AutoSitu growth playbook

Render each audience page's full copy on the server, as the product pages already are. Each page gets the buyer's problem, the modules that matter to them, the matching case-study result from the guides (60 staff hours for architects, $2.1M of scope found for GCs, entitlement 13 weeks sooner for developers), an FAQ, and a demo button that is a plain link. Do the same for the homepage. Then check each page with a no-JavaScript fetch and request reindexing in Search Console.

The pages already exist, sit in the sitemap and have the right titles. Making their content visible to crawlers and AI assistants is the cheapest discovery gain open to a two-person team. It also cuts JavaScript work on slower phones, where real visitors already wait 3.3 seconds for the main content.

[ ] 500+ words of body copy in the served HTML
[ ] Demo button is a plain link in the served HTML
[ ] One case-study result for this buyer, with its caveat
[ ] FAQ with structured data
[ ] Links to the matching product pages and guide
[ ] Title and meta description name the buyer
[ ] Checked with Search Console URL inspection after deploy

• Copy written for animated sections reads poorly as plain text

  Write the plain version first and animate on top of it.

• Five similar pages compete with each other

  Give each page a distinct buyer problem, result and FAQ.

• Your websiteAudience pages as served before JavaScript runs (checked November 10, 2026): /for-cities about 23 words, /for-architects about 20, /for-developers one heading and one sentence, /for-construction-gcs about 45, /for-planning-firms about 31. Product pages such as /compliance-review (about 700 words) and /completeness-review (about 750) and the four guides (1,500-1,600 words) are served in full.Source
• Your websiteHomepage (checked November 10, 2026): title "AutoSitu - AI Operating System for the Built World", headline "The first AI operating system for the built world." (in February 2026 it read "Check your plan in minutes"); the HTML served before JavaScript runs is about 200 words covering five modules (site feasibility, design review and QA/QC, code compliance, quantity takeoff, scope gap analysis) and an FAQ, with no demo link, customer names or numbers.Source
• Your websiteSitemap (27 URLs, checked November 10, 2026): homepage, about, six product pages, four guides (AI plan review, AI construction takeoff, real estate entitlement, data center feasibility), five audience pages, the blog with seven posts, privacy and terms. No pricing, customer story, careers or jurisdiction pages; every URL carries the same lastmod timestamp.Source
• PageSpeedGoogle PageSpeed (November 10, 2026): mobile performance 42, desktop 43; lab mobile LCP 5.4 s, total blocking time 11.1 s, 819 KB of JavaScript with about 238 KB unused; the LCP element is the "AutoSitu" wordmark text. Real Chrome visitors: LCP p75 3.3 s (above the 2.5 s target) and 26% slower than before, while INP (95 ms) and CLS (0.02) pass.Source

Pick one scheduler and replace every link with a shared component. Offer one event per buyer (city plan review, developer feasibility, architect QA/QC, GC takeoff and scope gaps, data center), each with two or three short intake questions: jurisdiction, project type and stage, and whether a drawing set is ready. Update the tagline to the current positioning and pass the source page in the booking URL.

Buyers book more readily when the event names their problem. The founders also get the context to prepare a demo on the buyer's own jurisdiction or drawings, which is AutoSitu's strongest proof.

• Intake questions lower bookings

  Keep three questions, with only the jurisdiction required.

• Meetings already booked on the old calendar

  Keep it live until those meetings have passed.

• Your websiteDemo links: the guides' buttons ("Book a plan review demo", "Book a takeoff demo", "Discuss a development site", "Discuss a data center project") go to cal.com/autositu, which offers two 30-minute events ("30 min meeting" and "30min AutoSitu") under the older tagline "AI native development plan review for faster approvals"; /scope-gap-analysis ("Schedule Private Demo", "Book Demo") goes to a Google Calendar appointment page instead.Source
• Your websiteHomepage (checked November 10, 2026): title "AutoSitu - AI Operating System for the Built World", headline "The first AI operating system for the built world." (in February 2026 it read "Check your plan in minutes"); the HTML served before JavaScript runs is about 200 words covering five modules (site feasibility, design review and QA/QC, code compliance, quantity takeoff, scope gap analysis) and an FAQ, with no demo link, customer names or numbers.Source
• Your websiteData center guide (about 1,500 words): power, utility agreements, phasing and permitting for data center campuses, citing JLL's 2026 outlook, Dominion's January 2026 process and Loudoun County's March 2025 rule changes; its button "Discuss a data center project" goes to cal.com/autositu.Source

Add a proof strip to the homepage, the about page and each audience page. It carries the three platform numbers (with definitions and an as-of date), the three case-study results from the guides with their caveats (60 staff hours saved, $2.1M of scope found before award, entitlement 13 weeks sooner), and the backers (Y Combinator, Pi Labs, Robinhood Ventures Fund II). Add customer names as clients agree to them.

Public agencies and AEC firms look for signs that others already rely on a tool, and the numbers already exist.

• Buyers question the numbers

  Footnote each definition and date.

