foundr.companyby Perea
← foundr.run·Beta

foundr.run — market insights

MECE market analysis. Numbers are point-in-time (May 2026) — sources linked so you can re-verify. TAM > SAM > SOM are nested slices, not aspirational forecasts.

TAMTotal addressable

~$9.1B / yr (2025), growing to ~$16.9B by 2030

Proxy

Global project management software market (issue trackers, work-OS, PPM). Adjacent: "AI in project management" $3.58B (2025) → $8.9B (2030); "agentic orchestration" $5.2B (2025) → $46.8B (2036).

Calc

PM software 2025 = $9.14B (TBRC) / $7.46B PPM (IDC) / $9.76B (Mordor). Midpoint ~$9B. Layer in agentic orchestration ($5B, 22% CAGR) for the agent-share.

Sources
SAMServiceable addressable

~$600M–$1.1B / yr (2026)

Proxy

AI-native devs + indie/solo founders running coding agents. Cursor's $1B ARR base (~1–2M paying devs) is the closest demographic proxy.

Calc

If 25–40% adopt an agent-shaped backlog at ARPU $15/mo workspace → 250–800k workspaces × $180/yr ≈ $45M–$144M direct; widened to all solo+small AI-native teams currently on Linear Free / GitHub Projects / Notion (~3–5M workspaces) at same ARPU = ~$540M–$900M.

Sources
SOMServiceable obtainable (3–5 yr)

$8M–$25M ARR by year 5

Proxy

Capturing 1–3% of the AI-native solo/indie segment via MCP-native + per-workspace pricing wedge.

Assumptions
  • 40k–100k paid workspaces
  • Solo $9 + Pro $29 blended ARPU ~$18/mo = $216/yr
  • Free tier drives viral acquisition; 3–5% conversion off a 1–3M free funnel
Analog precedent

Linear hit $8.4M ARR in 2023 (year 4) on 88 employees, $100M by year 6 — pure PLG, no sales. Plane reached "several hundred thousand $" within ~18 months on $799 lifetime.

Sources

The top 3 incumbents

Who controls the market — and why they can't pivot.

Each incumbent's vulnerabilities tagged by kind: technical, business model, regulatory / channel, cultural.

#1Linear

$100M ARR (Jun 2025), 25k+ orgs, 20k+ paid biz customers

  • Tech debt

    GraphQL-only API; no first-party MCP server. Community wrappers (linear-mcp) are GraphQL pass-throughs. Linear's Agent platform is bolted on top of seat-based identity.

  • Business model misalignment

    Pricing $10–$16 per user/mo with "Unlimited members" only on Free (capped 250 issues). 14k issues/week created by agents on their own platform — none of which are billable seats. Agent does the work, human pays the per-seat tax.

  • Cultural / incentive trap

    $100M ARR built on ~180% NRR from per-seat expansion within engineering orgs. Switching to workspace pricing collapses the upmarket motion that landed Coinbase / Oscar / Automattic.

#2Atlassian Jira

$5.2B FY25 revenue, 350k+ orgs, 2.3M AI MAU

  • Tech debt

    Documented automation caps (1,700 executions/mo Standard); JQL throttling on scheduled queries. Rovo + "Agents in Jira" GA Q3 FY26 — but governance bolted onto 20-year-old per-seat permissions.

  • Business model misalignment

    Agent automation runs growing 30% MoM, but priced per paid user. Indie founders / 1-person companies are functionally locked out — Jira's smallest unit of meaning is a "team."

  • Cultural / incentive trap

    Enterprise sales motion ($1M+ customer cohorts up 6× in 4yr) — cannot serve the $9/mo solo founder without cannibalizing upmarket.

#3GitHub Projects

Bundled into Copilot/Enterprise — closest AI-native default for solo devs

  • Tech debt

    Issue fields only hit public preview May 2026 — typed metadata is new. Projects v2 still requires GraphQL for non-trivial automation. Backlog UX is a spreadsheet over issues, not an intake queue.

  • Business model misalignment

    Free for public repos / bundled into Copilot — GitHub has no incentive to build an agent-shaped backlog as a product; ships as a feature when convenient.

  • Cultural / incentive trap

    Microsoft-owned, optimized for repo-scoped engineering work. Cross-project intake from a solo founder running 5 side-projects is a non-goal. Height tried and died Sep 2025.

Strategic moves (12 mo)

Ranked by leverage. Top of the list ships first.

Leverage is encoded in position — no fake score. #1 is the highest-leverage move we can make in the next quarter.

  1. 01

    Lead with "agent-native" as the category, not "yet another tracker"

    Q3

    Linear ships agent features behind beta gates ($16/user/mo Business); GitHub Projects v2 still has no API mutation to move a card between status columns. Own the wedge before Linear's Agent Platform matures out of beta.

