ProgrammableMarketing/tryscott.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 · 6
  • This quarter · 12
  • Big bets · 3
  • Wildcards · 2
Generated Nov 10, 2026
  1. Programmable
  2. Playbooks
  3. Scott AI
Scott AI logo

growth playbook · Nov 10, 2026

tryscott.ai (opens in a new tab)

Scott AI growth playbook

Diagnosis

Scott AI is a three-person Y Combinator Fall 2025 company with a clear thesis ("Coding agents amplify the cost of bad planning"), a working product documented in depth (sync, planspaces, reviews, PR design review) and two engineering roles open.

The problem is the front door: tryscott.ai now shows one line, "Scott: we help shops with estimations", tells search engines not to index it, and has no link to the docs, the install command or a demo, while the YC launch still sends people to an /apply page that returns 404.

This playbook first restores a real homepage, the launch link, crawlability and the install path, and explains exactly what sync uploads. It then adds pricing, comparison pages, activation and sharing loops built into planspaces and PR comments, and longer bets on a public plan-drift report, an enterprise tier and a free "where the agents disagree" tool.

Growth Score

5/10

Biggest Opportunity

Every planspace link and every PR comment is seen by teammates who have not installed Scott; making those surfaces invite them in turns each team into the next sign-up.

Biggest Risk

Agent workspaces, spec-driven IDEs and AI code review tools (Traycer, Kiro, CodeRabbit) are closing in from both sides, and a homepage that describes a different business makes Scott invisible while they grow.

01

Start here

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

  1. 1

    Put the coding-agent planning product back on tryscott.ai: the homepage now says "we help shops with estimations" and blocks search engines

    Remove the noindex today and ship a holding page with the YC one-liner, the install command and a docs link.

    High impact First result in 1 week
  2. 2

    Make the install command the main button, with a "first answer in five minutes" promise

    Time a clean install-to-first-answer run and write the promise from the real number.

    High impact First result in 2 weeks
  3. 3

    Publish exactly what "scott sync" uploads, with an exclude list, before teams ask

    Write down today's upload fields and retention.

    High impact First result in 2 weeks

02

Your idea portfolio

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

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

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

Effort vs impactEach number is an idea. Select one to open it.
Impact →
Quick wins · 3Big bets · 6Fill-ins · 9Reconsider · 5
Effort →

Showing 23 of 23 ideas

This week 6

Quick wins with a clear first action.

  • 01

    Every public path to Scott ends at tryscott.ai: the YC company page, the YC launch (122 upvotes) and both open job posts. Today the homepage shows one heading, "Scott: we help shops with estimations", and an email address, with no link to the docs, the install command, the app or a demo, and PageSpeed reports that the page is blocked from indexing. A developer who heard about "the planning layer for teams that build with coding agents" lands on a page about a different business with no next step, and a search for Scott cannot find the homepage at all. The product itself is live and well documented at docs.tryscott.ai.

    High impactLow effort First result in 1 week Founder/CEO

    First action

    Remove the noindex today and ship a holding page with the YC one-liner, the install command and a docs link.

    How to measure it

    Homepage to install-or-docs click rate

    Now
    –
    Target
    25% of visitors
    By
    30 days

    Steps

    1. 1

      Decide the message

      Confirm the homepage is for the planning layer; if shop estimations is a separate bet, give it its own URL.

      Founder/CEO 1 hour

    2. 2

      Remove the noindex

      Drop the robots noindex and set the page title and meta description.

      Founder/CTO 1 hour

    3. 3

      Write the page

      Headline, who it is for, and three blocks: Sync, Planspaces, PR design review.

      Founder/CEO 1 day

    4. 4

      Add the install block

      Copy button for the install command and "scott sync", with a link to the quickstart.

      Product Engineer 2 hours

    5. 5

      Add demo and trust links

      A team demo button, the trust center and the SOC 2 line from the docs.

      Founder/CEO 1 hour

    6. 6

      Add share image and favicon

      So links in chats, YC and social show the product, not a blank card.

      Design Engineer 2 hours

    7. 7

      Track clicks

      Count install-command copies, docs clicks and demo bookings.

      Founder/CTO 1 day

    8. 8

      Request indexing

      Submit the homepage in Search Console and check the result shows the new description.

      Founder/CTO 1 week

    First result
    1 week
    Cost
    $0, about two days of founder and engineer time
    Owner
    Founder/CEO

    Smallest test

    Swap the H1 for the YC one-liner, add the install command and a docs link, and remove the noindex today.

    Why it works

    Replace the stub with a one-screen product homepage that matches the YC description and the docs: the headline "The planning layer for teams that build with coding agents", three short blocks for Sync, Planspaces and PR design review, the one-line install command with a copy button, a docs link and a "Book a team demo" button. Remove the noindex, write a title and meta description, and add a share image and favicon. If shop estimations is a real new direction, give it its own page or domain (idea 21) instead of replacing the page that YC, the docs and the job posts all point to.

    The homepage is the one URL every channel shares. A clear headline and a single next step turn visits that the launch, the YC listing and the hiring already create into installs, and let search engines and AI assistants describe Scott correctly. The page is already fast (real visitors pass Core Web Vitals), so the only thing missing is the message.

    Ready to use

    ChecklistHomepage restore list

    [ ] H1: The planning layer for teams that build with coding agents
    [ ] One line on the problem: coding agents amplify the cost of bad planning
    [ ] Three blocks: Sync, Planspaces, PR design review
    [ ] Install command with copy button + quickstart link
    [ ] Book a team demo button
    [ ] Trust center link and SOC 2 line
    [ ] Title, meta description, share image, favicon
    [ ] Noindex removed; homepage submitted for indexing
    [ ] Events: install copy, docs click, demo booked

    Risks and how to handle them

    • The stub is deliberate while the team tests a new direction

      Run the test on its own URL or domain so the product the docs and jobs describe still has a homepage.

    • A quick page undersells the product

      Ship the holding page now and a designed page once the Design Engineer starts.

    Evidence

    • Your websiteHomepage (fetched twice on November 10, 2026): title "Scott | Shop Estimations", H1 "Scott: we help shops with estimations", and one line, "Reachout to [email protected]". No link to the docs, the app, the install command, a demo, pricing or the team.Source
    • PageSpeedMobile test on November 10, 2026: performance 100, accessibility 100 and best practices 100, but SEO 60 because the "Page is blocked from indexing"; no social share image or favicon; 7 requests and 108 KB with no third-party requests. Real visitors pass Core Web Vitals (75th percentile LCP 1.2 s, INP 45 ms, CLS 0).Source
    • Y CombinatorCompany page: "The planning layer for teams that build with coding agents"; founded 2025, Fall 2025 batch, team of 3, New York City, status Active, website tryscott.ai.Source
    • Y CombinatorLaunch post "Scott AI - Agentic Workspace for Software Spec Design" (122 upvotes): Scott "runs several coding agents side by side to expose their disagreements", shows where they disagree and exports a shareable plan to execution tools. Its calls to action are "Try Scott" (tryscott.ai) and "Apply for access" (www.tryscott.ai/apply), which returned 404 on November 10, 2026.Source
    • DocsDocs home: "Sync agent work, share design intent, gather team feedback, and connect planspaces to code." The docs index lists Quickstart, Core concepts, Teams, Documents and versions, Code context, Comments, Reviews and approvals, Sharing, a For You inbox, Team pulse ("design intent, review health, and execution drift"), the agent plugin (Sync, Push, Pull), the GitHub App, PR design review, "The Scott Method" and "The Scott app".Source
  • 02

    The YC launch post (122 upvotes) has two calls to action: "Try Scott" goes to the homepage stub and "Apply for access" goes to www.tryscott.ai/apply, which now returns 404. Launch posts keep getting read for months, and both of its links currently lead nowhere useful.

    Medium impactLow effort First result in 1 week Founder/CTO

    First action

    Add a redirect from /apply to the quickstart today.

    How to measure it

    /apply conversion

    Now
    –
    Target
    30% of /apply visits
    By
    30 days

    Steps

    1. 1

      Add the redirect

      /apply to the quickstart or a team-access page.

      Founder/CTO 1 hour

    2. 2

      Build team access

      A short page: team size, agents used, book a demo.

      Product Engineer 1 day

    3. 3

      Update the launch post

      Point both calls to action at live pages.

      Founder/CEO 1 hour

    4. 4

      Audit old links

      YC page, job posts, social bios, docs footer.