• Single-project results read as typical

  Keep the guides' caveats next to each result.

• Your websiteScope gap analysis page: "Find the work nobody priced."; the only page with platform numbers: "65M+ elements and quantities analyzed", "$6.4B+ construction scope reviewed", "25,000+ developments analyzed"; no customer names or logos.Source
• Your websiteHomepage (checked November 10, 2026): title "AutoSitu - AI Operating System for the Built World", headline "The first AI operating system for the built world." (in February 2026 it read "Check your plan in minutes"); the HTML served before JavaScript runs is about 200 words covering five modules (site feasibility, design review and QA/QC, code compliance, quantity takeoff, scope gap analysis) and an FAQ, with no demo link, customer names or numbers.Source
• Your websiteAbout page: "Increasing the GDP of the world" and "AutoSitu is building the AI operating system for real estate development"; the served page has no team, location, customers or call to action.Source
• Competitor researchRivals' own homepages: Archistar AI PreCheck says it is trusted by 30-plus cities and local governments, names County of Los Angeles and City of Austin with staff quotes, and claims up to 55% fewer permit review cycles; CivCheck, now a Clariti company, is free to city partners "forever" with applicants paying "a small prescreening fee"; Togal sells "Takeoff in Minutes. Not Days." with a self-guided tour; UpCodes offers "Try it for Free".See the research Source

Show date, author and reading time on the index and on each post. Correct the tax-base post's date. Emit real lastmod dates in the sitemap. Add a short signup ("one finding a month from real plan reviews") to the blog and the guides, and send the first issue from the weekly finding posts.

Dates and authors are basic trust signals for public-sector readers, accurate lastmod helps crawlers, and a list turns one-time guide readers into a nurture channel the founders control.

• No time to write each issue

  Reuse the weekly finding posts.

• Your websiteBlog: seven posts; the index shows no dates, authors or newsletter signup. "The Tax Base You're Losing to the Next City Over" is dated January 7, 2025 but cites a September 2025 survey, says AutoSitu is piloting compliance review with municipalities and design consultancies in California and Michigan, and argues approvals are "a competitive asset, not an administrative afterthought".Source
• Your websiteSitemap (27 URLs, checked November 10, 2026): homepage, about, six product pages, four guides (AI plan review, AI construction takeoff, real estate entitlement, data center feasibility), five audience pages, the blog with seven posts, privacy and terms. No pricing, customer story, careers or jurisdiction pages; every URL carries the same lastmod timestamp.Source
• Y CombinatorY Combinator profile: Winter 2026, founded 2025, team size 2, "The AI operating system for the built world". The launch post claims teams "reduce city comments 50–70%", "cut resubmittals from 2–3 → 0–1" and "shorten approvals by 3–6 months+", with $100k-$400k of avoided carrying costs for developers and $10k-$25k of preserved fee margin per project for architects; it names no customers and links a launch video.Source

Server-render the hero and above-the-fold copy, load animations and heavy client components after first paint, remove the unused JavaScript and CSS, preload the font used by the wordmark, and add a JavaScript budget check to deploys. Re-run PageSpeed and watch the field LCP over the next 28 days.

Faster first paint keeps mobile visitors from link previews and search on the page, and much of the work overlaps with idea 1.

• Visual polish is lost

  Keep animations, just load them after the text.

• PageSpeedGoogle PageSpeed (November 10, 2026): mobile performance 42, desktop 43; lab mobile LCP 5.4 s, total blocking time 11.1 s, 819 KB of JavaScript with about 238 KB unused; the LCP element is the "AutoSitu" wordmark text. Real Chrome visitors: LCP p75 3.3 s (above the 2.5 s target) and 26% slower than before, while INP (95 ms) and CLS (0.02) pass.Source
• Your websiteHomepage (checked November 10, 2026): title "AutoSitu - AI Operating System for the Built World", headline "The first AI operating system for the built world." (in February 2026 it read "Check your plan in minutes"); the HTML served before JavaScript runs is about 200 words covering five modules (site feasibility, design review and QA/QC, code compliance, quantity takeoff, scope gap analysis) and an FAQ, with no demo link, customer names or numbers.Source

Embed the launch video (loaded only on click, to protect speed) on the homepage and /about. Each week, post one redacted finding on LinkedIn and the blog: the drawing excerpt, the governing requirement, why it would have caused a correction, and what it would have cost in time. Rotate through cities, architects, GCs and developers.

Showing the output is the most convincing proof a review product has, and a weekly cadence builds an audience at almost no cost.

• A finding identifies a client

  Use hypothetical or fully redacted sets.