  2. 02

    Make per-workspace pricing the loudest line on the homepage

    A 50-person Linear team pays $4,800/yr; a Jira service-account costs a paid seat. Lead every comparison: "Your agent fleet doesn't pay rent." Solo $9 / Pro $29 = $29/mo for one human + unlimited claude/codex sessions.

  3. 03

    Ship an "MCP-native" certification + receipts page

    Most "MCP servers" for PM are read-only wrappers around GitHub/Linear APIs inheriting their rate limits. foundr.run was designed AS an MCP server. Publish the spec + benchmarks against github-mcp-server and Composio Linear wrappers.

  4. 04

    Build the "agent-claim lease" into the marketing story

    Q1

    Race-conditions filed against claude-code (#22239), gh-aw (#26836), paperclip (PR #2643) all describe two agents both claiming the same task. Turn `backlog_claim` into a 90-second demo: "two Claude windows, one ticket, no duplicate work."

  5. 05

    Cut a free tier that solo founders cannot outgrow

    Linear's 250-issue free cap forces upgrade within weeks. Free: unlimited issues, 1 workspace, unlimited agents, quote-tweet attribution. Convert on the second project, not the 251st issue.

  6. 06

    Open-source the MCP server, monetize the workspace

    Backlog.md / sdvg.io / alice-mcp playbook. MCP server MIT, self-hostable; hosted multi-workspace + realtime collab + analytics as paid. Neutralizes "you'll die like Height" objection.

  7. 07

    Win the agent-fleet operator persona by name

    Pieter Levels, Justin Paul (V4D, 34 agents), Maciek Marchlewski (Agent0). Ship case studies + presets ("the Levels.io workspace") + a CLI that drops into Claude Code / Codex / Cursor in one line.

Economic moats

What we can hold — and what we can't.

Honest split. We refuse to call cost-leadership or distribution a moat unless it actually defends.

Real (defensible)

  1. 01

    MCP-native data model

    Linear and Jira retrofitted MCP onto existing REST/GraphQL APIs. We designed the schema and tool surface around `backlog_claim` / `backlog_complete` from day zero. Retrofit is a year of API redesign for them.

  2. 02

    Atomic claim-lease primitive in the DB

    Race conditions filed against claude-code, gh-aw, multica, paperclip all stem from missing CAS semantics. Our agent_backlog table enforces single-winner claims via Postgres row locks.

  3. 03

    Per-workspace billing infra

    Atlassian, Linear, Salesforce, Microsoft 365 all have ARR forecasting pinned to seat counts ($285B SaaS valuation wiped Feb 2026 specifically because per-seat ARR is mispriced for agents). Switching costs them more than us.

Not real (incumbents can match)

  1. 01

    The backlog data model itself

    Kanban + status + priority + assignee is a 30-year-old schema. Backlog.md, Alice, tpm-mcp, kanban-mcp all ship this in a weekend.

  2. 02

    "Fast keyboard-first UI"

    Linear already owns this perception. Copying with shadcn + the Linear-clone tutorial circuit is solved.

  3. 03

    AI-generated issue triage / PRD parsing

    Linear Triage Intelligence, Atlassian Rovo, GitHub's kunwarVivek/mcp-github-project-manager all ship this. Undifferentiated.

Switching costs in our favor

  • Agents written against `backlog_*` MCP tools won't port to Linear's `LINEAR_CREATE_LINEAR_ISSUE` without rewriting every prompt and tool call

Switching costs against us

  • A human teammate who shows up later expects Linear/Jira UI vocabulary (sprints, epics, JQL) — we have to teach our model

Power-user pain

5 unaddressed pains, real voices.

Each pain has ≥3 independent quotes from Reddit / HN / GitHub / X. If an incumbent could fix it, they would have already.

Pain A

Per-seat pricing penalizes agent fleets

  • “My agents don't have butts.”

    — artisancraft.dev

  • “Per-seat pricing for an AI agent is pricing based on the thing that least correlates with cost.”

    — Helenmireille, Medium

  • “$285 billion in SaaS valuation was wiped in 48 hours — the model itself was mispriced.”

    — The SaaS Library

Why incumbents
can't fix

Atlassian / Linear / Salesforce sales comp, ARR forecasts, and Wall Street multiples are anchored to seat count. Moving off it means re-pricing existing contracts downward.

Coverage

✅ Shipped foundr.run prices per workspace ($9 Solo / $29 Pro), unlimited agent identities included.

Pain B

Agents cannot move cards through workflow states

  • “Issue content can be updated programmatically, but the board workflow still has to be moved manually in the GitHub UI.”

    — Charliepsycho, github/github-mcp-server #2255

  • “The mutation correctly updates the underlying data store, but fails to update the board-view's grouping index.”

    — devactivity.com bug writeup

  • “There is no way to trigger an agentic workflow when an issue is added to a project.”