      Founder/CEO 2 hours

    5. 5

      Watch 404s

      Log 404 paths on tryscott.ai weekly.

      Founder/CTO weekly

    6. 6

      Redirect the rest

      Any old path with traffic goes to its closest live page.

      Founder/CTO 1 hour

    First result
    1 week
    Cost
    $0
    Owner
    Founder/CTO

    Smallest test

    Redirect /apply to the quickstart and count the hits for two weeks.

    Why it works

    Redirect /apply to a live page: either the install quickstart or a short team-access form that books a demo. Then update the launch post's links and check every other link Scott has published (launch, YC page, job posts, social bios) for dead ends.

    Someone who clicks "Apply for access" has already decided they want Scott; a 404 loses them at the last step.

    Risks and how to handle them

    • Old applicants expect a waitlist reply

      Say on the page that access is open now and how to start.

    Evidence

    • Y CombinatorLaunch post "Scott AI - Agentic Workspace for Software Spec Design" (122 upvotes): Scott "runs several coding agents side by side to expose their disagreements", shows where they disagree and exports a shareable plan to execution tools. Its calls to action are "Try Scott" (tryscott.ai) and "Apply for access" (www.tryscott.ai/apply), which returned 404 on November 10, 2026.Source
    • Your websiteHomepage (fetched twice on November 10, 2026): title "Scott | Shop Estimations", H1 "Scott: we help shops with estimations", and one line, "Reachout to [email protected]". No link to the docs, the app, the install command, a demo, pricing or the team.Source
  • 03

    tryscott.ai has no robots.txt and no sitemap (both return 404), and the homepage links nowhere, so the well-written docs at docs.tryscott.ai (quickstart, planspaces, PR design review and more) get no authority from the main domain. The domain is a year old with almost no outside mentions, so internal links matter more than usual.

    Medium impactLow effort First result in 1 month Founder/CTO

    First action

    Publish robots.txt and a sitemap.

    How to measure it

    Indexed pages

    Now
    –
    Target
    All docs pages and the homepage indexed
    By
    45 days

    Steps

    1. 1

      Write robots.txt

      Allow crawling and reference the sitemap.

      Founder/CTO 30 minutes

    2. 2

      Publish a sitemap

      Homepage plus every product page as they ship.

      Founder/CTO 1 hour

    3. 3

      Link to the docs

      Homepage and footer links to Quickstart, Planspaces, PR design review.

      Product Engineer 1 hour

    4. 4

      Link back

      Docs header logo and footer link to tryscott.ai.

      Product Engineer 1 hour

    5. 5

      Verify in Search Console

      Both hosts, sitemaps submitted.

      Founder/CTO 1 hour

    6. 6

      Check coverage

      Indexed pages and top queries.

      Founder/CTO monthly

    First result
    1 month
    Cost
    $0
    Owner
    Founder/CTO

    Smallest test

    Ship robots.txt and a sitemap today and check Search Console next week.

    Why it works

    Add a robots.txt that allows crawling and points to a sitemap, a sitemap listing the homepage and the main product pages, and homepage and footer links to the docs' top pages. Check that the docs have their own sitemap and that docs pages link back to tryscott.ai.

    Search engines and AI assistants find and trust pages through links and sitemaps; connecting the main domain and the docs lets queries like "coding agent planning tool" reach Scott.

    Risks and how to handle them

    • Docs pages compete with the homepage for brand searches

      Give the homepage the brand title and the docs page titles their topic.

    Builds on

    Evidence

    • Your websiteSite check (November 10, 2026): /robots.txt and /sitemap.xml return 404; no pricing page, blog, newsletter, careers page or social profile links; 0 Hacker News mentions and no news in the last 30 days. Change tracking shows the homepage headline moved from "Coding Agent Plan Mode with Super Powers" to "Scott: we help shops with estimations"; the domain was registered in October 2025.Source
    • PageSpeedMobile test on November 10, 2026: performance 100, accessibility 100 and best practices 100, but SEO 60 because the "Page is blocked from indexing"; no social share image or favicon; 7 requests and 108 KB with no third-party requests. Real visitors pass Core Web Vitals (75th percentile LCP 1.2 s, INP 45 ms, CLS 0).Source
    • DocsDocs home: "Sync agent work, share design intent, gather team feedback, and connect planspaces to code." The docs index lists Quickstart, Core concepts, Teams, Documents and versions, Code context, Comments, Reviews and approvals, Sharing, a For You inbox, Team pulse ("design intent, review health, and execution drift"), the agent plugin (Sync, Push, Pull), the GitHub App, PR design review, "The Scott Method" and "The Scott app".Source
  • 04

    Scott's quickstart is short: one install command, "scott sync", then open the URL it prints and ask questions across your recent agent sessions. That is a strong first experience, yet nothing on tryscott.ai shows it; a visitor has to find the docs on their own.

    High impactLow effort First result in 2 weeks Product Engineer

    First action

    Time a clean install-to-first-answer run and write the promise from the real number.

    How to measure it

    Copy-to-first-sync rate

    Now
    –
    Target
    40%
    By
    60 days

    Steps

    1. 1

      Time the path

      Fresh machine: install, sync, first answer.

      Product Engineer 2 hours

    2. 2

      Build the block

      Copy button, supported agents, one-line promise.

      Product Engineer 1 day

    3. 3

      Record the clip

      30 seconds from sync to first answer.

      Design Engineer 1 day

    4. 4

      Mirror in docs

      Same block on the docs home.

      Product Engineer 1 hour

    5. 5

      Tag installs

      Record whether an install came from the homepage or docs.

      Founder/CTO 1 day

    6. 6

      Review weekly

      Copies, syncs and where people stop.

      Product Engineer weekly

    First result
    2 weeks
    Cost
    Engineering time
    Owner
    Product Engineer

    Smallest test

    Add the command block to the homepage and count copies for a week.

    Why it works

    Put the install command in a copyable block at the top of the homepage, with "Run scott sync and ask what your agents did this week" under it, and the list of supported agents. Add a 30-second clip of sync producing the first answer. Mirror the block at the top of the docs home.

    Developer tools win when the first useful result comes quickly; showing the exact command and the result removes the guesswork about setup.

    Risks and how to handle them

    • Teams wary of piping a script to the shell

      Link the script source and offer a package-manager alternative.

    Builds on

    Evidence

    • DocsQuickstart: install with one command (curl -fsSL https://tryscott.ai/install | sh), run "scott sync" and open the URL it prints; works with five named coding agents "and other supported agent sessions"; pushing a thread creates a planspace with a shareable app.tryscott.ai/workspace URL; the GitHub App works best with access to all repositories; Scott says it is SOC 2 compliant, does not persist repository contents and fetches code into ephemeral sandboxes only when needed. No page in the docs mentions pricing, plans or seats.Source
    • Your websiteHomepage (fetched twice on November 10, 2026): title "Scott | Shop Estimations", H1 "Scott: we help shops with estimations", and one line, "Reachout to [email protected]". No link to the docs, the app, the install command, a demo, pricing or the team.Source
  • 05

    Sync reads local coding-agent sessions from the last 14 days and "uploads the relevant context to Scott", but the sync page does not say which fields are sent, what is redacted, or how to keep a session out. Agent sessions often contain secrets, customer data and internal code, so this is the first question a security-minded lead will ask, and today the answer is not written down. The docs already make strong claims elsewhere (SOC 2, no persisted repository contents, ephemeral sandboxes).

    High impactLow effort First result in 2 weeks Founder/CTO

    First action

    Write down today's upload fields and retention.

    How to measure it

    Install-to-sync completion

    Now
    –
    Target
    +20%
    By
    60 days

    Steps

    1. 1

      List the fields

      Exactly what sync sends and keeps.

      Founder/CTO 1 day

    2. 2

      Add redaction

      Strip tokens, keys and environment files before upload.

      Product Engineer 3 days

    3. 3

      Add excludes

      Project paths and individual sessions.

      Product Engineer 2 days

    4. 4

      Add a dry run

      Print what would be uploaded.

      Product Engineer 1 day

    5. 5

      Document it

      A "What sync sends" section on the sync page.

      Founder/CTO 1 day

    6. 6

      Link it

      From the install block and the trust center.

      Founder/CEO 1 hour

    First result
    2 weeks
    Cost
    About a week of engineering
    Owner
    Founder/CTO

    Smallest test

    Publish the field list this week and watch for security questions in demos.

