both vendors and buyers, and the outbound team works supply. Every call starts with triage.

    Medium impactLow effort First result in 2 weeks Customer Operations Lead

    First action

    Count last month's calls by audience to size each path.

    How to measure it

    Call-to-opportunity rate

    Now
    –
    Target
    +25%
    By
    60 days

    Steps

    1. 1

      Size the mix

      Tag last month's calls as buyer, seller or builder.

      Customer Operations Lead 1 day

    2. 2

      Write the paths

      Two qualifying questions per path.

      Customer Operations Lead 1 day

    3. 3

      Build the chooser

      A modal on every "Book a call" button.

      Product Engineer, Frontend & Design 2 days

    4. 4

      Route calendars

      Each path books with the right owner.

      GTM Engineer 1 day

    5. 5

      Log the source

      Path and answers saved to the CRM.

      GTM Engineer 1 day

    6. 6

      Review

      Conversion by path each fortnight.

      Customer Operations Lead fortnightly

    First result
    2 weeks
    Cost
    Staff time only
    Owner
    Customer Operations Lead

    Smallest test

    Add the chooser on the homepage only for two weeks.

    Why it works

    Replace the single button with a three-way chooser: "I want to buy data or evals", "I want to sell data or code", "I'm building environments". Each path asks two qualifying questions and books with the right person: buyers with founders or an AE, sellers with the outbound team, builders with a forward-deployed engineer or the docs.

    Routing at the click means shorter calls, quicker follow-up and clean source data for the CRM.

    Risks and how to handle them

    • An extra click lowers bookings

      Keep it to one click and two questions, and measure.

    Evidence

    • Your websiteHomepage (checked November 9, 2026): H1 "The platform for building RL environments"; counters of 4,831,034 task runs, 13,764 environments built and 22,297,355 inference calls; "Start building for free" and "Create your first environment in minutes. No credit card required."; a "Turn tasks into revenue" section for listing on DataVendor, "where AI labs buy training and evaluation data"; two customer stories (Sharpe and UiPath), both "Coming soon"; FAQ includes "Can I monetize environments I build?"; contact paths are "Book a call" and a founders email.Source
    • HiringProduct Manager, Marketplace: "make it easier for data vendors to bring high-quality RL environments and training data to HUD", improve vendor onboarding and how research teams discover environments, and "Define and track measures of marketplace health". A Customer Operations Lead (posted October 2, 2026) owns vendor and buyer onboarding and "successful customer activation and engagement".Source
    • HiringOutbound Account Executive (remote, Pacific hours; also Singapore and SGT versions): persuade businesses and individuals to "list useful data on HUD", following buyer demand "usually from frontier AI labs", with daily cold calls, emails and LinkedIn messages and weekly listing targets; USD 60,000-90,000 on-target pay. Three Sales Development Representative roles sit alongside.Source
  • 05

    The homepage has no meta description and no social share image, so search results and every link shared on X, LinkedIn or Discord show whatever text the platform picks. HUD's growth runs on founder posts and social, so this is the first thing most people see.

    Medium impactLow effort First result in 2 weeks Marketing & Events Lead

    First action

    Add the homepage meta description and an og:image.

    How to measure it

    Search click-through on the homepage

    Now
    –
    Target
    +10%
    By
    60 days

    Steps

    1. 1

      Write the description

      Under 155 characters, naming both sides.

      Marketing & Events Lead 1 hour

    2. 2

      Design the share image

      Logo, line and counters.

      Design Engineer 1 day

    3. 3

      Add the tags

      Meta, Open Graph and X card tags.

      Product Engineer, Frontend & Design 2 hours

    4. 4

      Template the rest

      Guides, research and careers get per-page descriptions and images.

      Product Engineer, Frontend & Design 2 days

    5. 5

      Check previews

      X, LinkedIn, Slack and Discord.

      Marketing & Events Lead 1 hour

    6. 6

      Measure

      Search Console click-through.

      Marketing & Events Lead monthly

    First result
    2 weeks
    Cost
    $0
    Owner
    Marketing & Events Lead

    Smallest test

    Add the homepage tags today.

    Why it works

    Write a description that names both sides ("Build, QA and sell RL environments and training data that frontier labs trust") and a share image with the live counters (13,764 environments, 4.8M task runs). Do the same for the guides, research and careers templates.

    A clear preview lifts click-through on every shared link at no cost, and the numbers are proof in themselves.

    Risks and how to handle them

    • The counters in the image go stale

      Generate the image from live numbers or round them.

    Evidence

    • PageSpeedGoogle PageSpeed for www.hud.ai (November 9, 2026): mobile 40/100, desktop 53; real-visitor data fails Core Web Vitals on LCP (4.7s) and INP (243ms); in the mobile lab test the main content (a hero video) appears at 10.5s; 6.2 MB over 127 requests, about 2 MB of JavaScript; savings from image delivery (~2.1s), render-blocking requests (~1.7s), unused JavaScript (~0.9s) and redirects (~0.75s). The homepage has no meta description or social share image.Source
    • HiringMarketing & Events Lead (posted April 24, 2026): events that "generate pipeline", "Ghostwrite high-quality content for founders" and "Lead social media presence on Twitter/X and LinkedIn". Growth Lead (posted July 2, 2026): "help HUD meet the right people and evangelize our vision". Technical Writer: customer and partner case studies that "turn interviews, demos, and results into stories with concrete evidence".Source
  • 06

    The homepage names two customer stories, how Sharpe scales RL environments and how UiPath benchmarks enterprise agents against every frontier model, and both say "Coming soon". The sitemap has no customer-story pages. The CMO and Technical Writer roles are both hired to produce case studies.

    High impactMedium effort First result in 1 month Technical Writer

    First action

    Book a 45-minute interview with each customer this week.

    How to measure it

    Calls booked from story pages

    Now
    –
    Target
    15 a month
    By
    90 days

    Steps

    1. 1

      Interview

      One call per customer, recorded with permission.

      Technical Writer 1 week

    2. 2

      Pull the numbers

      Tasks, runs and QA findings from the platform, with the customer's approval.

      Forward Deployed Research Engineer 3 days

    3. 3

      Write

      Problem, build, results, quote.

      Technical Writer 1 week

    4. 4

      Approve

      Customer sign-off on every number.

      Customer Operations Lead 1 week

    5. 5

      Publish

      /customers/<name> pages, added to the sitemap, homepage cards switched to live.

      Product Engineer, Frontend & Design 1 day

    6. 6

      Distribute

      Founder posts and an outbound sequence.

      Marketing & Events Lead 1 week

    First result
    1 month
    Cost
    Staff time only
    Owner
    Technical Writer

    Smallest test

    Replace one "Coming soon" card with a three-paragraph story.

    Why it works

    Publish both stories as full pages: the problem, the environments built, the numbers (tasks, runs, issues caught by QA, time saved) and a quote. Give each its own URL and add them to the sitemap. Replace the "Coming soon" cards with the live stories.

    Lab and enterprise buyers want proof from teams like themselves, and a "Coming soon" card on the homepage reads as no proof at all.

    Risks and how to handle them

    • Customer approval takes too long

      Publish the shorter story first and add numbers later.