• Y CombinatorY Combinator profile: Winter 2026, founded 2025, team size 2, "The AI operating system for the built world". The launch post claims teams "reduce city comments 50–70%", "cut resubmittals from 2–3 → 0–1" and "shorten approvals by 3–6 months+", with $100k-$400k of avoided carrying costs for developers and $10k-$25k of preserved fee margin per project for architects; it names no customers and links a launch video.Source
• Your websiteHomepage (checked November 10, 2026): title "AutoSitu - AI Operating System for the Built World", headline "The first AI operating system for the built world." (in February 2026 it read "Check your plan in minutes"); the HTML served before JavaScript runs is about 200 words covering five modules (site feasibility, design review and QA/QC, code compliance, quantity takeoff, scope gap analysis) and an FAQ, with no demo link, customer names or numbers.Source

List in the served HTML: how data is isolated per organization, hosting region, retention and deletion, subprocessors, which documents are available and how to request them, and a security contact. State only what is true today. Link the page from /for-cities, /pricing and the pilot one-pager.

Security review is a standard gate for public agencies. Clear answers shorten it and show a small vendor can meet it.

• Overstating controls

  List certifications only once they are held.

• Your websiteThe homepage FAQ says each organization gets a customized, isolated AI environment and links the AutoSitu Trust Center at trust.autositu.com, which shows only a "Trust center" heading to readers that do not run JavaScript.Source
• Your websiteAudience pages as served before JavaScript runs (checked November 10, 2026): /for-cities about 23 words, /for-architects about 20, /for-developers one heading and one sentence, /for-construction-gcs about 45, /for-planning-firms about 31. Product pages such as /compliance-review (about 700 words) and /completeness-review (about 750) and the four guides (1,500-1,600 words) are served in full.Source

Create a /customers section with three anonymized stories: problem, project type, what was flagged (with redacted sheet excerpts), outcome and caveat. Add a city-side story from the California or Michigan pilots when a municipality agrees. Link each story from its audience page, guide and sales emails, and ask each client whether a name or logo can follow later.

A single-project result is more believable as a full story with real findings than as one sentence in a guide, and each buyer gets proof from someone like them.

• Clients decline to be named

  Describe them by project type and region.

• Overclaiming

  Keep the reported-by-client labels the guides already use.

• #3 Put the 25,000 developments and $6.4B of scope on the homepage, not only on the scope gap page

• Your websiteAI plan review guide: "catch the issues that send plans back"; compares AutoSitu with Avolve ProjectDox, Bluebeam and UpCodes Copilot ("not an independent product benchmark"); one anonymized mixed-use project reported "60 fewer staff hours spent on internal QA/QC and resubmission handling", details withheld "at the client's request"; FAQ "How much does AI plan review software cost?".Source
• Your websiteAI construction takeoff guide: compares Togal, STACK and Autodesk Forma Takeoff (vendor descriptions reviewed September 2026); one anonymized commercial project found about $2.1M of unassigned scope before award ("priced exposure, not a recorded saving"); the cost FAQ gives no prices and tells buyers to ask how reference quantities were established.Source
• Your websiteEntitlement and feasibility guide (about 1,600 words): compares TestFit and ARGUS EstateMaster; one multifamily developer reports entitlement 13 weeks ahead of its original schedule and about $350K of carrying-cost exposure avoided (a developer estimate).Source
• Y CombinatorY Combinator profile: Winter 2026, founded 2025, team size 2, "The AI operating system for the built world". The launch post claims teams "reduce city comments 50–70%", "cut resubmittals from 2–3 → 0–1" and "shorten approvals by 3–6 months+", with $100k-$400k of avoided carrying costs for developers and $10k-$25k of preserved fee margin per project for architects; it names no customers and links a launch video.Source
• Your websiteSitemap (27 URLs, checked November 10, 2026): homepage, about, six product pages, four guides (AI plan review, AI construction takeoff, real estate entitlement, data center feasibility), five audience pages, the blog with seven posts, privacy and terms. No pricing, customer story, careers or jurisdiction pages; every URL carries the same lastmod timestamp.Source

Put an address field on /feasibility-analysis and /for-developers. For jurisdictions where the code library is complete, return a short sourced snapshot: zoning district, key envelope limits with code citations, by-right or discretionary path, and two nearby precedents. Email the fuller sample report and offer the complete study on a call. Addresses in unsupported jurisdictions join a waitlist, which also shows where demand is.

It shows the product's sourcing on the buyer's own site and captures site-level intent at the moment a developer is weighing a parcel.

• A wrong answer on a public tool

  Supported jurisdictions only, every rule cited, and the page's decision-support label.

• Compute cost from casual use

  Email verification and a daily limit.