    Why it works

    Add a "What sync sends" section: the fields uploaded, what is stripped (keys and tokens, environment files), how long it is kept and who can see it. Ship a simple exclude list (by project path or session) and a dry-run flag that prints what would be uploaded. Link it from the homepage install block and the trust center.

    A clear answer and a dry run let a cautious developer try Scott without a security review, and give the lead who does the review the document they need.

    Risks and how to handle them

    • Excluding too much weakens answers

      Show which excluded sessions would have helped and let users opt back in.

    Evidence

    • DocsSync: reads local coding-agent sessions from the last 14 days and "uploads the relevant context to Scott"; the page does not describe what is excluded or redacted, or how to opt a session out.Source
    • DocsQuickstart: install with one command (curl -fsSL https://tryscott.ai/install | sh), run "scott sync" and open the URL it prints; works with five named coding agents "and other supported agent sessions"; pushing a thread creates a planspace with a shareable app.tryscott.ai/workspace URL; the GitHub App works best with access to all repositories; Scott says it is SOC 2 compliant, does not persist repository contents and fetches code into ephemeral sandboxes only when needed. No page in the docs mentions pricing, plans or seats.Source
  • 06

    Scott runs a Vanta trust center at trust.tryscott.ai, and the docs say Scott is SOC 2 compliant, does not persist repository contents and uses ephemeral sandboxes. But read as text, the trust center page is only its title, "Scott AI Trust Center", so search engines and AI assistants (where buyers increasingly ask "is Scott safe to give repo access?") find nothing.

    Medium impactLow effort First result in 2 weeks Founder/CEO

    First action

    Copy the security lines from the quickstart into a /security page.

    How to measure it

    Security-blocked deals

    Now
    –
    Target
    -50%
    By
    90 days

    Steps

    1. 1

      Gather claims

      SOC 2, data handling and access from the docs and Vanta.

      Founder/CEO 2 hours

    2. 2

      Explain repo access

      What the GitHub App reads and why all repositories helps.

      Founder/CTO 2 hours

    3. 3

      Write the page

      Plain HTML, one screen, with a report-request link.

      Founder/CEO 1 day

    4. 4

      Add subprocessors

      List and purpose for each.

      Founder/CTO 1 hour

    5. 5

      Link it

      Homepage, docs footer, GitHub App install screen.

      Product Engineer 1 hour

    6. 6

      Keep it current

      Review with each Vanta update.

      Founder/CTO quarterly

    First result
    2 weeks
    Cost
    $0
    Owner
    Founder/CEO

    Smallest test

    Publish a five-line /security page this week.

    Why it works

    Publish tryscott.ai/security in plain HTML: SOC 2 status and report request, what the GitHub App can access and why it asks for all repositories, no persisted repository contents, ephemeral sandboxes, what sync uploads (idea 5), subprocessors and contact. Link it from the homepage, docs footer and the trust center.

    Repository access is the biggest ask Scott makes; a readable answer shortens security review and shows up when buyers research.

    Risks and how to handle them

    • Claims overstate the audit scope

      Use the exact wording from the SOC 2 report.

    Builds on

    Evidence

    • Trust centertrust.tryscott.ai is a Vanta trust center; read as page text it shows only the title "Scott AI Trust Center", with no certifications, policies or subprocessors.Source
    • DocsQuickstart: install with one command (curl -fsSL https://tryscott.ai/install | sh), run "scott sync" and open the URL it prints; works with five named coding agents "and other supported agent sessions"; pushing a thread creates a planspace with a shareable app.tryscott.ai/workspace URL; the GitHub App works best with access to all repositories; Scott says it is SOC 2 compliant, does not persist repository contents and fetches code into ephemeral sandboxes only when needed. No page in the docs mentions pricing, plans or seats.Source

This quarter 12

The core plays for the next three months.

  • 07

    There is no pricing page on tryscott.ai and no page in the docs mentions plans, seats or billing, even though the Teams page defines Owner, Editor, Commenter and Viewer roles. The nearest rival, Traycer, publishes a free tier and a team plan at $16 per user a month billed yearly, so buyers comparing tools can price one and not the other.

    High impactMedium effort First result in 1 month Founder/CEO

    First action

    Decide which roles are paid and draft three tiers.

    How to measure it

    Free-to-team conversion

    Now
    –
    Target
    5% of active free workspaces
    By
    1 quarter

    Steps

    1. 1

      Pick the value metric

      Editor seats, with free commenters and viewers.

      Founder/CEO 2 days

    2. 2

      Set prices

      Check against Traycer and code review tools.

      Founder/CEO 2 days

    3. 3

      Build billing

      Seat counting and checkout.

      Product Engineer 2 weeks

    4. 4

      Design the page

      Three tiers, a role table, FAQs.

      Design Engineer 3 days

    5. 5

      Tell current users

      Early-user terms before the switch.

      Founder/CEO 1 week

    6. 6

      Review monthly

      Conversion by tier and team size.

      Founder/CEO monthly

    First result
    1 month
    Cost
    Billing tool fees plus engineering time
    Owner
    Founder/CEO

    Smallest test

    Publish the page with a "talk to us" team price and count demo requests.

    Why it works

    Publish a simple page: Free for one developer (sync, private planspaces), Team per editor seat (shared planspaces, reviews and approvals, PR design review), and Enterprise (idea 19). Make Commenter and Viewer seats free so reviewers never block adoption. Show a monthly and yearly price and a "book a demo" for larger teams.

    Published prices let a team lead start without a sales call, and free reviewer seats match how planspaces spread: one author, many reviewers.

    Risks and how to handle them

    • Pricing too early limits learning

      Keep a generous free tier and revisit after one quarter.

    Builds on

    Evidence

    • DocsTeams and concepts: a planspace combines a chat, a living document, source context, comments, reviews, approvals and members; roles are Owner, Editor, Commenter and Viewer; reviews have pending, in progress, approved and rejected states. Nothing on invitations, seats or billing.Source
    • Competitor researchRivals' own homepages: Traycer, "The <Nerve Center/> for Agentic Coding", a free open-source desktop app with a team plan at $16 per user a month billed yearly ($20 monthly) and 19 coding agents supported; Kiro, "Specs capture what you meant before code is written and verify results after"; CodeRabbit, "The future isn't writing code. It's reviewing it.", with 17K customers, 6M repositories and a claim to be the most installed AI app on GitHub.Source
    • Your websiteSite check (November 10, 2026): /robots.txt and /sitemap.xml return 404; no pricing page, blog, newsletter, careers page or social profile links; 0 Hacker News mentions and no news in the last 30 days. Change tracking shows the homepage headline moved from "Coding Agent Plan Mode with Super Powers" to "Scott: we help shops with estimations"; the domain was registered in October 2025.Source
  • 08

    Scott sits between three loud categories: agent workspaces (Traycer, "The Nerve Center for Agentic Coding"), spec-driven IDEs (Kiro, specs "before code is written") and AI code review (CodeRabbit, 17K customers). Its own pitch, "Does this implementation match the design intent the team reviewed?", is a distinct answer, but there is no page that says so.

    Medium impactMedium effort First result in 2 months Founder/CEO

    First action

    Ask the last ten demo calls which tools they compared.

    How to measure it

    Comparison-page installs

    Now
    –
    Target
    50 a month
    By
    1 quarter

    Steps

    1. 1

      Pick rivals

      The three named most in calls.

      Founder/CEO 1 day

    2. 2

      Check facts

      Each rival's current features and prices from their own pages.

      Founder/CEO 2 days

    3. 3

      Write pages

      Fair tables and a use-both section.

      Founder/CEO 1 week

    4. 4

      Design template

      One reusable layout.

      Design Engineer 2 days

    5. 5

      Publish

      Add to the sitemap and footer.

      Product Engineer 1 day

    6. 6

      Refresh

      Re-check rival facts.

      Founder/CEO quarterly

    First result
    2 months
    Cost
    Writing time
    Owner
    Founder/CEO

    Smallest test

    Publish the Scott and CodeRabbit page first and track installs.

    Why it works

    Write three fair pages: Scott and Traycer (plan and approve vs run agents side by side), Scott and Kiro (works with the agents you already use vs one environment), Scott and CodeRabbit (checks PRs against the approved design vs code quality, and how they work together). Each has a short table, a "use both" section and the install block.

    Buyers search for alternatives and comparisons once they know the category; clear pages also teach AI assistants where Scott fits.