    Evidence

    • Your websiteHomepage (checked November 9, 2026): H1 "The platform for building RL environments"; counters of 4,831,034 task runs, 13,764 environments built and 22,297,355 inference calls; "Start building for free" and "Create your first environment in minutes. No credit card required."; a "Turn tasks into revenue" section for listing on DataVendor, "where AI labs buy training and evaluation data"; two customer stories (Sharpe and UiPath), both "Coming soon"; FAQ includes "Can I monetize environments I build?"; contact paths are "Book a call" and a founders email.Source
    • Your websiteSitemap (37 URLs, checked November 9, 2026): homepage, careers, 2 blog posts, 7 research posts, 20 guides and legal pages; no pricing, customer story, comparison or environment pages.Source
    • HiringChief Marketing Officer: "Develop and refine HUD's positioning for frontier-lab buyers and data vendors" and own case studies, partnerships, launches and announcements. Every post says HUD is "Backed by $16M from top VCs and YC (W25)", has "~25 people currently", is "scaling profitably and quickly to meet very strong demand", and that its marketplace and platform "are used by startups, Fortune 500 companies and frontier labs".Source
  • 07

    Real-visitor data for www.hud.ai fails Core Web Vitals on both loading (LCP 4.7s) and responsiveness (INP 243ms). On a phone the largest element, a hero video, appears at 10.5s in Google's lab test, and the page weighs 6.2 MB over 127 requests. PageSpeed lists about 2.1s of image savings, 1.7s of render-blocking requests, 0.9s of unused JavaScript and 0.75s of redirects.

    Medium impactMedium effort First result in 1 month Product Engineer

    First action

    Swap the hero video for a poster image plus a deferred video and rerun PageSpeed.

    How to measure it

    Core Web Vitals pass

    Now
    –
    Target
    LCP under 2.5s and INP under 200ms
    By
    60 days

    Steps

    1. 1

      Poster first

      A compressed still as the LCP element.

      Product Engineer, Frontend & Design 1 day

    2. 2

      Defer the video

      Load after first paint, and not at all on slow connections.

      Product Engineer, Frontend & Design 1 day

    3. 3

      Fix images

      AVIF/WebP with explicit sizes.

      Design Engineer 1 day

    4. 4

      Trim JavaScript

      Ship static sections without JavaScript and defer the tag manager.

      Product Engineer, Frontend & Design 3 days

    5. 5

      Cut redirects

      Link and redirect straight to the final host.

      Senior Software Engineer, Infrastructure 2 hours

    6. 6

      Verify

      PageSpeed weekly and field data after 28 days.

      Product Engineer, Frontend & Design 4 weeks

    First result
    1 month
    Cost
    About a week of frontend time
    Owner
    Product Engineer, Frontend & Design

    Smallest test

    Poster plus deferred video on the homepage only.

    Why it works

    Show a compressed poster image as the LCP element and start the video only after load. Serve modern image formats, defer non-critical scripts and the tag manager, split the JavaScript per section, and send visitors straight to the final URL without a redirect chain.

    A large share of first visits come from shared links on phones. A faster first screen keeps those visitors and helps search rankings for the guides.

    Risks and how to handle them

    • A static hero looks less alive

      Keep the video, just load it after the poster.

    Evidence

    • PageSpeedGoogle PageSpeed for www.hud.ai (November 9, 2026): mobile 40/100, desktop 53; real-visitor data fails Core Web Vitals on LCP (4.7s) and INP (243ms); in the mobile lab test the main content (a hero video) appears at 10.5s; 6.2 MB over 127 requests, about 2 MB of JavaScript; savings from image delivery (~2.1s), render-blocking requests (~1.7s), unused JavaScript (~0.9s) and redirects (~0.75s). The homepage has no meta description or social share image.Source

This quarter 11

The core plays for the next three months.

  • 08

    HUD's own guide already ranks the platforms for publishing RL environments (hud.ai, Harbor, Prime Intellect, Gymnasium, RLlib), but there's no head-to-head comparison, and the sitemap has no comparison pages. Prime Intellect promotes an Environment Hub of 2,500+ community environments next to its compute and training.

    High impactMedium effort First result in 6 weeks Chief Marketing Officer

    First action

    Ask the outbound team which alternative prospects name most often.

    How to measure it

    Comparison-page signups

    Now
    –
    Target
    100 a month
    By
    1 quarter

    Steps

    1. 1

      Pick rivals

      From calls and the existing guide.

      Chief Marketing Officer 1 day

    2. 2

      Check facts

      Each rival's current features and pricing from its own site.

      Research Engineer, Benchmarks 3 days

    3. 3

      Write

      A table, when each tool is the better choice, and migration steps.

      Technical Writer 2 weeks

    4. 4

      Review

      Claims and trademarks.

      Business Operations 3 days

    5. 5

      Publish

      Linked from the guides and docs.

      Product Engineer, Frontend & Design 1 day

    6. 6

      Refresh

      Every quarter.

      Technical Writer quarterly

    First result
    6 weeks
    Cost
    Staff time only
    Owner
    Chief Marketing Officer

    Smallest test

    Publish HUD vs Harbor first, since HUD already imports Harbor tasks.

    Why it works

    Write three fair pages: HUD vs Prime Intellect Environment Hub, HUD vs Harbor (and how to bring Harbor tasks into HUD) and HUD vs Gymnasium for LLM RL. Each covers hosting, QA for reward hacking, team review, the marketplace, pricing and a migration guide. Date every page and cite each rival's own pages.

    Comparison searches come from people who are already choosing. HUD's QA agents and marketplace are clear points of difference that a table makes obvious.

    Risks and how to handle them

    • Unfair or outdated claims hurt credibility with researchers

      Say where the rival is better and date every table.

    Evidence

    • Competitor researchPrime Intellect's homepage: "The Open Superintelligence Stack", with hosted RL training, sandboxes, an Environment Hub of 2,500+ community environments and GPUs by the hour (H100 at $3.50/hr). HUD's own guide to publishing RL environments ranks hud.ai, Harbor Framework, Prime Intellect, Gymnasium and RLlib, with no head-to-head comparison table.Source
    • Your websiteSitemap (37 URLs, checked November 9, 2026): homepage, careers, 2 blog posts, 7 research posts, 20 guides and legal pages; no pricing, customer story, comparison or environment pages.Source
    • HiringChief Marketing Officer: "Develop and refine HUD's positioning for frontier-lab buyers and data vendors" and own case studies, partnerships, launches and announcements. Every post says HUD is "Backed by $16M from top VCs and YC (W25)", has "~25 people currently", is "scaling profitably and quickly to meet very strong demand", and that its marketplace and platform "are used by startups, Fortune 500 companies and frontier labs".Source
  • 09

    HUD's guides say good codebases start at about $30,000 per license and talk about how HUD "grades and sells codebases", but a founder can't see a grade without talking to sales. The GTM Engineer is hired to build systems that "find specific kinds of data supply at scale". A self-serve grade would bring that supply in by itself.

    High impactMedium effort First result in 6 weeks GTM Engineer

    First action

    Write the grading rubric from the last 20 codebases sold or rejected.

    How to measure it

    Grade-to-listing rate

    Now
    –
    Target
    15%
    By
    1 quarter

    Steps

    1. 1

      Write the rubric

      What buyers paid for and what they rejected.

      Lead Research Engineer, Data Quality 1 week

    2. 2

      Set ranges

      Wide, honest price bands per grade.

      Business Development Lead 3 days

    3. 3

      Build the form

      Questionnaire first; read-only repo connection later.

      Product Engineer 2 weeks

    4. 4

      Score and route

      High grades booked with outbound, low grades sent tips.

      GTM Engineer 1 week

    5. 5

      Put it in the guides

      Inside the seller CTA from idea 1.

      Product Engineer, Frontend & Design 1 day

    6. 6

      Calibrate

      Compare grades with actual sales each month.

      Lead Research Engineer, Data Quality monthly

    First result
    6 weeks
    Cost
    About three weeks of engineering
    Owner
    GTM Engineer

    Smallest test

    A ten-question form scored by hand for two weeks.