• Your websiteFeasibility page: "Know what a site can really become."; an address or parcel number is enough for a first pass; one sourced report covering zoning envelope, precedents, massing, pro forma and entitlement pathway "in days instead of months"; no sample report; "a decision-support tool, not a substitute for entitlement counsel or a licensed engineer of record"; the same site model carries into design review, compliance and takeoff.Source
• Competitor researchRivals' own homepages: Archistar AI PreCheck says it is trusted by 30-plus cities and local governments, names County of Los Angeles and City of Austin with staff quotes, and claims up to 55% fewer permit review cycles; CivCheck, now a Clariti company, is free to city partners "forever" with applicants paying "a small prescreening fee"; Togal sells "Takeoff in Minutes. Not Days." with a self-guided tour; UpCodes offers "Try it for Free".See the research Source
• Your websiteHomepage (checked November 10, 2026): title "AutoSitu - AI Operating System for the Built World", headline "The first AI operating system for the built world." (in February 2026 it read "Check your plan in minutes"); the HTML served before JavaScript runs is about 200 words covering five modules (site feasibility, design review and QA/QC, code compliance, quantity takeoff, scope gap analysis) and an FAQ, with no demo link, customer names or numbers.Source

Publish /pricing with structures and starting ranges: a fixed-fee pilot credited against the first year, per-project pricing for developers and GCs, per-set pricing for architects, and an annual city licence scaled by permit volume. Add what a pilot includes, how long it takes, security documents and how public agencies can buy.

Price ranges qualify leads before the call and shorten city buying cycles, where the budget question decides everything.

• Ranges anchor too low

  Show starting prices and credit the pilot fee against year one.

• A rival undercuts the published price

  Lead with the outcome per project, such as hours and weeks saved.

• Your websiteAI plan review guide: "catch the issues that send plans back"; compares AutoSitu with Avolve ProjectDox, Bluebeam and UpCodes Copilot ("not an independent product benchmark"); one anonymized mixed-use project reported "60 fewer staff hours spent on internal QA/QC and resubmission handling", details withheld "at the client's request"; FAQ "How much does AI plan review software cost?".Source
• Your websiteAI construction takeoff guide: compares Togal, STACK and Autodesk Forma Takeoff (vendor descriptions reviewed September 2026); one anonymized commercial project found about $2.1M of unassigned scope before award ("priced exposure, not a recorded saving"); the cost FAQ gives no prices and tells buyers to ask how reference quantities were established.Source
• Your websiteSitemap (27 URLs, checked November 10, 2026): homepage, about, six product pages, four guides (AI plan review, AI construction takeoff, real estate entitlement, data center feasibility), five audience pages, the blog with seven posts, privacy and terms. No pricing, customer story, careers or jurisdiction pages; every URL carries the same lastmod timestamp.Source
• Competitor researchRivals' own homepages: Archistar AI PreCheck says it is trusted by 30-plus cities and local governments, names County of Los Angeles and City of Austin with staff quotes, and claims up to 55% fewer permit review cycles; CivCheck, now a Clariti company, is free to city partners "forever" with applicants paying "a small prescreening fee"; Togal sells "Takeoff in Minutes. Not Days." with a self-guided tour; UpCodes offers "Try it for Free".See the research Source

Standardize the pilot: one permit type, first-pass review of a set number of submittals, a baseline and pilot comparison (staff hours per set, correction cycles, days to first comments), human authority preserved, a fixed fee and an end-of-pilot report the city can present. Publish the outline on /for-cities with a downloadable one-pager, and use the same template for every city conversation.

Cities buy what they can explain to council. A fixed scope and an agreed measure turn "try our AI" into a low-risk, reportable decision.

Scope: first-pass review of [N] [permit type] submittals over 90 days
Who decides: city reviewers keep every determination; AutoSitu findings cite code section and sheet
Baseline (2 weeks before): staff hours per set, correction cycles, days to first comments
Pilot measures: the same three, plus findings accepted or rejected by reviewers
Data: isolated environment; city data not used for other customers
Fee: fixed [$], credited to year one if the city continues
End report: one page of results for council, plus a findings appendix

• Pilots never convert to paid contracts

  Agree the post-pilot price and buying path at the start.

• Baseline data is weak

  Measure two weeks of manual review before the pilot.

• #1 Make the five audience pages readable before JavaScript: today /for-cities serves 23 words and /for-architects 20

• Your websiteBlog: seven posts; the index shows no dates, authors or newsletter signup. "The Tax Base You're Losing to the Next City Over" is dated January 7, 2025 but cites a September 2025 survey, says AutoSitu is piloting compliance review with municipalities and design consultancies in California and Michigan, and argues approvals are "a competitive asset, not an administrative afterthought".Source
• Your websiteAudience pages as served before JavaScript runs (checked November 10, 2026): /for-cities about 23 words, /for-architects about 20, /for-developers one heading and one sentence, /for-construction-gcs about 45, /for-planning-firms about 31. Product pages such as /compliance-review (about 700 words) and /completeness-review (about 750) and the four guides (1,500-1,600 words) are served in full.Source
• Competitor researchRivals' own homepages: Archistar AI PreCheck says it is trusted by 30-plus cities and local governments, names County of Los Angeles and City of Austin with staff quotes, and claims up to 55% fewer permit review cycles; CivCheck, now a Clariti company, is free to city partners "forever" with applicants paying "a small prescreening fee"; Togal sells "Takeoff in Minutes. Not Days." with a self-guided tour; UpCodes offers "Try it for Free".See the research Source

Publish one page per jurisdiction where coverage is strong, starting with ten. Each page covers adopted codes and local amendments, a summary of the intake checklist, common correction themes, the typical timeline from published sources, and links to the parcel first pass, the completeness check and a demo. Show a reviewed-on date and source links on every page.