    Risks and how to handle them

    • Claims about rivals go out of date

      Date each page and cite the rival's own page.

    Builds on

    Evidence

    • Competitor researchRivals' own homepages: Traycer, "The <Nerve Center/> for Agentic Coding", a free open-source desktop app with a team plan at $16 per user a month billed yearly ($20 monthly) and 19 coding agents supported; Kiro, "Specs capture what you meant before code is written and verify results after"; CodeRabbit, "The future isn't writing code. It's reviewing it.", with 17K customers, 6M repositories and a claim to be the most installed AI app on GitHub.Source
    • DocsPR design review: "The planspace URL is the contract between design intent and execution." Scott asks "Does this implementation match the design intent the team reviewed?", posts a GitHub comment on architectural alignment, missing requirements and drift, and links back to the planspace. It only runs when a PR includes a planspace URL and does not block merges or add a required status check by default.Source
  • 09

    Scott's value is in shared planspaces with reviews and approvals, but the first run ends in a personal Sync Chat over 14 days of sessions. The step from "interesting answer" to "shared planspace my team reviews" is left to the user.

    High impactMedium effort First result in 1 month Product Engineer

    First action

    Measure how many syncs today lead to a planspace within a week.

    How to measure it

    7-day planspace activation

    Now
    –
    Target
    35%
    By
    1 quarter

    Steps

    1. 1

      Measure today

      Sync to first planspace and first review.

      Founder/CTO 2 days

    2. 2

      Rank threads

      Size, recency and open questions.

      Product Engineer 1 week

    3. 3

      Design the card

      Three suggestions with one button.

      Design Engineer 3 days

    4. 4

      Add the invite

      Push and invite a reviewer in one step.

      Product Engineer 3 days

    5. 5

      A/B test

      Suggestions vs current Sync Chat.

      Product Engineer 3 weeks

    6. 6

      Iterate

      Tune ranking from accepted suggestions.

      Product Engineer monthly

    First result
    1 month
    Cost
    Engineering time
    Owner
    Product Engineer

    Smallest test

    Show one suggested thread after sync for a week and count pushes.

    Why it works

    At the end of the first sync, show three suggested threads (largest, most recent unfinished, most debated) with a one-click "push to planspace and invite a reviewer". Track time from install to first shared planspace and first review as the activation measure.

    A concrete next step on the user's own work beats a blank prompt, and the first reviewer invite starts team adoption.

    Risks and how to handle them

    • Suggestions surface private work

      Pushing stays opt-in and private until the user shares.

    Builds on

    Evidence

    • DocsQuickstart: install with one command (curl -fsSL https://tryscott.ai/install | sh), run "scott sync" and open the URL it prints; works with five named coding agents "and other supported agent sessions"; pushing a thread creates a planspace with a shareable app.tryscott.ai/workspace URL; the GitHub App works best with access to all repositories; Scott says it is SOC 2 compliant, does not persist repository contents and fetches code into ephemeral sandboxes only when needed. No page in the docs mentions pricing, plans or seats.Source
    • DocsSync: reads local coding-agent sessions from the last 14 days and "uploads the relevant context to Scott"; the page does not describe what is excluded or redacted, or how to opt a session out.Source
    • DocsTeams and concepts: a planspace combines a chat, a living document, source context, comments, reviews, approvals and members; roles are Owner, Editor, Commenter and Viewer; reviews have pending, in progress, approved and rejected states. Nothing on invitations, seats or billing.Source
  • 10

    Each pushed thread becomes a planspace with a shareable app.tryscott.ai/workspace URL, and planspaces are built for other people: commenters, reviewers and approvers. Every reviewer who opens a link is a teammate seeing Scott at its best, but nothing says the docs give that viewer a path to their own planspaces.

    High impactMedium effort First result in 1 month Design Engineer

    First action

    Count how many unique non-users opened planspace links last month.

    How to measure it

    Reviewer-to-author conversion

    Now
    –
    Target
    15% within 30 days
    By
    1 quarter

    Steps

    1. 1

      Measure viewers

      Non-user opens per planspace.

      Founder/CTO 1 day

    2. 2

      Design the frame

      Light header and sign-in to comment.

      Design Engineer 1 week

    3. 3

      Build sign-in

      Comment after one-click sign-in.

      Product Engineer 1 week

    4. 4

      Prompt after review

      Install and sync your own sessions.

      Design Engineer 3 days

    5. 5

      Respect privacy

      No frame on private or enterprise planspaces.

      Founder/CTO 1 day

    6. 6

      Track the loop

      Invites per user and conversion.

      Founder/CTO weekly

    First result
    1 month
    Cost
    Engineering and design time
    Owner
    Design Engineer

    Smallest test

    Add a one-line "Plan with Scott" link to shared planspaces for two weeks.

    Why it works

    On a planspace opened by someone without an account, add a light frame: "Reviewed with Scott: plan your next agent task the same way", one-click sign-in to comment, and after their first review a prompt to install and sync their own sessions. Count invites per active user and reviewer-to-author conversion as the loop's health.

    Collaboration tools grow through the people invited to collaborate; the reviewer arrives with context and a reason to come back.

    Risks and how to handle them

    • Authors dislike branding on their work

      Keep it small and let paid teams turn it off.

    Evidence

    • DocsQuickstart: install with one command (curl -fsSL https://tryscott.ai/install | sh), run "scott sync" and open the URL it prints; works with five named coding agents "and other supported agent sessions"; pushing a thread creates a planspace with a shareable app.tryscott.ai/workspace URL; the GitHub App works best with access to all repositories; Scott says it is SOC 2 compliant, does not persist repository contents and fetches code into ephemeral sandboxes only when needed. No page in the docs mentions pricing, plans or seats.Source
    • DocsTeams and concepts: a planspace combines a chat, a living document, source context, comments, reviews, approvals and members; roles are Owner, Editor, Commenter and Viewer; reviews have pending, in progress, approved and rejected states. Nothing on invitations, seats or billing.Source
  • 11

    PR design review posts a GitHub comment on alignment, missing requirements and drift, but only when a PR includes a planspace URL, and it does not add a status check by default. That comment is read by every reviewer on the PR, including teammates who never installed Scott, and a check is what teams wire into their merge rules.

    Medium impactMedium effort First result in 1 month Founder/CTO

    First action

    Count linked vs unlinked PRs in connected repositories.

    How to measure it

    PR link rate

    Now
    –
    Target
    40%
    By
    1 quarter

    Steps

    1. 1

      Measure link rate

      Linked PRs per connected repository.

      Founder/CTO 1 day

    2. 2

      Build the check

      Pass, warn or fail on drift; off by default.

      Product Engineer 2 weeks

    3. 3

      Add the footer

      Link to the planspace and to Scott.

      Product Engineer 1 day

    4. 4

      Nudge unlinked PRs

      Optional, one comment per PR.

      Product Engineer 3 days

    5. 5

      Document it

      Update the PR design review page.

      Founder/CTO 1 day

    6. 6

      Review noise

      Track dismissals and opt-outs.

      Founder/CTO weekly

    First result
    1 month
    Cost
    Engineering time
    Owner
    Founder/CTO

    Smallest test

    Add the comment footer and watch new installs from GitHub referrals.

    Why it works

    Add an optional "design drift" status check teams can make required, a short comment footer ("Plan linked with Scott: see the approved design"), and a gentle nudge on unlinked PRs in connected repositories: "No plan linked. Push your agent thread to Scott." Report how many PRs are linked per team as a usage measure.

    Required checks make a tool part of the workflow, and PR comments put the product in front of the whole team every day.

    Risks and how to handle them

    • Nudges feel like spam

      Off by default, one per PR, easy to mute.

    Evidence

    • DocsPR design review: "The planspace URL is the contract between design intent and execution." Scott asks "Does this implementation match the design intent the team reviewed?", posts a GitHub comment on architectural alignment, missing requirements and drift, and links back to the planspace. It only runs when a PR includes a planspace URL and does not block merges or add a required status check by default.Source
  • 12

    Scott already tracks "design intent, review health, and execution drift" in Team pulse and has a For You inbox, but both live inside the app. A team lead who does not open Scott every day never sees the value it creates.

    Medium impactLow effort First result in 1 month Product Engineer

    First action

    Draft the email from one active team's last week.

    How to measure it

    Week-4 team retention

    Now
    –
    Target
    +15%
    By
    1 quarter

    Steps

    1. 1

      Pick the numbers

      Approvals, drift caught, waiting reviews.