    Why it works

    Build a free tool on the seller guides: the founder describes the repo (languages, size, age, tests, domain, contributors), or connects a read-only repo, and gets a grade and a likely license range with the reasons. High grades get a booked call with the outbound team. Low grades get tips on what would raise the value.

    "What is it worth?" is the question every seller asks first. Answering it up front turns a cold call into a warm follow-up and gives each lead a score.

    Risks and how to handle them

    • Price ranges taken as a promise

      Show bands, the reasons, and that the final price depends on buyer demand.

    • Founders hesitate to share code

      Start with a questionnaire that needs no code.

    Builds on

    Evidence

    • Your website"How to Sell Source Code to AI Companies" (September 25, 2026): a good codebase starts at about "$30,000 per license" on DataVendor and some single-codebase deals have passed $10 million; readers are told to "add it to DataVendor Inventory"; the page then closes with "Build environments that frontier labs trust" and "Start building for free".Source
    • Your website"Acquire.com Alternatives for Founders Who Can't (or Won't) Sell the Whole Company" (July 7, 2026): HUD is Option 1 for selling code and IP; the only action link is "See how HUD grades and sells codebases at" datavendor.ai; the page closes with "Start building for free" and "Book a call".Source
    • HiringGTM Engineer (posted July 13, 2026): "Build sourcing systems, such as scrapers and enrichment pipelines, that find specific kinds of data supply at scale" and "connect our CRM, marketplace and product data into a clean, reliable source of truth for pipeline, listings and revenue".Source
  • 10

    The GTM Engineer is hired to connect CRM, marketplace and product data into one "source of truth for pipeline, listings and revenue", while the outbound team works to weekly listing targets. Without the source of each lead in that view, HUD can't tell whether a guide, an event or a cold call produced a listing.

    Medium impactMedium effort First result in 1 month GTM Engineer

    First action

    Add a source field to listings and fill it in for every listing created since July.

    How to measure it

    Listings with source attribution

    Now
    –
    Target
    95%
    By
    60 days

    Steps

    1. 1

      Define sources

      A fixed list of values.

      GTM Engineer 1 day

    2. 2

      Capture at entry

      Forms, call paths and outbound tools.

      GTM Engineer 1 week

    3. 3

      Join the data

      CRM, marketplace and payouts.

      GTM Engineer 2 weeks

    4. 4

      Backfill

      Recent listings tagged by hand.

      GTM Associate - Singapore 1 week

    5. 5

      Dashboard

      Listings, sales and revenue by source.

      GTM Engineer 1 week

    6. 6

      Monthly review

      Shift effort to the best sources.

      Business Development Lead monthly

    First result
    1 month
    Cost
    Staff time and existing tools
    Owner
    GTM Engineer

    Smallest test

    Tag only the seller-guide leads for one month.

    Why it works

    Give every lead a source (guide name, call path, event, outbound list, referral), carry it through listing submission, approval, first sale and payout, and build one dashboard by source and by asset type. Use it to set the content and outbound plans each month.

    A two-sided marketplace grows by putting effort where supply turns into sales. This shows which sources produce listings that actually sell.

    Risks and how to handle them

    • Messy self-reported sources

      Prefer tracked links and only ask when no tracking exists.

    Builds on

    Evidence

    • HiringGTM Engineer (posted July 13, 2026): "Build sourcing systems, such as scrapers and enrichment pipelines, that find specific kinds of data supply at scale" and "connect our CRM, marketplace and product data into a clean, reliable source of truth for pipeline, listings and revenue".Source
    • HiringOutbound Account Executive (remote, Pacific hours; also Singapore and SGT versions): persuade businesses and individuals to "list useful data on HUD", following buyer demand "usually from frontier AI labs", with daily cold calls, emails and LinkedIn messages and weekly listing targets; USD 60,000-90,000 on-target pay. Three Sales Development Representative roles sit alongside.Source
  • 11

    HUD's research is already co-authored with Sixtyfour, Browserbase, Dreamscale and Sepal AI, and on November 6 Toptal's ExpertCrowd and HUD released the ExpertCrowd-SWE benchmark as "the first release of an ongoing collaboration". Each release earned partner reach, but it happened deal by deal.

    High impactMedium effort First result in 2 months GTM Lead

    First action

    Draft a one-page partner offer from the ExpertCrowd-SWE release.

    How to measure it

    Partner benchmark releases

    Now
    –
    Target
    3 a quarter
    By
    2 quarters

    Steps

    1. 1

      Write the offer

      Who brings what, timeline and the joint announcement.

      GTM Lead, Commercial Partnerships 3 days

    2. 2

      Partner page

      Past releases and an application form.

      Product Engineer, Frontend & Design 1 week

    3. 3

      Pipeline

      Agent tool companies, expert networks, vertical software.

      GTM Lead, Commercial Partnerships 2 weeks

    4. 4

      Run the playbook

      Tasks, QA, results and write-up.

      Research Engineer, Benchmarks 4 weeks per release

    5. 5

      Launch together

      Joint posts and an event.

      Marketing & Events Lead per release

    6. 6

      List it

      Offer the environment on DataVendor where the partner agrees.

      Product Manager, Marketplace per release

    First result
    2 months
    Cost
    Research engineering time per release
    Owner
    GTM Lead, Commercial Partnerships

    Smallest test

    Pitch two current co-authors on a second release.

    Why it works

    Package it as "Benchmark with HUD": the partner brings tasks and experts, HUD hosts, runs QA and publishes the results, both sides announce it, and the environment can be listed on DataVendor. Publish a partner page with past releases and an application form, and aim for one release a month.

    Every co-published benchmark reaches the partner's audience, shows off HUD's QA, and puts new environments on the marketplace.

    Risks and how to handle them

    • Low-quality partner tasks hurt HUD's name

      Every release must pass HUD's QA agents and a research review.

    Evidence

    • Your websiteResearch page: 7 posts, including RECON (October 21, 2026; 22 systems tested), "Evaluating Robot Models in the Cloud" (October 16), "Can you trust your browser eval?" (September 28), "What is a good task?" (September 22) and Assemble Bench for robot assembly (July 31), co-authored with Sixtyfour, Browserbase, Dreamscale and Sepal AI.Source
    • PressToptal press release (November 6, 2026): Toptal's ExpertCrowd and HUD released ExpertCrowd-SWE, a software engineering benchmark; Toptal's experts wrote the tasks and grading rubrics, HUD hosted the environments and verified the benchmark against its QA pipelines; it is "the first release of an ongoing collaboration" with more initiatives in development.Source
  • 12

    HUD's homepage reports 13,764 environments built and shows community environments for coding, browser, computer use, deep research, robotics and ARC-AGI-3. None of them has a page in the 37-URL sitemap, while Prime Intellect markets an Environment Hub of 2,500+ community environments.

    High impactMedium effort First result in 2 months Product Manager

    First action

    Pick the 20 most-run public environments for the first pages.

    How to measure it

    Environment-page signups

    Now
    –
    Target
    300 a month
    By
    1 quarter

    Steps

    1. 1

      Pick 20

      Most-run public environments.

      Product Manager, Platform 1 day

    2. 2

      Design the template

      Description, tasks, scores, QA badge, run command.

      Design Engineer 1 week

    3. 3

      Generate pages

      From environment metadata.

      Product Engineer 2 weeks

    4. 4

      Write intros

      A human-written summary for each of the 20.

      Technical Writer 1 week

    5. 5

      Add to sitemap

      And link from the guides.

      Product Engineer, Frontend & Design 1 day

    6. 6

      Expand

      Open it to builders who want their environment listed.