City-specific searches carry high intent, and few vendors answer them with sourced, current detail. The pages also give AI assistants something specific to cite.

• Out-of-date code references

  A reviewed-on date and a source link for every rule.

• Templated thin pages

  Publish only where there is original detail.

• Your websiteCarrying-cost post (February 3, 2026): on a $10 million project at 7.5-9.5% interest, about $62,500-$79,000 a month; median permitting times of 7.5 months in Boston, 13 in Los Angeles, 16 in Seattle, 30 in New York City and 33 in San Francisco; no calculator.Source
• Your websiteCompleteness review page: "Nothing missing before the clock starts."; checks submittals against each jurisdiction's published checklists for missing sheets, unsigned forms and absent studies, and compares resubmittals with the prior run; built for cities and applicants working from one shared checklist; no demo button in the served page.Source
• Your websiteSitemap (27 URLs, checked November 10, 2026): homepage, about, six product pages, four guides (AI plan review, AI construction takeoff, real estate entitlement, data center feasibility), five audience pages, the blog with seven posts, privacy and terms. No pricing, customer story, careers or jurisdiction pages; every URL carries the same lastmod timestamp.Source

Ask the Michigan investors for introductions to municipal leaders, economic development offices and active developers in the state. Target the cities around the pilot city as a cluster, publish the Michigan pilot result (with permission) on a Michigan page, and present it at state planning and municipal events. Repeat the cluster approach in California once Michigan has two or three cities.

Local investors open doors faster than cold outreach, and a cluster of nearby cities gives developers a reason to ask their other cities for the same review.

• The pilot city declines publicity

  Use an anonymized result and private references.

• Founder time split across two states

  Lead with Michigan until two cities sign.

• #8 A 90-day city pilot package with a written success measure that plan review managers can take to council

• Your websiteAugust 6, 2026 post: AutoSitu joined the initial Robinhood Ventures Fund II portfolio, adding to "a seed round led by Pi Labs" with Y Combinator, Detroit Venture Partners and Michigan Rise.Source
• Your websiteBlog: seven posts; the index shows no dates, authors or newsletter signup. "The Tax Base You're Losing to the Next City Over" is dated January 7, 2025 but cites a September 2025 survey, says AutoSitu is piloting compliance review with municipalities and design consultancies in California and Michigan, and argues approvals are "a competitive asset, not an administrative afterthought".Source

Build a small calculator: project cost, loan rate, taxes and insurance, and the expected months in review (prefilled with the city median where known). It shows the monthly carrying cost and the cost of one extra correction cycle, then offers a feasibility or pre-submission review. Email the result as a one-page PDF.

It turns a general argument into the buyer's own number, the figure a developer takes to a partner or lender meeting.

• The numbers look like a promise

  Label every result as an estimate based on the user's inputs.

• Your websiteCarrying-cost post (February 3, 2026): on a $10 million project at 7.5-9.5% interest, about $62,500-$79,000 a month; median permitting times of 7.5 months in Boston, 13 in Los Angeles, 16 in Seattle, 30 in New York City and 33 in San Francisco; no calculator.Source
• Y CombinatorY Combinator profile: Winter 2026, founded 2025, team size 2, "The AI operating system for the built world". The launch post claims teams "reduce city comments 50–70%", "cut resubmittals from 2–3 → 0–1" and "shorten approvals by 3–6 months+", with $100k-$400k of avoided carrying costs for developers and $10k-$25k of preserved fee margin per project for architects; it names no customers and links a launch video.Source
• Your websiteEntitlement and feasibility guide (about 1,600 words): compares TestFit and ARGUS EstateMaster; one multifamily developer reports entitlement 13 weeks ahead of its original schedule and about $350K of carrying-cost exposure avoided (a developer estimate).Source

Write a fair guide for permitting directors and building officials: the models on offer (applicant-paid prescreen, city licence, add-ons to electronic plan review), who makes determinations, how findings are sourced, how data is handled, what a pilot costs and how results are measured. Use the same transparency note as the existing guides, date it, and have the claims about rivals checked against their current pages.

Comparison searches are high intent, and AutoSitu already writes careful comparison guides; this one targets the buyer it most wants to win.

• Claims about rivals go out of date

  Date the table and link each source.

• Your websiteAI plan review guide: "catch the issues that send plans back"; compares AutoSitu with Avolve ProjectDox, Bluebeam and UpCodes Copilot ("not an independent product benchmark"); one anonymized mixed-use project reported "60 fewer staff hours spent on internal QA/QC and resubmission handling", details withheld "at the client's request"; FAQ "How much does AI plan review software cost?".Source
• Competitor researchRivals' own homepages: Archistar AI PreCheck says it is trusted by 30-plus cities and local governments, names County of Los Angeles and City of Austin with staff quotes, and claims up to 55% fewer permit review cycles; CivCheck, now a Clariti company, is free to city partners "forever" with applicants paying "a small prescreening fee"; Togal sells "Takeoff in Minutes. Not Days." with a self-guided tour; UpCodes offers "Try it for Free".See the research Source

Package a "data center site screen": parcel, target IT megawatts, operating date and delivery model in; an approval matrix, local rule changes, power milestones to confirm, and phase risks out, delivered within days at a fixed price. Offer the guide's approval matrix as a downloadable template in exchange for an email, and run outbound to data center developers and site selection consultants.