      Founder/CEO 1 day

    2. 2

      Design the email

      One screen, one call to action.

      Design Engineer 2 days

    3. 3

      Build the job

      Weekly per-team send.

      Product Engineer 3 days

    4. 4

      Add settings

      Frequency and opt-out.

      Product Engineer 1 day

    5. 5

      Test

      Half of teams first.

      Product Engineer 4 weeks

    6. 6

      Compare

      Return visits and retention.

      Founder/CTO monthly

    First result
    1 month
    Cost
    Email service plus engineering time
    Owner
    Product Engineer

    Smallest test

    Hand-send the digest to five teams for two weeks.

    Why it works

    Send a Monday email per team: plans approved and shipped, PRs where drift was caught, reviews waiting longest, and one suggested next plan. Include a forwardable summary for the engineering manager.

    A weekly digest brings people back and makes the tool's value visible to whoever approves the budget.

    Risks and how to handle them

    • Empty weeks look bad

      Skip the send when nothing happened and suggest a first plan instead.

    Evidence

    • DocsDocs home: "Sync agent work, share design intent, gather team feedback, and connect planspaces to code." The docs index lists Quickstart, Core concepts, Teams, Documents and versions, Code context, Comments, Reviews and approvals, Sharing, a For You inbox, Team pulse ("design intent, review health, and execution drift"), the agent plugin (Sync, Push, Pull), the GitHub App, PR design review, "The Scott Method" and "The Scott app".Source
    • DocsPR design review: "The planspace URL is the contract between design intent and execution." Scott asks "Does this implementation match the design intent the team reviewed?", posts a GitHub comment on architectural alignment, missing requirements and drift, and links back to the planspace. It only runs when a PR includes a planspace URL and does not block merges or add a required status check by default.Source
  • 13

    The docs index lists "The Scott Method", and the job posts carry a sharp thesis: "Coding agents amplify the cost of bad planning", because agents write far more code while teams review and align at the same pace. That is a point of view other engineering leaders would share, but it sits in docs and job posts, not on the site.

    Medium impactLow effort First result in 1 month Founder/CEO

    First action

    Move the docs version to a public URL on tryscott.ai with a byline.

    How to measure it

    Essay-to-install rate

    Now
    –
    Target
    3%
    By
    1 quarter

    Steps

    1. 1

      Draft

      Problem, four moves, one example.

      Founder/CEO 3 days

    2. 2

      Add a diagram

      Sync, push, review, pull, check.

      Design Engineer 1 day

    3. 3

      Publish

      On tryscott.ai with the install block.

      Product Engineer 1 day

    4. 4

      Homepage summary

      Second screen links to it.

      Design Engineer 1 day

    5. 5

      Distribute

      Founder posts and engineering forums.

      Founder/CEO 1 week

    6. 6

      Measure

      Readers, shares and installs.

      Founder/CTO monthly

    First result
    1 month
    Cost
    Writing time
    Owner
    Founder/CEO

    Smallest test

    Post the thesis and the four moves as one founder thread and count clicks.

    Why it works

    Turn the method into a short essay on tryscott.ai: the problem, the four moves (sync, push to a planspace, review and approve, pull into execution and check the PR), and a worked example. Put a summary on the homepage and post the essay where engineering leaders read.

    A named method gives people a way to talk about the problem and a reason to link to Scott, and it sells the product's workflow without a demo.

    Risks and how to handle them

    • Reads as a sales pitch

      Make the method useful even without Scott.

    Builds on

    Evidence

    • DocsDocs home: "Sync agent work, share design intent, gather team feedback, and connect planspaces to code." The docs index lists Quickstart, Core concepts, Teams, Documents and versions, Code context, Comments, Reviews and approvals, Sharing, a For You inbox, Team pulse ("design intent, review health, and execution drift"), the agent plugin (Sync, Push, Pull), the GitHub App, PR design review, "The Scott Method" and "The Scott app".Source
    • HiringTwo open roles on Y Combinator's job board: Product Engineer ($140K-$250K, 0.50%-2.00% equity) and Design Engineer ($140K-$250K, 0.50%-1.50%), New York or San Francisco, new grads welcome. The posts' thesis: "Coding agents amplify the cost of bad planning"; agents write far more code while teams still review and align on intent at the same pace.Source
  • 14

    Scott's launch hook was showing where several coding agents disagree about a design, and it earned 122 upvotes. Since then Scott has had no Hacker News mentions and no news coverage, and the site links to no social profiles. Those disagreements are naturally interesting content for developers.

    Medium impactLow effort First result in 1 month Founder/CEO

    First action

    Write the first post from one of Scott's own planspaces.

    How to measure it

    Series-sourced installs

    Now
    –
    Target
    30 a month
    By
    1 quarter

    Steps

    1. 1

      Pick a task

      From Scott's own work first.

      Founder/CTO weekly

    2. 2

      Write it up

      Task, options, divergence, decision.

      Founder/CEO 2 hours

    3. 3

      Make one image

      The disagreement view as a screenshot.

      Design Engineer 1 hour

    4. 4

      Post

      Founder and company accounts, with a tracked link.

      Founder/CEO weekly

    5. 5

      Collect

      Archive page on tryscott.ai.

      Product Engineer 1 day

    6. 6

      Invite teams

      Ask users to share anonymized examples.

      Founder/CEO monthly

    First result
    1 month
    Cost
    Two hours a week
    Owner
    Founder/CEO

    Smallest test

    Publish four weekly posts and compare installs with the month before.

    Why it works

    Each week, publish one short post: a real task (from Scott's own work or a team that opts in, anonymized), the two or three designs the agents proposed, where they diverged and which choice the humans made and why. Post it from the founder accounts and the company account, and collect them on a page on tryscott.ai.

    Concrete design debates are shareable, show the product without a pitch, and build an audience that the site currently lacks.

    Risks and how to handle them

    • Exposing customer code

      Use Scott's own work or written opt-in with anonymization.

    Evidence

    • Y CombinatorLaunch post "Scott AI - Agentic Workspace for Software Spec Design" (122 upvotes): Scott "runs several coding agents side by side to expose their disagreements", shows where they disagree and exports a shareable plan to execution tools. Its calls to action are "Try Scott" (tryscott.ai) and "Apply for access" (www.tryscott.ai/apply), which returned 404 on November 10, 2026.Source
    • Your websiteSite check (November 10, 2026): /robots.txt and /sitemap.xml return 404; no pricing page, blog, newsletter, careers page or social profile links; 0 Hacker News mentions and no news in the last 30 days. Change tracking shows the homepage headline moved from "Coding Agent Plan Mode with Super Powers" to "Scott: we help shops with estimations"; the domain was registered in October 2025.Source
  • 15

    Scott's GitHub App is how PR design review works, and AI code review is a crowded install category on GitHub: CodeRabbit says it is the most installed AI app on GitHub, with 6M repositories. Teams looking for help with agent-written PRs browse GitHub's app listings, so a clear listing that explains the difference (checks the PR against the approved design) can catch them.

    Medium impactLow effort First result in 1 month Founder/CTO

    First action

    Check how the app appears when searched on GitHub today.

    How to measure it

    Listing installs

    Now
    –
    Target
    20 a month
    By
    1 quarter

    Steps

    1. 1

      Check today

      How the app appears in GitHub search.

      Founder/CTO 1 hour

    2. 2

      Write the listing

      Value line, permissions, works-alongside line.

      Founder/CEO 1 day

    3. 3

      Add screenshots

      A drift comment and a planspace link.

      Design Engineer 1 day

    4. 4

      Link security

      The /security page from idea 6.

      Founder/CTO 1 hour

    5. 5

      Submit

      Marketplace review.

      Founder/CTO 1 week

    6. 6

      Track

      Installs by source.

      Founder/CTO monthly

    First result
    1 month
    Cost
    $0
    Owner
    Founder/CTO

    Smallest test

    Rewrite the app's public description and watch installs for a month.

    Why it works

    Publish or polish a public GitHub Marketplace listing for the Scott GitHub App: one-line value ("Check every PR against the design your team approved"), screenshots of a drift comment, the permissions and why, and a link to the security page. Add a "Works alongside your code review bot" line so it is a complement, not a swap.

    Developers look for tools where they will use them; a listing with clear permissions and a distinct promise turns browsing into installs.

    Risks and how to handle them

    • Installs without a planspace see nothing

      Show an install screen that leads to sync and the first push.