      Product Manager, Platform monthly

    First result
    2 months
    Cost
    About three weeks of engineering
    Owner
    Product Manager, Platform

    Smallest test

    Five hand-made pages for the homepage categories.

    Why it works

    Publish a page for each public environment (what it tests, example tasks, the models run on it with scores, QA status, how to fork it or run it in one command), grouped by category. Add them to the sitemap and link them from the guides.

    Researchers search for environments by task ("computer use environment", "spreadsheet agent benchmark"). Catalogue pages win those searches and turn browsing into forks and runs.

    Risks and how to handle them

    • Thin auto-generated pages

      Only publish environments with a description, tasks and scores.

    Evidence

    • Your websiteHomepage (checked November 9, 2026): H1 "The platform for building RL environments"; counters of 4,831,034 task runs, 13,764 environments built and 22,297,355 inference calls; "Start building for free" and "Create your first environment in minutes. No credit card required."; a "Turn tasks into revenue" section for listing on DataVendor, "where AI labs buy training and evaluation data"; two customer stories (Sharpe and UiPath), both "Coming soon"; FAQ includes "Can I monetize environments I build?"; contact paths are "Book a call" and a founders email.Source
    • Your websiteSitemap (37 URLs, checked November 9, 2026): homepage, careers, 2 blog posts, 7 research posts, 20 guides and legal pages; no pricing, customer story, comparison or environment pages.Source
    • Competitor researchPrime Intellect's homepage: "The Open Superintelligence Stack", with hosted RL training, sandboxes, an Environment Hub of 2,500+ community environments and GPUs by the hour (H100 at $3.50/hr). HUD's own guide to publishing RL environments ranks hud.ai, Harbor Framework, Prime Intellect, Gymnasium and RLlib, with no head-to-head comparison table.Source
  • 13

    HUD is hiring an Outbound AE, an SDR and a GTM Associate on-site in Singapore, plus remote SGT roles, and most engineering roles list Singapore and Remote (Asia). But its events and founder content are run from San Francisco. The Asia team will mostly be cold-calling.

    Medium impactMedium effort First result in 1 month Marketing & Events Lead

    First action

    Book a venue and a co-host for the first evening.

    How to measure it

    Asia-sourced listings

    Now
    –
    Target
    20 a quarter
    By
    2 quarters

    Steps

    1. 1

      Find a co-host

      A local AI community or university lab.

      Marketing & Events Lead 2 weeks

    2. 2

      Design the night

      A talk, a hands-on build, a vendor Q&A.

      Forward Deployed Research Engineer 1 week

    3. 3

      Office hours

      Weekly on Discord in Asia hours.

      Delivery Engineer weekly

    4. 4

      Promote

      Local channels and founder posts.

      GTM Associate - Singapore 2 weeks

    5. 5

      Follow up

      Attendees contacted within 48 hours.

      Outbound Account Executive - Singapore per event

    6. 6

      Measure

      Leads, signups and listings per event.

      GTM Engineer monthly

    First result
    1 month
    Cost
    A few thousand dollars per event
    Owner
    Marketing & Events Lead

    Smallest test

    One evening and four office hours, then measure.

    Why it works

    Run a monthly "build an RL environment" evening in Singapore with a local co-host, weekly office hours in Asia time zones on Discord, and a Singapore page that gathers them. Invite data vendors as well as builders, so the outbound team meets suppliers in person.

    In-person technical events build trust quickly with both builders and data owners, and they give a new regional team warm leads from its first month.

    Risks and how to handle them

    • Low turnout at first

      Co-host with an established community.

    Evidence

    • HiringCareers page: 38 open roles. 14 in Engineering & Research, including Research Engineer roles for Robotics Data, Robotics Evals, Benchmarks, and Privacy and Anonymization; 11 in GTM, including an Outbound Account Executive, a Sales Development Representative and a GTM Associate on-site in Singapore; 5 in Design & Product (Product Manager, Marketplace and Product Manager, Platform); 3 in Marketing; 4 in Operations.Source
    • HiringMarketing & Events Lead (posted April 24, 2026): events that "generate pipeline", "Ghostwrite high-quality content for founders" and "Lead social media presence on Twitter/X and LinkedIn". Growth Lead (posted July 2, 2026): "help HUD meet the right people and evangelize our vision". Technical Writer: customer and partner case studies that "turn interviews, demos, and results into stories with concrete evidence".Source
  • 14

    HUD says vendors on its marketplace are earning millions a month, and it is hiring a whole outbound team to find more of them. Vendors who have been paid know other people with sellable data, and their word carries more weight than a cold call.

    Medium impactMedium effort First result in 2 months Product Manager

    First action

    Ask the ten highest-earning vendors whether they would refer someone.

    How to measure it

    Referred listings that sell

    Now
    –
    Target
    10 a quarter
    By
    2 quarters

    Steps

    1. 1

      Interview vendors

      What reward would motivate them.

      Product Manager, Marketplace 1 week

    2. 2

      Set terms

      Paid only on the referred vendor's first sales.

      Business Operations 3 days

    3. 3

      Build links

      Referral codes in the vendor dashboard.

      Product Engineer 2 weeks

    4. 4

      Trigger at payout

      Invite sent with the first payout email.

      Customer Operations Lead 2 days

    5. 5

      Track

      In the source dashboard from idea 10.

      GTM Engineer 1 day

    6. 6

      Review

      Cost per referred sale against outbound.

      Product Manager, Marketplace quarterly

    First result
    2 months
    Cost
    A share of fees on referred sales only
    Owner
    Product Manager, Marketplace

    Smallest test

    Manual referral codes for ten vendors.

    Why it works

    After a vendor's first payout, offer a referral link: a bonus or a smaller platform fee for a set period on the referred vendor's first sales, plus a short "how I sold my data" note they can forward. Track it in the listings dashboard.

    Supply-side referrals from paid vendors come with built-in proof and cost only when a referral actually earns.

    Risks and how to handle them

    • Low-quality referrals

      Pay only after the referred listing sells.

    Builds on

    Evidence

    • Your websiteSeries A post (June 30, 2026): $16M led by Standard Capital with Y Combinator and others; 1.3M+ task runs, 2,500+ environments and 15M+ inference calls at the time; 50+ businesses on the platform; SheetBench-50 validated by PwC, Cisco, Charles Schwab and Fannie Mae; OSWorld-Verified fixed 300+ issues in the original benchmark; marketplace vendors earning millions a month; the goal is "to let anyone, anywhere build and sell frontier-grade RL environments and post-training data".Source
    • HiringOutbound Account Executive (remote, Pacific hours; also Singapore and SGT versions): persuade businesses and individuals to "list useful data on HUD", following buyer demand "usually from frontier AI labs", with daily cold calls, emails and LinkedIn messages and weekly listing targets; USD 60,000-90,000 on-target pay. Three Sales Development Representative roles sit alongside.Source
  • 15

    HUD bills by usage (environment runtime, model calls plus 5%, QA runs) with Pro at $50 a month and auto top-up. Usage-billed teams churn quietly when they stop running tasks, and HUD's QA agents produce findings (reward hacking, false positives, grader mistakes) that a team may never open.

    Medium impactLow effort First result in 1 month Product Manager

    First action

    Write the email from one active team's data by hand.

    How to measure it

    Week-8 active paying teams

    Now
    –
    Target
    +15%
    By
    1 quarter

    Steps

    1. 1

      Pick the signals

      Runs, spend, credit left, QA findings.

      Product Manager, Platform 2 days

    2. 2

      Draft the email

      One screen with one next step.

      Technical Writer 2 days

    3. 3

      Build the job

      Weekly per team, with opt-out.