    — johnpreed, github/gh-aw #25336

Why incumbents
can't fix

GitHub Projects v2's GraphQL `updateProjectV2ItemFieldValue` has been broken for the MOVE path for months; the GitHub MCP server's `projects` toolset is read-mostly.

Coverage

✅ Shipped foundr-backlog status transitions are atomic Postgres updates via the same MCP tool that fetches them. No drag, no race.

Pain C

Two agents claim the same ticket and both do the work

  • “3-worker swarm, 6 tasks. Both alpha and beta saw Task #1 available, both claimed it, both executed it.”

    — odysseus0, anthropics/claude-code #22239

  • “Two sessions execute the same step. The losing session does not detect that it lost the claim.”

    — rileywhite, gastownhall/gascity #1052

  • “Both passed the guard. Two issues created when only one was permitted.”

    — github-actions[bot], github/gh-aw #26836

Why incumbents
can't fix

Linear's API has no claim primitive — agents simulate it via `assignee` updates (a TOCTOU window). Adding atomic claim semantics requires a schema change incumbents won't risk.

Coverage

✅ Shipped `backlog_claim` enforces single-winner via Postgres row lock + lease timeout.

Pain D

Automation execution quotas throttle real workloads

  • “We are on the standard plan at 1,700 execution a month and we are struggling to not exceed the monthly limit.”

    — Atlassian Community thread

  • “For my case it's a bullet in the foot. We have 8 users and my usage is about 15000 per month.”

    — same thread

  • “Atlassian cripples Jira automation for all but enterprise customers.”

    — HN #37595898, 100 points

Why incumbents
can't fix

Atlassian's automation cap is a deliberate upsell lever to Premium/Enterprise — it IS the pricing model, not a bug.

Coverage

⚠️ Partial foundr.run has no per-rule cap today, but needs a published "no execution quota on agent actions" guarantee on pricing.foundr.run.

Pain E

Service accounts / bots cost a full paid seat

  • “I'd prefer not to pay for a full user license for this. You're out of luck. Access to the REST API is via a licensed user account.”

    — Atlassian Community thread

  • “GitHub App bots cannot be added as collaborators to private Projects v2, blocking them from accessing project items.”

    — devactivity.com analysis

  • “Bot IDs are not accepted. Manage access UI does not list GitHub App bot users as inviteable collaborators.”

    — GitHub GraphQL error documentation

Why incumbents
can't fix

GitHub's `ProjectV2Actor` GraphQL union literally omits the `Bot` type. Atlassian Jira Software still consumes a license per assignment. Fixing requires schema changes.

Coverage

✅ Shipped Every agent identity in foundr.run is a free principal in the workspace; the workspace itself is the billable unit.

Synthesis

Where SAM × incumbent vulnerability × unaddressed pain converges.

A wedge counts only when all three columns align. Status = what we've actually shipped against it.

WedgeSAM segmentIncumbent vulnPain solvedStatus
MCP-as-the-API (not REST wrapper)Solo + small AI-native teams running Claude Code / Cursor / Codexgithub-mcp-server STILL cannot update Projects v2 status (Issue #2255 open)Agent can finish workflow without a human dragging a card✅ Shipped
Per-workspace, not per-seatSolo founders with 3–15 agent identitiesLinear at 50 seats = $4.8k/yr; Atlassian service accounts require paid license"My agents don't have butts" pricing✅ Shipped
Atomic claim-leaseMulti-agent fleetsclaude-code #22239, gh-aw #26836, paperclip PR #2643 all shipping fixes for duplicate executionNo two agents grab the same ticket✅ Shipped
Quote-tweet free tierFirst-time builders from X/HNLinear's 250-issue cap; Jira's 10-user capFrictionless 0→1 without credit card⚠️ PartialNeeds more virality loop
Self-hostable MCP open coreEngineers burned by Height shutdown (Sept 2025)Height deleted all data; 6mo migration windowVendor lock-in / shutdown risk❌ Gap
Agent-session lifecycle as first-class objectOperators running 5+ projectsLinear AgentSession is OAuth-app-only, Business tier; GitHub Projects has no projects_v2_item agentic triggerOne source of truth across agents and humans⚠️ Partial
Native dedup + claim webhook receiptsOperators paranoid about agent budget burnAtlassian automation throttles at 1,700 execs/mo on StandardPredictable cost per agent action, not per seat❌ Gap
Workspace-as-podcast (publish backlog to /changelog)Build-in-public foundersNone of Linear/Jira/GH ship a public "this is what shipped" surfaceMarketing + accountability in one place⚠️ PartialAdjacent — foundr-world has it

Capture strategy

Where foundr.run actually wins.

Each angle ties SOM capture to a specific incumbent vulnerability above.

See how we sell into that gap.

The market thesis lives here. The pricing, MCP surface, and feature list live on the features page.

Full features & pricing← Back to foundr.run