    Builds on

    Evidence

    • DocsPR design review: "The planspace URL is the contract between design intent and execution." Scott asks "Does this implementation match the design intent the team reviewed?", posts a GitHub comment on architectural alignment, missing requirements and drift, and links back to the planspace. It only runs when a PR includes a planspace URL and does not block merges or add a required status check by default.Source
    • Competitor researchRivals' own homepages: Traycer, "The <Nerve Center/> for Agentic Coding", a free open-source desktop app with a team plan at $16 per user a month billed yearly ($20 monthly) and 19 coding agents supported; Kiro, "Specs capture what you meant before code is written and verify results after"; CodeRabbit, "The future isn't writing code. It's reviewing it.", with 17K customers, 6M repositories and a claim to be the most installed AI app on GitHub.Source
  • 16

    The Scott agent plugin is "the bundle of skills and local scott commands" that moves work in and out of Scott, and the quickstart names five supported coding agents. Traycer advertises 19. Developers increasingly discover add-ons inside their agent's own plugin and skills directories, not on vendor websites.

    Medium impactMedium effort First result in 2 months Founder/CTO

    First action

    List which supported agents have a plugin or skills directory and its submission rules.

    How to measure it

    Directory-sourced installs

    Now
    –
    Target
    25% of new installs
    By
    2 quarters

    Steps

    1. 1

      Map directories

      Official and community lists per agent.

      Founder/CTO 1 day

    2. 2

      Package

      Each agent's plugin format.

      Product Engineer 2 weeks

    3. 3

      Write listings

      Two lines and a quickstart link.

      Founder/CEO 1 day

    4. 4

      Supported agents page

      Setup per agent on tryscott.ai.

      Design Engineer 3 days

    5. 5

      Submit

      One directory at a time.

      Founder/CTO 2 weeks

    6. 6

      Add agents

      Next agents by user requests.

      Product Engineer quarterly

    First result
    2 months
    Cost
    Engineering time
    Owner
    Founder/CTO

    Smallest test

    List the plugin in one directory and track installs for a month.

    Why it works

    Package the plugin for each supported agent's official or community directory, with a two-line description and a link to the quickstart. Publish a "Supported agents" page on tryscott.ai with setup per agent, and add agents in order of user demand.

    Directory listings put Scott in front of developers at the moment they are extending their agent, and a supported-agents page answers "does it work with mine?" before they leave.

    Risks and how to handle them

    • Directory rules change often

      Keep one owner and review listings quarterly.

    Evidence

    • DocsQuickstart: install with one command (curl -fsSL https://tryscott.ai/install | sh), run "scott sync" and open the URL it prints; works with five named coding agents "and other supported agent sessions"; pushing a thread creates a planspace with a shareable app.tryscott.ai/workspace URL; the GitHub App works best with access to all repositories; Scott says it is SOC 2 compliant, does not persist repository contents and fetches code into ephemeral sandboxes only when needed. No page in the docs mentions pricing, plans or seats.Source
    • Competitor researchRivals' own homepages: Traycer, "The <Nerve Center/> for Agentic Coding", a free open-source desktop app with a team plan at $16 per user a month billed yearly ($20 monthly) and 19 coding agents supported; Kiro, "Specs capture what you meant before code is written and verify results after"; CodeRabbit, "The future isn't writing code. It's reviewing it.", with 17K customers, 6M repositories and a claim to be the most installed AI app on GitHub.Source
  • 17

    A planspace combines a chat, a living document, source context, comments, reviews and approvals, which is hard to picture from text. Today a visitor has to install a CLI and connect GitHub to see one, and the homepage shows nothing at all.

    Medium impactMedium effort First result in 1 month Design Engineer

    First action

    Pick one internal plan that is safe to show.

    How to measure it

    Example-to-install rate

    Now
    –
    Target
    10%
    By
    1 quarter

    Steps

    1. 1

      Pick the plan

      One of Scott's own, safe to publish.

      Founder/CTO 1 day

    2. 2

      Clean it

      Remove anything internal.

      Founder/CTO 1 day

    3. 3

      Build public view

      Read-only, no sign-in.

      Design Engineer 1 week

    4. 4

      Annotate

      Short callouts on each part.

      Design Engineer 2 days

    5. 5

      Link it

      Homepage, docs home, launch post.

      Product Engineer 1 hour

    6. 6

      Measure

      Views to installs.

      Founder/CTO monthly

    First result
    1 month
    Cost
    Design and engineering time
    Owner
    Design Engineer

    Smallest test

    Share screenshots of one planspace on the homepage first.

    Why it works

    Build one public example planspace from Scott's own work (for example, how PR design review was planned): the agent thread, the document versions, comments, an approval, and the linked PR with its drift comment. Link it from the homepage as "See a real planspace".

    Seeing a finished example answers "what would my team get?" in a minute, without asking for repository access first.

    Risks and how to handle them

    • Example goes stale

      Refresh it with each major release.

    Builds on

    Evidence

    • DocsTeams and concepts: a planspace combines a chat, a living document, source context, comments, reviews, approvals and members; roles are Owner, Editor, Commenter and Viewer; reviews have pending, in progress, approved and rejected states. Nothing on invitations, seats or billing.Source
    • Your websiteHomepage (fetched twice on November 10, 2026): title "Scott | Shop Estimations", H1 "Scott: we help shops with estimations", and one line, "Reachout to [email protected]". No link to the docs, the app, the install command, a demo, pricing or the team.Source
    • HiringTwo open roles on Y Combinator's job board: Product Engineer ($140K-$250K, 0.50%-2.00% equity) and Design Engineer ($140K-$250K, 0.50%-1.50%), New York or San Francisco, new grads welcome. The posts' thesis: "Coding agents amplify the cost of bad planning"; agents write far more code while teams still review and align on intent at the same pace.Source
  • 23

    The homepage loads no third-party tools at all (7 requests, none third-party), and Scott's journey has clear steps (install, sync, push to a planspace, review and approve, link a PR). Without events at each step, the team cannot tell whether the homepage, the sync, or the first review is where people drop off.

    Medium impactLow effort First result in 2 weeks Founder/CTO

    First action

    Write the six event names and where each fires.

    How to measure it

    Steps instrumented

    Now
    –
    Target
    6 of 6
    By
    2 weeks

    Steps

    1. 1

      Name events

      Six steps with a source tag.

      Founder/CTO 1 hour

    2. 2

      Pick a tool

      First-party, privacy-friendly, light.

      Founder/CTO 1 day

    3. 3

      Instrument web

      Homepage and install copy.

      Product Engineer 1 day

    4. 4

      Instrument product

      Sync, planspace, approval, linked PR.

      Product Engineer 2 days

    5. 5

      Build the chart

      Weekly cohorts.

      Founder/CTO 1 day

    6. 6

      Review weekly

      Pick the next step to fix.

      Founder/CEO weekly

    First result
    2 weeks
    Cost
    Free tier of an analytics tool
    Owner
    Founder/CTO

    Smallest test

    Log install copies and first syncs this week.

    Why it works

    Define six events (homepage visit, install command copied, first sync, first planspace, first approved review, first linked PR) with a source tag, and one weekly chart by cohort. Use a privacy-friendly, first-party setup so the page stays fast and no session content is collected.

    A three-person team has to pick one fix at a time; a funnel shows which step to fix first and proves whether each idea in this playbook works.

    Risks and how to handle them

    • Tracking worries privacy-minded developers

      Count events only, never session content, and say so on the security page.

    Evidence

    • PageSpeedMobile test on November 10, 2026: performance 100, accessibility 100 and best practices 100, but SEO 60 because the "Page is blocked from indexing"; no social share image or favicon; 7 requests and 108 KB with no third-party requests. Real visitors pass Core Web Vitals (75th percentile LCP 1.2 s, INP 45 ms, CLS 0).Source
    • DocsQuickstart: install with one command (curl -fsSL https://tryscott.ai/install | sh), run "scott sync" and open the URL it prints; works with five named coding agents "and other supported agent sessions"; pushing a thread creates a planspace with a shareable app.tryscott.ai/workspace URL; the GitHub App works best with access to all repositories; Scott says it is SOC 2 compliant, does not persist repository contents and fetches code into ephemeral sandboxes only when needed. No page in the docs mentions pricing, plans or seats.Source
    • DocsTeams and concepts: a planspace combines a chat, a living document, source context, comments, reviews, approvals and members; roles are Owner, Editor, Commenter and Viewer; reviews have pending, in progress, approved and rejected states. Nothing on invitations, seats or billing.Source

Big bets 3

Larger plays that can change the trajectory.