      Product Engineer 1 week

    4. 4

      Alert on silence

      No runs in 14 days goes to Customer Operations.

      GTM Engineer 2 days

    5. 5

      Test

      Half of teams for a month.

      Product Manager, Platform 1 month

    6. 6

      Roll out

      All paying teams.

      Product Manager, Platform 1 day

    First result
    1 month
    Cost
    About a week of engineering
    Owner
    Product Manager, Platform

    Smallest test

    Hand-written digests to 20 teams for three weeks.

    Why it works

    Send a weekly email per team: runs and spend this week, credit left, the top QA findings with links to the traces, and one suggested next step (fix a flagged task, add a model, schedule a regression run). Flag teams that go quiet to Customer Operations.

    Showing the value delivered each week, and the problems QA caught, keeps teams running. Spotting silence early lets Customer Operations step in before a team churns.

    Risks and how to handle them

    • Email fatigue

      Skip weeks with no activity and let teams choose the frequency.

    Evidence

    • Your websiteBilling docs: Pay as you go ("No subscription"), Pro at $50 a month adding $50 of credits, and Enterprise at custom pricing; credits pay for environment runtime by the hour, model calls at the model's price plus a 5% platform fee (5% of the equivalent cost with your own provider key) and QA runs; no free tier is described. hud.ai/pricing opens the sign-in page.Source
    • Your websiteHomepage (checked November 9, 2026): H1 "The platform for building RL environments"; counters of 4,831,034 task runs, 13,764 environments built and 22,297,355 inference calls; "Start building for free" and "Create your first environment in minutes. No credit card required."; a "Turn tasks into revenue" section for listing on DataVendor, "where AI labs buy training and evaluation data"; two customer stories (Sharpe and UiPath), both "Coming soon"; FAQ includes "Can I monetize environments I build?"; contact paths are "Book a call" and a founders email.Source
    • HiringProduct Manager, Marketplace: "make it easier for data vendors to bring high-quality RL environments and training data to HUD", improve vendor onboarding and how research teams discover environments, and "Define and track measures of marketplace health". A Customer Operations Lead (posted October 2, 2026) owns vendor and buyer onboarding and "successful customer activation and engagement".Source
  • 16

    The hud package on PyPI is at version 0.6.22 (released November 3, 2026), but the hud-python repo (325 stars, 91 forks) has no GitHub releases, and the site has no changelog. Developers who watch the repo never see what shipped, and the stars stay low next to how much the platform is used (4.8M task runs).

    Medium impactLow effort First result in 1 month Product Engineer

    First action

    Add a release step to the publish pipeline.

    How to measure it

    GitHub stars

    Now
    –
    Target
    1,000
    By
    2 quarters

    Steps

    1. 1

      Automate releases

      One per PyPI version.

      Product Engineer 1 day

    2. 2

      Backfill

      Notes for the 0.6 line.

      Technical Writer 3 days

    3. 3

      Changelog page

      /changelog on hud.ai, generated from releases.

      Product Engineer, Frontend & Design 2 days

    4. 4

      Monthly post

      What shipped, with one example each.

      Technical Writer monthly

    5. 5

      Ask for stars

      One line in the CLI after a successful run, and in the quickstart.

      Product Engineer 1 day

    6. 6

      Share

      Release posts on X and Discord.

      Marketing & Events Lead per release

    First result
    1 month
    Cost
    $0
    Owner
    Product Engineer

    Smallest test

    Publish releases for the next four versions and watch the watcher count.

    Why it works

    Automate a GitHub release for every PyPI version, with human-written notes for minor versions. Publish a /changelog page and a monthly "what shipped" post, and ask developers to star the repo from the quickstart and the CLI.

    Visible releases signal that a project is alive and earn notifications for every watcher. Stars and changelogs are how developers judge a tool before they try it.

    Risks and how to handle them

    • Notes nobody reads

      Lead each release with one user-facing change.

    Evidence

    • Developer channelsGitHub hud-evals/hud-python: 325 stars, 91 forks, MIT licence, no GitHub releases published; the PyPI package "hud" is at 0.6.22, released November 3, 2026. One Hacker News post ("HUD: The platform for building RL environments", March 30, 2026) reached 3 points.Source
    • Your websiteHomepage (checked November 9, 2026): H1 "The platform for building RL environments"; counters of 4,831,034 task runs, 13,764 environments built and 22,297,355 inference calls; "Start building for free" and "Create your first environment in minutes. No credit card required."; a "Turn tasks into revenue" section for listing on DataVendor, "where AI labs buy training and evaluation data"; two customer stories (Sharpe and UiPath), both "Coming soon"; FAQ includes "Can I monetize environments I build?"; contact paths are "Book a call" and a founders email.Source
  • 17

    HUD's QA agents look for five failure types: failure analysis, prompt-grader alignment, false negatives, false positives and reward hacking. Its research asks "What is a good task?" and "Can you trust your browser eval?". The Marketing & Events Lead is hired to ghostwrite founder content for X and LinkedIn. One Hacker News launch reached 3 points.

    Medium impactMedium effort First result in 1 month Marketing & Events Lead

    First action

    Pick four examples from public benchmarks HUD has already fixed.

    How to measure it

    Signups from series links

    Now
    –
    Target
    150 a quarter
    By
    1 quarter

    Steps

    1. 1

      Collect examples

      Public benchmarks first.

      Research Engineer, QC Automation 1 week

    2. 2

      Format

      Hook, trace screenshot, fix.

      Marketing & Events Lead 2 days

    3. 3

      Ghostwrite

      Founder voice, reviewed by research.

      Marketing & Events Lead weekly

    4. 4

      Blog version

      Short post with a link to run the environment.

      Technical Writer weekly

    5. 5

      Quarterly guide

      Best examples as a guide page.

      Technical Writer quarterly

    6. 6

      Measure

      Reach and tracked signups.

      Growth Lead weekly

    First result
    1 month
    Cost
    Staff time only
    Owner
    Marketing & Events Lead

    Smallest test

    Four weekly posts, then compare reach with the usual posts.

    Why it works

    Every week, publish one real, anonymised example (with permission, or from a public benchmark) of an agent gaming a grader or a broken task, the trace that shows it, and the fix. Post it as a founder thread with a short write-up on the blog. Collect the best ones into a guide each quarter.

    Researchers share good failure stories, and each post shows why HUD's QA matters without a sales pitch.

    Risks and how to handle them

    • Exposing customer data

      Use public benchmarks or customer-approved, anonymised cases only.

    Evidence

    • Your websiteResearch page: 7 posts, including RECON (October 21, 2026; 22 systems tested), "Evaluating Robot Models in the Cloud" (October 16), "Can you trust your browser eval?" (September 28), "What is a good task?" (September 22) and Assemble Bench for robot assembly (July 31), co-authored with Sixtyfour, Browserbase, Dreamscale and Sepal AI.Source
    • HiringMarketing & Events Lead (posted April 24, 2026): events that "generate pipeline", "Ghostwrite high-quality content for founders" and "Lead social media presence on Twitter/X and LinkedIn". Growth Lead (posted July 2, 2026): "help HUD meet the right people and evangelize our vision". Technical Writer: customer and partner case studies that "turn interviews, demos, and results into stories with concrete evidence".Source
    • Developer channelsGitHub hud-evals/hud-python: 325 stars, 91 forks, MIT licence, no GitHub releases published; the PyPI package "hud" is at 0.6.22, released November 3, 2026. One Hacker News post ("HUD: The platform for building RL environments", March 30, 2026) reached 3 points.Source
  • 18

    The homepage's own FAQ asks "Can I monetize environments I build?" and promotes "Turn tasks into revenue" on DataVendor. But the docs never mention DataVendor or selling: the builder path ends at QA. Builders with QA-clean tasksets are exactly the supply HUD's outbound team is hired to find, and they are already inside the product.