A defined deliverable at a known price is easier to buy than a platform demo, and the guide already proves the domain knowledge.

• Utility facts can't be verified from public data

  List them as items to confirm with the utility, as the guide does.

• Distraction from the core city and developer work

  Cap it at five screens until one converts.

• #2 One booking link everywhere, with an event for each buyer, instead of two calendars and an old tagline

• Your websiteData center guide (about 1,500 words): power, utility agreements, phasing and permitting for data center campuses, citing JLL's 2026 outlook, Dominion's January 2026 process and Loudoun County's March 2025 rule changes; its button "Discuss a data center project" goes to cal.com/autositu.Source
• Your websiteFeasibility page: "Know what a site can really become."; an address or parcel number is enough for a first pass; one sourced report covering zoning envelope, precedents, massing, pro forma and entitlement pathway "in days instead of months"; no sample report; "a decision-support tool, not a substitute for entitlement counsel or a licensed engineer of record"; the same site model carries into design review, compliance and takeoff.Source
• Your websiteDemo links: the guides' buttons ("Book a plan review demo", "Book a takeoff demo", "Discuss a development site", "Discuss a data center project") go to cal.com/autositu, which offers two 30-minute events ("30 min meeting" and "30min AutoSitu") under the older tagline "AI native development plan review for faster approvals"; /scope-gap-analysis ("Schedule Private Demo", "Book Demo") goes to a Google Calendar appointment page instead.Source

Offer a "pre-submission review": upload one drawing set, get cited findings (code, firm standards, coordination, completeness) within 48 hours for a fixed price, with a discount on the first set for new firms. Put it on /for-architects with a sample findings page, and offer it first to firms that submit in pilot jurisdictions. Convert repeat buyers to an annual plan.

A fixed price per set is a small, budgeted decision for a project architect, and one avoided correction cycle more than pays for it.

• 48 hours is too tight at volume

  Cap weekly slots and set a waitlist.

• Liability worries

  Keep the decision-support language: the architect of record decides.

• #1 Make the five audience pages readable before JavaScript: today /for-cities serves 23 words and /for-architects 20

• Y CombinatorY Combinator profile: Winter 2026, founded 2025, team size 2, "The AI operating system for the built world". The launch post claims teams "reduce city comments 50–70%", "cut resubmittals from 2–3 → 0–1" and "shorten approvals by 3–6 months+", with $100k-$400k of avoided carrying costs for developers and $10k-$25k of preserved fee margin per project for architects; it names no customers and links a launch video.Source
• Your websiteAI plan review guide: "catch the issues that send plans back"; compares AutoSitu with Avolve ProjectDox, Bluebeam and UpCodes Copilot ("not an independent product benchmark"); one anonymized mixed-use project reported "60 fewer staff hours spent on internal QA/QC and resubmission handling", details withheld "at the client's request"; FAQ "How much does AI plan review software cost?".Source
• Your websiteAudience pages as served before JavaScript runs (checked November 10, 2026): /for-cities about 23 words, /for-architects about 20, /for-developers one heading and one sentence, /for-construction-gcs about 45, /for-planning-firms about 31. Product pages such as /compliance-review (about 700 words) and /completeness-review (about 750) and the four guides (1,500-1,600 words) are served in full.Source

When a feasibility study closes, schedule a check-in for schematic design. Offer the first compliance and completeness run on the same project at a bundled price, and show what carried over (envelope, conditions, precedents). Track how many projects use two or more modules, and price annual developer plans by project rather than by module.

The second module costs far less to sell than the first, and multi-module projects show the connected-platform claim in practice.

• Projects stall between stages

  Keep the offer open for 12 months.

• Your websiteFeasibility page: "Know what a site can really become."; an address or parcel number is enough for a first pass; one sourced report covering zoning envelope, precedents, massing, pro forma and entitlement pathway "in days instead of months"; no sample report; "a decision-support tool, not a substitute for entitlement counsel or a licensed engineer of record"; the same site model carries into design review, compliance and takeoff.Source
• HiringFounding Engineer (San Francisco or remote US, $120K-$250K, 1-3% equity, on the Y Combinator jobs board): builds autonomous development-review agents, BIM-native systems and a "persistent project intelligence layer" from feasibility through construction, plus evaluation tooling for high-stakes outputs, working directly with architects, engineers, developers and public-sector reviewers; AutoSitu supports public agencies and private teams on projects worth "hundreds of millions of dollars in development value".Source
• Your websiteHomepage (checked November 10, 2026): title "AutoSitu - AI Operating System for the Built World", headline "The first AI operating system for the built world." (in February 2026 it read "Check your plan in minutes"); the HTML served before JavaScript runs is about 200 words covering five modules (site feasibility, design review and QA/QC, code compliance, quantity takeoff, scope gap analysis) and an FAQ, with no demo link, customer names or numbers.Source

Launch a small partner program: a firm licence at a partner price, completeness reports the firm can deliver to its own clients under shared branding, and a referral fee for developers it brings to feasibility or compliance. Start with five firms in the pilot states, and put the program on /for-planning-firms once that page is server-rendered.