  • 18

    Scott is the rare tool that sees both the approved design (the planspace) and the code that ships (the linked PR), and it already classifies drift as unexpected or intentional. The thesis in its job posts, that agents write far more code than teams can align on, is widely felt but rarely measured.

    High impactHigh effort First result in 2 quarters Founder/CEO

    First action

    Count how many linked PRs and drift findings exist today and get opt-in from willing teams.

    How to measure it

    Report-sourced installs

    Now
    –
    Target
    200 in the first month
    By
    2 quarters

    Steps

    1. 1

      Check the data

      Enough linked PRs and drift findings.

      Founder/CTO 1 week

    2. 2

      Get consent

      Opt-in terms for aggregate use.

      Founder/CEO 1 month

    3. 3

      Define measures

      Drift types and rates.

      Founder/CTO 1 week

    4. 4

      Analyze

      Aggregate and anonymize.

      Product Engineer 2 weeks

    5. 5

      Design

      Report page and charts.

      Design Engineer 2 weeks

    6. 6

      Launch

      Post, press and talks.

      Founder/CEO 1 month

    7. 7

      Offer benchmarks

      Teams compare their own rate.

      Product Engineer 1 month

    First result
    2 quarters
    Cost
    About a month of team time
    Owner
    Founder/CEO

    Smallest test

    Publish drift numbers from Scott's own repositories as a blog post.

    Why it works

    With customer opt-in and full anonymization, publish a twice-yearly report: share of linked PRs with drift, the most common kinds (missing requirements, new dependencies, changed abstractions), how much drift is intentional, and how approval before execution changes the rate. Release the method openly and offer teams their own benchmark.

    Original data earns links, press and conference talks, and makes Scott the reference for a problem every engineering leader is asking about.

    Risks and how to handle them

    • Too little data at this stage

      Start with Scott's own repositories and a small opt-in cohort, labelled as such.

    • Customer privacy

      Aggregate only, with written opt-in.

    Builds on

    Evidence

    • DocsPR design review: "The planspace URL is the contract between design intent and execution." Scott asks "Does this implementation match the design intent the team reviewed?", posts a GitHub comment on architectural alignment, missing requirements and drift, and links back to the planspace. It only runs when a PR includes a planspace URL and does not block merges or add a required status check by default.Source
    • HiringTwo open roles on Y Combinator's job board: Product Engineer ($140K-$250K, 0.50%-2.00% equity) and Design Engineer ($140K-$250K, 0.50%-1.50%), New York or San Francisco, new grads welcome. The posts' thesis: "Coding agents amplify the cost of bad planning"; agents write far more code while teams still review and align on intent at the same pace.Source
    • DocsDocs home: "Sync agent work, share design intent, gather team feedback, and connect planspaces to code." The docs index lists Quickstart, Core concepts, Teams, Documents and versions, Code context, Comments, Reviews and approvals, Sharing, a For You inbox, Team pulse ("design intent, review health, and execution drift"), the agent plugin (Sync, Push, Pull), the GitHub App, PR design review, "The Scott Method" and "The Scott app".Source
  • 19

    Scott asks for the two things large companies guard most: local agent session history (sync uploads the last 14 days) and access to all repositories (the GitHub App works best that way). It already has SOC 2, a Vanta trust center, no persisted repository contents and ephemeral sandboxes, which is the foundation larger buyers need, but none of it is packaged or priced.

    High impactHigh effort First result in 2 quarters Founder/CEO

    First action

    Ask the three largest current teams what blocks a wider rollout.

    How to measure it

    Enterprise contracts

    Now
    –
    Target
    3
    By
    2 quarters

    Steps

    1. 1

      Interview teams

      Rollout blockers at the largest accounts.

      Founder/CEO 2 weeks

    2. 2

      Build SSO

      SSO and SCIM.

      Product Engineer 1 month

    3. 3

      Add audit logs

      Sync, push, pull, review events.

      Founder/CTO 1 month

    4. 4

      Add scoping

      Repository and session limits, retention.

      Product Engineer 1 month

    5. 5

      Paper

      DPA and security questionnaire answers.

      Founder/CEO 2 weeks

    6. 6

      Price and sell

      Enterprise tier on the pricing page and demos.

      Founder/CEO ongoing

    First result
    2 quarters
    Cost
    A quarter of engineering plus legal review
    Owner
    Founder/CEO

    Smallest test

    Offer audit logs and SSO to one growing team as a paid pilot.

    Why it works

    Package an Enterprise plan: SSO and SCIM, audit logs for every sync, push, pull and PR review, repository and session scoping, data retention settings, a data processing agreement and a named contact. Sell it through demos to engineering organizations that already allow coding agents, starting with existing teams that are growing.

    Planning and design approval matter most in large teams, which pay more per seat, and security controls are what turn a team pilot into a company rollout.

    Risks and how to handle them

    • Enterprise work slows the core product

      Build only what a signed or late-stage deal needs.

    Builds on

    Evidence

    • DocsQuickstart: install with one command (curl -fsSL https://tryscott.ai/install | sh), run "scott sync" and open the URL it prints; works with five named coding agents "and other supported agent sessions"; pushing a thread creates a planspace with a shareable app.tryscott.ai/workspace URL; the GitHub App works best with access to all repositories; Scott says it is SOC 2 compliant, does not persist repository contents and fetches code into ephemeral sandboxes only when needed. No page in the docs mentions pricing, plans or seats.Source
    • Trust centertrust.tryscott.ai is a Vanta trust center; read as page text it shows only the title "Scott AI Trust Center", with no certifications, policies or subprocessors.Source
    • DocsSync: reads local coding-agent sessions from the last 14 days and "uploads the relevant context to Scott"; the page does not describe what is excluded or redacted, or how to opt a session out.Source
  • 20

    Scott launched on "runs several coding agents side by side to expose their disagreements" and the homepage headline was once "Coding Agent Plan Mode with Super Powers". The current product centres on sync and planspaces, which is stronger for teams but harder to try in a minute. The original hook is still the most shareable thing Scott does.

    High impactHigh effort First result in 1 quarter Product Engineer

    First action

    Prototype the comparison on three public repositories.

    How to measure it

    Tool-to-install rate

    Now
    –
    Target
    8%
    By
    2 quarters

    Steps

    1. 1

      Prototype

      Three public repositories, two agents each.

      Product Engineer 1 week

    2. 2

      Design the view

      Side-by-side divergences, one decision per row.

      Design Engineer 2 weeks

    3. 3

      Set limits

      Rate limits and cost caps.

      Founder/CTO 3 days

    4. 4

      Add the handoff

      Result to planspace to install.

      Product Engineer 1 week

    5. 5

      Launch

      A second launch post and founder demo.

      Founder/CEO 1 week

    6. 6

      Measure

      Runs, shares, installs, cost per install.

      Founder/CTO monthly

    First result
    1 quarter
    Cost
    Model usage per run plus a quarter of engineering
    Owner
    Product Engineer

    Smallest test

    Run the comparison by hand for ten developers who reply to a founder post.

    Why it works

    Build a free web tool on tryscott.ai: describe a task (or paste a public repository URL), two agents draft plans, and Scott shows the key divergences side by side. The result ends with "Turn this into a planspace for your team", which leads to install and sync. Rate-limit it and require sign-in after the first run.

    A free, instantly useful tool gets shared and indexed, and every result is a natural starting point for a planspace.

    Risks and how to handle them

    • Usage costs climb

      Cap free runs and require sign-in after the first.

    • Distracts from team features

      Reuse the existing planning engine; no separate product.

    Builds on

    Evidence

    • Y CombinatorLaunch post "Scott AI - Agentic Workspace for Software Spec Design" (122 upvotes): Scott "runs several coding agents side by side to expose their disagreements", shows where they disagree and exports a shareable plan to execution tools. Its calls to action are "Try Scott" (tryscott.ai) and "Apply for access" (www.tryscott.ai/apply), which returned 404 on November 10, 2026.Source
    • Your websiteSite check (November 10, 2026): /robots.txt and /sitemap.xml return 404; no pricing page, blog, newsletter, careers page or social profile links; 0 Hacker News mentions and no news in the last 30 days. Change tracking shows the homepage headline moved from "Coding Agent Plan Mode with Super Powers" to "Scott: we help shops with estimations"; the domain was registered in October 2025.Source

Wildcards 2

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

  • 21

    The homepage now says "Scott: we help shops with estimations", while YC, the launch, the docs and both job posts describe a planning layer for coding agents. If estimations for shops is a genuine new direction, the two audiences (shop owners and software teams) need different words, proof and calls to action, and today neither gets a page that serves them.