    High impactMedium effort First result in 2 months Product Manager

    First action

    Count the tasksets that passed QA last month and whose owners never listed.

    How to measure it

    In-product listings

    Now
    –
    Target
    50 a quarter
    By
    2 quarters

    Steps

    1. 1

      Size it

      Tasksets that passed QA but weren't listed.

      Product Manager, Marketplace 2 days

    2. 2

      Design the prompt

      Shown after a clean QA run.

      Design Engineer 1 week

    3. 3

      Draft listing

      Prefilled from the taskset metadata.

      Product Engineer 2 weeks

    4. 4

      Docs path

      "Sell your environment" on the docs home.

      Technical Writer 3 days

    5. 5

      Review flow

      Rights and quality checks before going live.

      Customer Operations Lead 1 week

    6. 6

      Measure

      Prompt views, drafts, live listings and sales.

      GTM Engineer monthly

    First result
    2 months
    Cost
    About three weeks of engineering
    Owner
    Product Manager, Marketplace

    Smallest test

    Email last month's QA-clean taskset owners an invitation to list.

    Why it works

    When a taskset passes QA, show a prompt: "This taskset passed QA. List it on DataVendor?", with an estimate of buyer demand in that category and one-click creation of a draft listing. Add a "Sell your environment" path to the docs home and a section to the quickstart.

    It turns the builder base into marketplace supply at the moment of highest confidence, without a sales call.

    Risks and how to handle them

    • Enterprise teams don't want to sell their internal evals

      Show the prompt only to individual and startup workspaces, and let admins switch it off.

    Evidence

    • Your websiteDocs home: "The platform for building high-quality RL tasks and benchmarks at scale", with paths for no-code tasks, Harbor imports, the SDK, QA agents and teams; it does not mention DataVendor or selling environments.Source
    • Your websiteHomepage (checked November 9, 2026): H1 "The platform for building RL environments"; counters of 4,831,034 task runs, 13,764 environments built and 22,297,355 inference calls; "Start building for free" and "Create your first environment in minutes. No credit card required."; a "Turn tasks into revenue" section for listing on DataVendor, "where AI labs buy training and evaluation data"; two customer stories (Sharpe and UiPath), both "Coming soon"; FAQ includes "Can I monetize environments I build?"; contact paths are "Book a call" and a founders email.Source
    • HiringProduct Manager, Marketplace: "make it easier for data vendors to bring high-quality RL environments and training data to HUD", improve vendor onboarding and how research teams discover environments, and "Define and track measures of marketplace health". A Customer Operations Lead (posted October 2, 2026) owns vendor and buyer onboarding and "successful customer activation and engagement".Source

Big bets 3

Larger plays that can change the trajectory.

  • 19

    HUD's credibility comes from fixing other people's benchmarks: OSWorld-Verified fixed 300+ issues in the original, and its research asks whether browser evals can be trusted. Its QA agents already detect false positives, false negatives and reward hacking. Nobody publishes a regular, independent quality score for popular public environments and benchmarks.

    High impactHigh effort First result in 1 quarter Senior Research Manager

    First action

    Pick ten widely used public benchmarks and run QA on a sample of tasks from each.

    How to measure it

    Enterprise and lab enquiries attributed to the index

    Now
    –
    Target
    20 a quarter
    By
    3 quarters

    Steps

    1. 1

      Choose benchmarks

      The ten most-cited public evals.

      Senior Research Manager 1 week

    2. 2

      Set the method

      Sampling, detectors and human checks of every flagged task.

      Lead Research Engineer, Data Quality 2 weeks

    3. 3

      Run QA

      Agents plus human review.

      Research Engineer, QC Automation 4 weeks

    4. 4

      Contact authors

      Share the findings before publishing.

      Senior Research Manager 2 weeks

    5. 5

      Publish

      An index page, a report and fixed versions on HUD.

      Technical Writer 2 weeks

    6. 6

      Launch

      Founder threads, press and an event.

      Chief Marketing Officer 1 week

    7. 7

      Repeat

      Every quarter, with trends.

      Senior Research Manager quarterly

    First result
    1 quarter
    Cost
    Research team time and compute for QA runs
    Owner
    Senior Research Manager

    Smallest test

    Publish one benchmark's audit as a research post.

    Why it works

    Each quarter, run HUD's QA agents over the most-used public benchmarks and environments and publish the results: a score per benchmark, the share of broken or hackable tasks, the worst examples, and fixed versions people can run on HUD. Invite benchmark authors to respond before publishing.

    It makes HUD the reference for whether an eval can be trusted, a position that brings both lab buyers and builders, earns press each quarter, and shows the QA product working at scale.

    Risks and how to handle them

    • Disputes with benchmark authors

      Share findings early, publish the method and correct errors quickly.

    • Seen as self-serving

      Open-source the method and the flagged task list.

    Evidence

    • Your websiteSeries A post (June 30, 2026): $16M led by Standard Capital with Y Combinator and others; 1.3M+ task runs, 2,500+ environments and 15M+ inference calls at the time; 50+ businesses on the platform; SheetBench-50 validated by PwC, Cisco, Charles Schwab and Fannie Mae; OSWorld-Verified fixed 300+ issues in the original benchmark; marketplace vendors earning millions a month; the goal is "to let anyone, anywhere build and sell frontier-grade RL environments and post-training data".Source
    • Your websiteResearch page: 7 posts, including RECON (October 21, 2026; 22 systems tested), "Evaluating Robot Models in the Cloud" (October 16), "Can you trust your browser eval?" (September 28), "What is a good task?" (September 22) and Assemble Bench for robot assembly (July 31), co-authored with Sixtyfour, Browserbase, Dreamscale and Sepal AI.Source
    • Your websiteHomepage (checked November 9, 2026): H1 "The platform for building RL environments"; counters of 4,831,034 task runs, 13,764 environments built and 22,297,355 inference calls; "Start building for free" and "Create your first environment in minutes. No credit card required."; a "Turn tasks into revenue" section for listing on DataVendor, "where AI labs buy training and evaluation data"; two customer stories (Sharpe and UiPath), both "Coming soon"; FAQ includes "Can I monetize environments I build?"; contact paths are "Book a call" and a founders email.Source
  • 20

    HUD's SheetBench-50 was validated by PwC, Cisco, Charles Schwab and Fannie Mae, and its homepage previews a UiPath story on benchmarking enterprise agents "against every frontier model". Enterprises buying or building agents need to know which model works on their own workflows, but HUD sells this through custom Enterprise deals with no packaged offer.

    High impactHigh effort First result in 1 quarter Business Development Lead

    First action

    Write the scope and price from the SheetBench-50 and UiPath projects.

    How to measure it

    Enterprise benchmark deals

    Now
    –
    Target
    6
    By
    2 quarters

    Steps

    1. 1

      Define scope

      Task count, models, deliverables, timeline.

      Forward Deployed Research Engineer 1 week

    2. 2

      Price it

      Fixed fee with platform credits included.

      Business Development Lead 3 days

    3. 3

      Build the template

      Report, dashboard and regression suite.

      Delivery Engineer 3 weeks

    4. 4

      Proof page

      SheetBench validation and the UiPath story.

      Technical Writer 2 weeks

    5. 5

      Pilot

      Two enterprises at an introductory price.

      Business Development Lead 6 weeks

    6. 6

      Convert

      From regression suite to platform subscription.