Partners already hold the client relationship and the submission calendar. AutoSitu gains volume without adding sales staff.

• Partners see AutoSitu as a competitor

  Position it as their tool, with their branding.

• #1 Make the five audience pages readable before JavaScript: today /for-cities serves 23 words and /for-architects 20

• Your websiteAudience pages as served before JavaScript runs (checked November 10, 2026): /for-cities about 23 words, /for-architects about 20, /for-developers one heading and one sentence, /for-construction-gcs about 45, /for-planning-firms about 31. Product pages such as /compliance-review (about 700 words) and /completeness-review (about 750) and the four guides (1,500-1,600 words) are served in full.Source
• Your websiteCompleteness review page: "Nothing missing before the clock starts."; checks submittals against each jurisdiction's published checklists for missing sheets, unsigned forms and absent studies, and compares resubmittals with the prior run; built for cities and applicants working from one shared checklist; no demo button in the served page.Source
• Your websiteBlog: seven posts; the index shows no dates, authors or newsletter signup. "The Tax Base You're Losing to the Next City Over" is dated January 7, 2025 but cites a September 2025 survey, says AutoSitu is piloting compliance review with municipalities and design consultancies in California and Michigan, and argues approvals are "a competitive asset, not an administrative afterthought".Source

Publish a /how-we-measure page: test sets per module, how reference answers were set (licensed reviewers, approved plan sets), finding precision and recall, the share of items marked "needs review", how human corrections are counted, and the date of the last run. Update it every quarter from the evaluation tooling, and link it from /pricing, /for-cities and every guide.

City procurement and senior QA/QC leads must justify a tool to others. A published method gives them that justification and holds rivals to the same standard.

• The numbers look worse than rivals' marketing claims

  Explain the method; rigorous measures read as trustworthy to expert buyers.

• Your websiteAI construction takeoff guide: compares Togal, STACK and Autodesk Forma Takeoff (vendor descriptions reviewed September 2026); one anonymized commercial project found about $2.1M of unassigned scope before award ("priced exposure, not a recorded saving"); the cost FAQ gives no prices and tells buyers to ask how reference quantities were established.Source
• HiringFounding Engineer (San Francisco or remote US, $120K-$250K, 1-3% equity, on the Y Combinator jobs board): builds autonomous development-review agents, BIM-native systems and a "persistent project intelligence layer" from feasibility through construction, plus evaluation tooling for high-stakes outputs, working directly with architects, engineers, developers and public-sector reviewers; AutoSitu supports public agencies and private teams on projects worth "hundreds of millions of dollars in development value".Source
• Competitor researchRivals' own homepages: Archistar AI PreCheck says it is trusted by 30-plus cities and local governments, names County of Los Angeles and City of Austin with staff quotes, and claims up to 55% fewer permit review cycles; CivCheck, now a Clariti company, is free to city partners "forever" with applicants paying "a small prescreening fee"; Togal sells "Takeoff in Minutes. Not Days." with a self-guided tour; UpCodes offers "Try it for Free".See the research Source

In one launch jurisdiction where pilots already run (California or Michigan), let an applicant upload a drawing set and get an item-by-item completeness report for a fixed per-set fee, with the first run free. Show the city aggregate results: the share of incoming sets pre-checked and the most common missing items. Use that applicant volume to open the city conversation, then add a second jurisdiction.

Applicants feel a rejected intake directly and can pay at once. Their usage builds the evidence a city needs before it buys.

• An applicant blames AutoSitu for a rejection

  Keep the page's statement that a clean run does not guarantee acceptance.

• The city sees it as interference

  Brief the city early and share the aggregate results.