    Medium impactMedium effort First result in 1 month Founder/CEO

    First action

    Write one sentence on who tryscott.ai is for, and share it with the team.

    How to measure it

    Primary-action rate per audience

    Now
    –
    Target
    20% on each page
    By
    60 days

    Steps

    1. 1

      Decide

      Which product owns tryscott.ai.

      Founder/CEO 1 day

    2. 2

      Pick the second home

      A separate domain or a /shops path.

      Founder/CEO 1 day

    3. 3

      Build the shop page

      Headline, example estimate, sign-up.

      Design Engineer 1 week

    4. 4

      Align channels

      YC page, docs and job posts match the decision.

      Founder/CEO 1 day

    5. 5

      Separate tracking

      Each product's funnel on its own.

      Founder/CTO 1 day

    6. 6

      Review

      Traffic and conversion by audience.

      Founder/CEO monthly

    First result
    1 month
    Cost
    A domain and a week of work
    Owner
    Founder/CEO

    Smallest test

    Move the shop line to /shops and put the planning product back on the homepage.

    Why it works

    Decide in writing which product tryscott.ai is for. If estimations is a test, give it its own domain or a /shops page with its own headline, example estimate and sign-up, and keep the planning product on the homepage. If it replaces the planning product, update the YC page, the docs and the job posts together so every channel tells one story.

    One domain, one audience, one promise: each product gets a page that converts, and neither confuses the other's visitors or search results.

    Risks and how to handle them

    • Splitting focus across two products

      Set a date to review the test and stop the weaker one.

    Builds on

    Evidence

    • Your websiteHomepage (fetched twice on November 10, 2026): title "Scott | Shop Estimations", H1 "Scott: we help shops with estimations", and one line, "Reachout to [email protected]". No link to the docs, the app, the install command, a demo, pricing or the team.Source
    • Your websiteSite check (November 10, 2026): /robots.txt and /sitemap.xml return 404; no pricing page, blog, newsletter, careers page or social profile links; 0 Hacker News mentions and no news in the last 30 days. Change tracking shows the homepage headline moved from "Coding Agent Plan Mode with Super Powers" to "Scott: we help shops with estimations"; the domain was registered in October 2025.Source
    • Y CombinatorCompany page: "The planning layer for teams that build with coding agents"; founded 2025, Fall 2025 batch, team of 3, New York City, status Active, website tryscott.ai.Source
  • 22

    Scott is hiring a Product Engineer and a Design Engineer ($140K-$250K plus meaningful equity, new grads welcome) only through YC's job board. Those posts carry the clearest statement of the problem ("Coding agents amplify the cost of bad planning"), while tryscott.ai has no careers page or product description at all.

    Low impactLow effort First result in 1 month Founder/CEO

    First action

    Copy the thesis and the two roles onto /careers.

    How to measure it

    Applications from /careers

    Now
    –
    Target
    30% of applications
    By
    60 days

    Steps

    1. 1

      Write the page

      Thesis, product, roles.

      Founder/CEO 1 day

    2. 2

      Show how you work

      One planspace from the team's own work.

      Founder/CTO 1 day

    3. 3

      Design

      Same template as the homepage.

      Founder/CTO 1 day

    4. 4

      Link roles

      To the YC job posts.

      Founder/CEO 1 hour

    5. 5

      Footer link

      From the homepage and docs.

      Founder/CTO 1 hour

    6. 6

      Track

      Ask applicants where they found the role.

      Founder/CEO ongoing

    First result
    1 month
    Cost
    $0
    Owner
    Founder/CEO

    Smallest test

    Add the thesis and roles as a section on the homepage.

    Why it works

    Add /careers with the thesis, what the team is building (sync, planspaces, PR design review), the two roles linking to the YC posts, and how the team uses Scott to plan its own work. Link it from the homepage footer.

    Engineers who would join are also the users Scott wants; a careers page that explains the product serves both, and gives candidates the site they will check first.

    Risks and how to handle them

    • Page goes stale after hires

      Keep a general "we're always talking to engineers" line.

    Builds on

    Evidence

    • HiringTwo open roles on Y Combinator's job board: Product Engineer ($140K-$250K, 0.50%-2.00% equity) and Design Engineer ($140K-$250K, 0.50%-1.50%), New York or San Francisco, new grads welcome. The posts' thesis: "Coding agents amplify the cost of bad planning"; agents write far more code while teams still review and align on intent at the same pace.Source
    • Your websiteSite check (November 10, 2026): /robots.txt and /sitemap.xml return 404; no pricing page, blog, newsletter, careers page or social profile links; 0 Hacker News mentions and no news in the last 30 days. Change tracking shows the homepage headline moved from "Coding Agent Plan Mode with Super Powers" to "Scott: we help shops with estimations"; the domain was registered in October 2025.Source
More example playbooks and reviews

More growth playbooks

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

  • OVERTHESEA growth strategyA Kyiv fashion brand making handmade chiffon, satin and cotton dresses, tops and capes inspired by the sea, festivals and weddings, sold…22 growth ideas
  • Phylo growth strategyAn applied research lab that makes Biomni Lab, an agentic AI platform (the "Integrated Biology Environment") where biomedical scientists…21 growth ideas
  • Way of Wade growth strategyThe official direct-to-consumer store for Way of Wade, the Li-Ning and Dwyane Wade basketball brand: signature shoes (Way of Wade 12, Wade…22 growth ideas
  • Butternut Box growth strategyA B Corp fresh dog food subscription: gently cooked, portioned meals delivered frozen, plus treats and extras; sister brand Marro does…21 growth ideas
  • Weave growth strategyAn AI platform that measures engineering output, for human engineers and coding agents, and shows what AI tool spend is buying; plus an…23 growth ideas
  • Kiwii growth strategyA free daily scratch-card app: a new card unlocks every day at 20:00 and users win discounts, vouchers and deals from brands, redeemable…20 growth ideas
  • Warehouse-Lighting.com growth strategyAn online LED lighting distributor and manufacturer based in New Berlin, Wisconsin: industrial, commercial, outdoor, architectural and…22 growth ideas
  • Reshoevn8r growth strategySneaker cleaning and care products (kits with a patented laundry bag, solution, brushes, protection sprays and refills), a large YouTube…22 growth ideas

See every playbook and website review

Independent research. Programmable wrote this playbook from public sources, as they stood on November 10, 2026, and cites them throughout. Scott AI did not commission, review or endorse it, and Programmable is not affiliated with Scott AI. 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 Scott AI runs its business.

Something wrong, or do you represent Scott AI? 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

    Restore the homepage, the /apply link, crawlability and the install path; document what sync uploads; add a plain-text security page; start funnel tracking.

  2. Days 31 to 6012 ideas

    Pricing, comparison pages, sync-to-planspace activation, the planspace and PR-comment loops, the weekly pulse email, the Scott Method essay, disagreement posts, GitHub and plugin directories, an example planspace and a careers page.

  3. Days 61 to 904 ideas

    Commit to the big bets: the plan-drift report, an enterprise tier and the free disagreement tool; decide where shop estimations lives.

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 seeing where several agents disagree before any code is written. The most cited problem: agents write code faster than teams can align on intent.

  • A shareable plan the whole team can review and approve
  • Agents over-architect, pick the wrong abstractions or obscure libraries
  • In their words: “design intent”

Market

Competitive landscape

3 competitors compared: Traycer, Kiro and CodeRabbit.

  • Gap to exploit: Team sign-off on design before agents execute
  • Traycer: Execution workspace first; design approval and PR checks against a plan are not its focus
  • Kiro: Tied to its own environment rather than the agents a team already uses

Hiring signals

Scott is a three-person team hiring its first two engineers, both full-stack and open to new grads: a Product Engineer and a Design Engineer.

  • Product Engineer: First product hire on the planning layer; $140K-$250K, 0.50%-2.00% equity. Ideas 4, 5, 9, 12, 20.
  • Design Engineer: Design-minded engineer for the product and its surfaces. Ideas 1, 10, 17.

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 first (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)
  • 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