      Customer Operations Lead ongoing

    First result
    1 quarter
    Cost
    Forward-deployed engineering time per deal
    Owner
    Business Development Lead

    Smallest test

    Offer the package to three enterprises already in the pipeline.

    Why it works

    Package a fixed-scope offer: HUD and the enterprise turn 50-100 of its real workflows into a private taskset, run every leading model and agent framework on it, QA the results, and deliver a ranked report plus a reusable regression suite on HUD. Fixed price, 30 days, then a platform subscription.

    A fixed scope and price shortens enterprise buying, uses HUD's forward-deployed engineers, and leaves a taskset the customer keeps running, which creates recurring usage.

    Risks and how to handle them

    • Scope creep turns it into consulting

      Fix the task count and deliverables, and charge for anything extra.

    Builds on

    Evidence

    • Your websiteSeries A post (June 30, 2026): $16M led by Standard Capital with Y Combinator and others; 1.3M+ task runs, 2,500+ environments and 15M+ inference calls at the time; 50+ businesses on the platform; SheetBench-50 validated by PwC, Cisco, Charles Schwab and Fannie Mae; OSWorld-Verified fixed 300+ issues in the original benchmark; marketplace vendors earning millions a month; the goal is "to let anyone, anywhere build and sell frontier-grade RL environments and post-training data".Source
    • Your websiteHomepage (checked November 9, 2026): H1 "The platform for building RL environments"; counters of 4,831,034 task runs, 13,764 environments built and 22,297,355 inference calls; "Start building for free" and "Create your first environment in minutes. No credit card required."; a "Turn tasks into revenue" section for listing on DataVendor, "where AI labs buy training and evaluation data"; two customer stories (Sharpe and UiPath), both "Coming soon"; FAQ includes "Can I monetize environments I build?"; contact paths are "Book a call" and a founders email.Source
    • Your websiteBilling docs: Pay as you go ("No subscription"), Pro at $50 a month adding $50 of credits, and Enterprise at custom pricing; credits pay for environment runtime by the hour, model calls at the model's price plus a 5% platform fee (5% of the equivalent cost with your own provider key) and QA runs; no free tier is described. hud.ai/pricing opens the sign-in page.Source
  • 21

    With Toptal's ExpertCrowd, experts wrote the tasks and rubrics and HUD hosted and QA'd them. That one partnership brought in a whole pool of expert task-writers at once. HUD's goal is to let "anyone, anywhere" build and sell environments, and its outbound team works one vendor at a time.

    High impactHigh effort First result in 1 quarter GTM Lead

    First action

    List ten expert networks by domain and size and contact the top three.

    How to measure it

    Partner-sourced listings

    Now
    –
    Target
    100
    By
    3 quarters

    Steps

    1. 1

      Map networks

      By domain, size and lab demand.

      GTM Lead, Commercial Partnerships 2 weeks

    2. 2

      Write the deal

      Revenue share, QA standard, rights.

      Business Operations 2 weeks

    3. 3

      Co-branded onboarding

      Partner landing page and taskset templates.

      Product Manager, Marketplace 3 weeks

    4. 4

      Pilot

      One network, one domain, 50 tasks.

      Forward Deployed Research Engineer 6 weeks

    5. 5

      QA and list

      Through HUD's QA agents and human review.

      Lead Research Engineer, Data Quality ongoing

    6. 6

      Announce

      Joint release, as with ExpertCrowd-SWE.

      Chief Marketing Officer per partner

    First result
    1 quarter
    Cost
    Partnership and engineering time, plus a revenue share
    Owner
    GTM Lead, Commercial Partnerships

    Smallest test

    One pilot with one network in one domain.

    Why it works

    Sign expert and freelance networks in fields where labs want data (finance, law, medicine-adjacent operations, engineering) as supply partners. Their experts build tasks on HUD, HUD runs QA and lists the results, and revenue is shared. Give each partner a co-branded onboarding path and a dashboard.

    Supply arrives in batches instead of one listing at a time, and every partner's experts become HUD builders.

    Risks and how to handle them

    • Uneven task quality across networks

      A QA pass rate threshold before anything is listed.

    Builds on

    Evidence

    • PressToptal press release (November 6, 2026): Toptal's ExpertCrowd and HUD released ExpertCrowd-SWE, a software engineering benchmark; Toptal's experts wrote the tasks and grading rubrics, HUD hosted the environments and verified the benchmark against its QA pipelines; it is "the first release of an ongoing collaboration" with more initiatives in development.Source
    • Your websiteSeries A post (June 30, 2026): $16M led by Standard Capital with Y Combinator and others; 1.3M+ task runs, 2,500+ environments and 15M+ inference calls at the time; 50+ businesses on the platform; SheetBench-50 validated by PwC, Cisco, Charles Schwab and Fannie Mae; OSWorld-Verified fixed 300+ issues in the original benchmark; marketplace vendors earning millions a month; the goal is "to let anyone, anywhere build and sell frontier-grade RL environments and post-training data".Source
    • HiringOutbound Account Executive (remote, Pacific hours; also Singapore and SGT versions): persuade businesses and individuals to "list useful data on HUD", following buyer demand "usually from frontier AI labs", with daily cold calls, emails and LinkedIn messages and weekly listing targets; USD 60,000-90,000 on-target pay. Three Sales Development Representative roles sit alongside.Source

Wildcards 2

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

  • 22

    HUD is hiring Research Engineers for Robotics Data and Robotics Evals, and has published "Evaluating Robot Models in the Cloud" and Assemble Bench for contact-rich assembly. Robotics is one of six homepage categories, but there is no page for robotics teams, and the guides are about language agents.

    Medium impactHigh effort First result in 1 quarter Research Engineer

    First action

    Turn the two robotics research posts into a /robotics landing page.

    How to measure it

    Robotics teams running evals

    Now
    –
    Target
    10
    By
    2 quarters

    Steps

    1. 1

      Landing page

      From the two research posts.

      Technical Writer 1 week

    2. 2

      Onboarding path

      Bring a simulator or policy and run an eval.

      Research Engineer, Robotics Evals 4 weeks

    3. 3

      Seed data vendors

      Robotics datasets for DataVendor.

      Research Engineer, Robotics Data 6 weeks

    4. 4

      Workshop talk

      Assemble Bench results.

      Senior Research Manager per event

    5. 5

      Target list

      Robotics model teams for outbound.

      GTM Engineer 2 weeks

    6. 6

      Review

      Decide after two quarters whether to scale.

      Senior Research Manager 2 quarters

    First result
    1 quarter
    Cost
    Research time plus simulation compute
    Owner
    Research Engineer, Robotics Evals

    Smallest test

    Publish the page and count sign-ups for a month.

    Build the prospect list. The list compiled from public sources, verified, deduplicated and delivered to your CRM. Usually live within days.

    Book a working session

    Why it works

    Launch a /robotics page (cloud evaluation of robot models, Assemble Bench, how to bring a simulator or dataset) with a sign-up for robotics teams to evaluate a model in the cloud. Recruit robotics data vendors for DataVendor at the same time. Present Assemble Bench results at a robotics workshop.

    Robotics evaluation is less crowded than language-agent evals, and HUD already has the research and the hires to be credible early.

    Risks and how to handle them

    • Spreads a 25-person team too thin

      Cap it at the two robotics hires and decide after two quarters.