• Your websiteCompleteness review page: "Nothing missing before the clock starts."; checks submittals against each jurisdiction's published checklists for missing sheets, unsigned forms and absent studies, and compares resubmittals with the prior run; built for cities and applicants working from one shared checklist; no demo button in the served page.Source
• Competitor researchRivals' own homepages: Archistar AI PreCheck says it is trusted by 30-plus cities and local governments, names County of Los Angeles and City of Austin with staff quotes, and claims up to 55% fewer permit review cycles; CivCheck, now a Clariti company, is free to city partners "forever" with applicants paying "a small prescreening fee"; Togal sells "Takeoff in Minutes. Not Days." with a self-guided tour; UpCodes offers "Try it for Free".See the research Source
• Y CombinatorY Combinator profile: Winter 2026, founded 2025, team size 2, "The AI operating system for the built world". The launch post claims teams "reduce city comments 50–70%", "cut resubmittals from 2–3 → 0–1" and "shorten approvals by 3–6 months+", with $100k-$400k of avoided carrying costs for developers and $10k-$25k of preserved fee margin per project for architects; it names no customers and links a launch video.Source
• Your websiteBlog: seven posts; the index shows no dates, authors or newsletter signup. "The Tax Base You're Losing to the Next City Over" is dated January 7, 2025 but cites a September 2025 survey, says AutoSitu is piloting compliance review with municipalities and design consultancies in California and Michigan, and argues approvals are "a competitive asset, not an administrative afterthought".Source

Aggregate anonymized findings into an annual report: the most common correction categories by discipline and project type, the items most often missing at intake, typical resubmittal counts, and which issues cost the most time. Cut it by region only where sample sizes and agreements allow. Launch it with trade press, a webinar for building officials and a version for architects, and refresh it each year.

Original data earns links, press and citations from AI assistants, and it positions AutoSitu as the authority on plan quality, which is the thing it sells.

• Client data concerns

  Aggregate only, with minimum sample sizes and no project details.

• The dataset skews to a few jurisdictions

  State the sample and its limits.

• #10 Fix the blog basics: dates and authors on the index, the 2025 date error, and a newsletter signup

• Your websiteScope gap analysis page: "Find the work nobody priced."; the only page with platform numbers: "65M+ elements and quantities analyzed", "$6.4B+ construction scope reviewed", "25,000+ developments analyzed"; no customer names or logos.Source
• Your websiteBlog: seven posts; the index shows no dates, authors or newsletter signup. "The Tax Base You're Losing to the Next City Over" is dated January 7, 2025 but cites a September 2025 survey, says AutoSitu is piloting compliance review with municipalities and design consultancies in California and Michigan, and argues approvals are "a competitive asset, not an administrative afterthought".Source
• Your websiteCarrying-cost post (February 3, 2026): on a $10 million project at 7.5-9.5% interest, about $62,500-$79,000 a month; median permitting times of 7.5 months in Boston, 13 in Los Angeles, 16 in Seattle, 30 in New York City and 33 in San Francisco; no calculator.Source

Using permit records that cities publish on open-data portals, compute time to first comments, correction cycles and total time for common permit types. Frame the index around predictability rather than speed alone, publish the method, and pair each city's figures with an improvement playbook. Offer every listed city a private briefing before publication, and let developers look up their target cities.

It turns the company's thesis into something a city leader can act on, while developers use the index to plan sites and return to AutoSitu for feasibility.

• Cities feel shamed

  Brief them first, frame results around improvement, and highlight the cities that gained.

• Inconsistent open data

  Include only cities whose data meets a published quality bar.

• #9 Jurisdiction pages for the cities in the code library: what gets plans sent back there, and how long approval takes

• Your websiteBlog: seven posts; the index shows no dates, authors or newsletter signup. "The Tax Base You're Losing to the Next City Over" is dated January 7, 2025 but cites a September 2025 survey, says AutoSitu is piloting compliance review with municipalities and design consultancies in California and Michigan, and argues approvals are "a competitive asset, not an administrative afterthought".Source
• Your websiteCarrying-cost post (February 3, 2026): on a $10 million project at 7.5-9.5% interest, about $62,500-$79,000 a month; median permitting times of 7.5 months in Boston, 13 in Los Angeles, 16 in Seattle, 30 in New York City and 33 in San Francisco; no calculator.Source

Offer applicants an optional one-page cover sheet that summarizes the completeness run: checklist version and date, items present, open items with reasons, and a link the reviewer can use to see the item-by-item report. Pilot it only in jurisdictions that agree. Track how many reviewers open the link, and follow up with cities where many pre-checked sets arrive.

Every applicant who uses AutoSitu puts it in front of the person who decides whether the city should buy it, with that city's own checklist as proof.

• Cities see it as advertising

  Use it only where a city agrees, and keep it plain and factual.

• Reviewers read it as an approval

  State clearly that it is not a determination of compliance.

• #6 A self-serve completeness pre-check for architects and permit applicants in one launch jurisdiction

• Your websiteCompleteness review page: "Nothing missing before the clock starts."; checks submittals against each jurisdiction's published checklists for missing sheets, unsigned forms and absent studies, and compares resubmittals with the prior run; built for cities and applicants working from one shared checklist; no demo button in the served page.Source
• Y CombinatorY Combinator profile: Winter 2026, founded 2025, team size 2, "The AI operating system for the built world". The launch post claims teams "reduce city comments 50–70%", "cut resubmittals from 2–3 → 0–1" and "shorten approvals by 3–6 months+", with $100k-$400k of avoided carrying costs for developers and $10k-$25k of preserved fee margin per project for architects; it names no customers and links a launch video.Source