    Evidence

    • HiringCareers page: 38 open roles. 14 in Engineering & Research, including Research Engineer roles for Robotics Data, Robotics Evals, Benchmarks, and Privacy and Anonymization; 11 in GTM, including an Outbound Account Executive, a Sales Development Representative and a GTM Associate on-site in Singapore; 5 in Design & Product (Product Manager, Marketplace and Product Manager, Platform); 3 in Marketing; 4 in Operations.Source
    • Your websiteResearch page: 7 posts, including RECON (October 21, 2026; 22 systems tested), "Evaluating Robot Models in the Cloud" (October 16), "Can you trust your browser eval?" (September 28), "What is a good task?" (September 22) and Assemble Bench for robot assembly (July 31), co-authored with Sixtyfour, Browserbase, Dreamscale and Sepal AI.Source
    • Your websiteSitemap (37 URLs, checked November 9, 2026): homepage, careers, 2 blog posts, 7 research posts, 20 guides and legal pages; no pricing, customer story, comparison or environment pages.Source
  • 23

    HUD's seller guides point readers to the UK ICO's anonymisation guidance, and HUD is hiring a Research Engineer for Privacy and Anonymization. A company deciding whether to license its data or code first has to know whether it contains personal data or secrets. Today it has to work that out alone before anyone at HUD can help.

    Medium impactMedium effort First result in 1 quarter Research Engineer

    First action

    Ask the outbound team how often sellers raise privacy or confidentiality as an objection.

    How to measure it

    Listings completed after a privacy objection

    Now
    –
    Target
    +30%
    By
    2 quarters

    Steps

    1. 1

      Size the objection

      Tag privacy objections in the CRM.

      Business Development Lead 2 weeks

    2. 2

      Define checks

      Personal data, credentials, licences.

      Research Engineer, Privacy and Anonymization 2 weeks

    3. 3

      Build the scan

      Runs on the sample during listing creation.

      Product Engineer 4 weeks

    4. 4

      Write the explainer

      Plain language for legal teams.

      Technical Writer 1 week

    5. 5

      Badge listings

      Shown to buyers on DataVendor.

      Product Manager, Marketplace 1 week

    6. 6

      Measure

      Completion rates before and after.

      GTM Engineer monthly

    First result
    1 quarter
    Cost
    About six weeks of engineering
    Owner
    Research Engineer, Privacy and Anonymization

    Smallest test

    Run the check by hand for the next ten sellers who raise privacy.

    Why it works

    Offer an automated pre-check as part of listing: scan the sample for personal data and credentials, suggest what to strip or anonymise, and show a "privacy-checked" badge on listings that pass. Explain what was checked in plain language for the seller's legal team.

    It removes one of the main reasons sellers stall, and gives lab buyers a trust signal on every listing.

    Risks and how to handle them

    • A badge read as a legal guarantee

      Describe it as an automated check, not legal advice, and list exactly what it covers.

    Builds on

    Evidence

    • HiringCareers page: 38 open roles. 14 in Engineering & Research, including Research Engineer roles for Robotics Data, Robotics Evals, Benchmarks, and Privacy and Anonymization; 11 in GTM, including an Outbound Account Executive, a Sales Development Representative and a GTM Associate on-site in Singapore; 5 in Design & Product (Product Manager, Marketplace and Product Manager, Platform); 3 in Marketing; 4 in Operations.Source
    • Your website"How to Sell Data to Model Labs" (September 25, 2026): in-body links point to DataVendor's inventory, listing and buying guides, plus one external link to the UK ICO's anonymisation guidance; sellers are told to "Confirm seller payout timing separately from the buyer's checkout payment"; it closes with "Build environments that frontier labs trust" and "Start building for free".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.

Have it built

ProgrammableAgency

1 of the 23 ideas are things Programmable builds and runs for its customers: pages, outreach, prospect lists, booking flows and monitors. Pick one, or send the whole list.

Prospect lists1The list compiled from public sources, verified, deduplicated and delivered to your CRM.

A 30-minute session with the engineer who would build it, not a sales call. You leave with a plan whether or not you hire us.

Book a working session

Usually live within days.

Tools for the rest

ProgrammableTalk

Fill the events in this plan

These ideas are events. Talk gets registrants for them and tells you who's coming and what their company does before the day.

Plan an event (opens programmable.talk in a new tab)
ProgrammableDigital

Show it, don't just describe it

These ideas need something to show: a demo, a video, an animation, graphics or a deck. Digital turns screenshots or the site into all of them, in the company's own brand.

Make a demo (opens programmable.digital 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.

  • Flo Health growth strategyFlo is the world's most downloaded women's health app: period, ovulation and pregnancy tracking, Flo for Perimenopause, Secret Chats…22 growth ideas
  • Quintagroup growth strategyA Python and JavaScript software development company from Lviv, Ukraine, known for e-procurement and auction systems (the OpenProcurement…23 growth ideas
  • Cortland growth strategyA vertically integrated multifamily owner, developer and manager that runs hospitality-led apartment communities and offers third-party…21 growth ideas
  • Dorian Webb growth strategyAn Oakland, California fine jewelry designer making 18k gold, diamond, sterling and gemstone jewelry, engagement rings, custom pieces and…22 growth ideas
  • Ordinary Habit growth strategyA design-led puzzle brand selling 100, 500 and 1,000-piece jigsaw puzzles, matching cards, journals and home goods with original art by…21 growth ideas
  • ELEGOO growth strategyShenzhen-based maker of consumer 3D printers (Centauri, Neptune and OrangeStorm FDM; Saturn and Mars resin), filament, resin, accessories…23 growth ideas
  • Upwork growth strategyThe Upwork Marketplace connects businesses with independent talent across 10,000+ skills, with contracts, payment protection, Business…21 growth ideas
  • Assured growth strategyAI for property and casualty insurance claims: digital and voice first notice of loss, recorded statements, messaging, service assignment…21 growth ideas

See every playbook and website review

Independent research. Programmable wrote this playbook from public sources, as they stood on November 9, 2026, and cites them throughout. HUD did not commission, review or endorse it, and Programmable is not affiliated with HUD. 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 HUD runs its business.

Something wrong, or do you represent HUD? 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

    Fix each audience's call to action: seller CTAs on the seller guides, starter credits, a public pricing page, audience-specific call booking, the homepage description and speed, and the two customer stories.

  2. Days 31 to 6011 ideas

    Comparison pages, a free codebase grade, one CRM view of inbound sellers, the co-authored benchmark program, an environment directory, Singapore events, vendor referrals, Pro usage digests, releases, a founder content series and a sell step in the product.

  3. Days 61 to 905 ideas

    Commit to the big bets (a public environment quality index, a packaged enterprise benchmark, expert-network supply partnerships) and test two wildcards: robotics evals buyers and a privacy check for sellers.

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 hosted environments with traces they can inspect and debug. The most cited problem: benchmarks with broken or contaminated tasks.

  • QA agents that catch grader mistakes and reward hacking
  • Unreliable reward signals
  • In their words: “RL environments”

Market

Competitive landscape

4 competitors compared: Prime Intellect, Harbor Framework, Gymnasium and RLlib and Large data-labelling vendors.

  • Gap to exploit: A trusted place for companies to license code and data without selling the company
  • Prime Intellect: Built mainly around its own training and compute, not a data marketplace with QA
  • Harbor Framework: No hosting, marketplace or team review (HUD already imports Harbor tasks)

Hiring signals

HUD (about 25 people) has 38 open roles, and they describe a two-sided marketplace moving into scale.

  • Outbound Account Executive (Remote PT, Singapore, Remote SGT): Getting businesses to list data on HUD, with weekly listing targets. Ideas 1, 9, 14.
  • Chief Marketing Officer (CMO): Positioning for lab buyers and data vendors; case studies. Ideas 6, 8.
  • GTM Engineer: Sourcing supply at scale; one CRM, marketplace and product view. Ideas 9, 10.

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