For boards. For founders. For AI agents.

Meet Rogue. The I'mBoard Board Ontology.

A typed, versioned registry of the metrics every board reports — defined, sourced, and benchmarked, so humans and AI agents are finally speaking the same language about semantic drift, dilution, ARR, and runway.

curl https://app.imboard.ai/api/ontology.json

The problem: semantic drift

Every founder has sat in a board meeting where twenty minutes were burned debating how ARR was calculated, not what it means. ARR, NRR, CAC payback, founder dilution — each has multiple common definitions, and the "right" one depends on the source you cite. Call this semantic drift: the gap between what a board thinks a metric means and what the underlying spreadsheet actually computed.

Humans tolerate that drift. We paper over it with footnotes, follow-ups, and "I'll get back to you on that one." AI agents don't. The moment you ask a CFO agent to compute runway or prepare a fundraising pack, semantic drift breaks the workflow — hallucinated formulas, broken API calls, or, worst of all, silently wrong numbers that look right.

Rogue is the contract that fixes both. One typed registry. One canonical definition per KPI. Every definition cited to an external standard. Every standard absorbed into the ontology so a human reader or an agent can read it from a single endpoint.

Why we call it Rogue

Named for the X-Men character whose mutant ability lets her absorb the powers, memories, and life force of whoever she touches. The ontology does the same thing with external standards: it touches the SaaS Metrics Standards Board, Carta's State of Private Markets, KBCM/Sapphire's SaaS survey, Mercer's turnover data, SaaS Capital's annual survey, Retently's NPS benchmarks, the NVCA Model Legal Documents, the YC Post-Money SAFE, and SHRM's talent-acquisition benchmarking — and absorbs their formulas, definitions, and accumulated wisdom into a single typed registry. One memory, sharable.

What's inside Rogue

Each entry in the registry is a typed contract — not just a definition, but the surrounding context an analyst or an agent needs to use it correctly.

~140 KPIs across 7 domains, 27 cited today

Fundraising, sales, customers, HR, operations, finance, and product — one registry, one shape. New citations land each release as standards catch up.

Every definition cites a source

SMSB, Carta, KBCM/Sapphire, Mercer, SaaS Capital, Retently, NVCA Model Documents, YC SAFE, or SHRM — each cited KPI traces back to an external standard. See the full source list on methodology.

Embedded interpretation, not just definitions

Every KPI carries a `whyItMatters` paragraph and `interpretationGuidance` with concrete benchmarks, common pitfalls, and escalation rules.

Benchmarks where mapped

p25 / median / p75 ranges drawn from public methodology sources, attributed at the KPI level.

Stage-aware

Same registry; different "core" set per stage. A pre-seed founder and a Series C CFO see the right KPIs without forking the schema.

Strict typing, versioned, machine-readable

Namespaced IDs (`fundraising.founder_dilution`), explicit `fieldType`, served from a single JSON endpoint. Agent-ready.

Anatomy of a KPI

One real entry from the published registry, end-to-end. This is the copy an agent — or your CFO — reads when founder dilution shows up in a board pack.

Founder Dilution

fundraising.founder_dilution
percentage (%) preSeed: core seed: core seriesA: core seriesB: core seriesC: core
Description
Percentage of founders' fully-diluted ownership that is given up in the new round, including any pre-close option-pool top-up (the "option pool shuffle" — option-pool expansion taken in the pre-money dilutes existing holders rather than new investors). Common pitfall: founders often quote the "investor dilution" (new money / post-money) and forget the option-pool top-up component. The Carta State of Private Markets quarterly reports publish stage-typical dilution ranges that boards should use as a sanity check.
Formula
founder_dilution_pct = (founder_shares_pre − founder_shares_post) / founder_shares_pre × 100
Includes both new-money dilution and any pre-close option-pool top-up borne in the pre-money. Per Carta State of Private Markets methodology.
Why it matters
Tracks founder skin-in-the-game over time — sustained ownership matters for long-term motivation and signaling to future investors. Boards balance dilution discipline against capital needs.
Interpretation guidance
Per Carta State of Private Markets benchmarks, typical per-round dilution for the priced round (excluding pool top-up) is 18–22% at seed, 18–22% at A, 12–18% at B, 10–15% at C+. Out-of-band dilution either signals weak negotiating position or a strategic priced-up next-round set-up.
Benchmark
p25 12% · median 18% · p75 24%
Related KPIs
fundraising.pre_money_valuation fundraising.post_money_valuation fundraising.total_round_size

Multiply that pattern across the registry. This is what an agent reads when you ask it to prep your board pack.

Where to use it

Who Rogue is for

For founders

Set up your board KPIs right the first time. Stop debating definitions in meetings — cite Rogue.

Browse the catalog

For AI and MCP builders

Agents need contracts, not vibes. Strict types, namespaced IDs, embedded interpretation guidance — one endpoint, zero ambiguity. Free signup, token in 30 seconds — see the MCP setup snippet below.

Connect via MCP

For researchers and journalists

Cite a versioned, sourced ontology. Every definition links back to its external standard.

Read the methodology

MCP setup — free signup, token in 30 seconds

Create a free I'mBoard account, grab a personal access token at Account → API Access, drop the snippet below into your MCP-compatible client. Read-only against the public catalog by default; no data about your company leaves your machine.

{
  "mcpServers": {
    "imboard": {
      "command": "npx",
      "args": ["-y", "@imboard.ai/mcp-server"],
      "env": {
        "IMBOARD_API_BASE_URL": "https://app.imboard.ai",
        "IMBOARD_API_TOKEN": "imb_pat_YOUR_TOKEN"
      }
    }
  }
}

Or in Claude Code:

claude mcp add imboard \
  -e IMBOARD_API_BASE_URL=https://app.imboard.ai \
  -e IMBOARD_API_TOKEN=imb_pat_YOUR_TOKEN \
  -- npx -y @imboard.ai/mcp-server

Full setup, tool reference, and the rest of the imboard MCP tools (most of which need the token to act on your board): docs.imboard.ai/docs/guides/claude-code-and-mcp

Frequently asked questions

What does "Rogue" mean?

Named for the X-Men character whose mutant ability lets her temporarily absorb the powers, memories, personality, and life force of anyone she touches. The ontology does the same with external standards — SMSB, Carta, KBCM/Sapphire, Mercer, SaaS Capital, Retently, NVCA Model Documents, YC SAFE, SHRM — absorbing their formulas and accumulated wisdom into a single typed registry.

How does Rogue fix semantic drift?

Semantic drift is the gap between what a board thinks a metric means and what the underlying spreadsheet actually computed. Rogue closes the gap by giving every KPI a single canonical definition, a cited source, an explicit formula, and embedded interpretation guidance — so humans and agents reading the same record reach the same conclusion.

Is Rogue open?

The spec is published and free to read. A formal open-source license and a community contribution flow are on the roadmap; until those land, contact us for usage rights beyond reading.

Do you charge for access?

No. Rogue is free to read at the published endpoint. The I'mBoard product uses Rogue internally; the registry itself is not metered.

How often does Rogue change?

Versioned. Definitions update when their cited source publishes new methodology (for example, Carta's quarterly State of Private Markets). The JSON endpoint reflects the current version.

What if I disagree with a definition?

File an issue or contact us. Industry-backed KPIs (tier: "published" in the JSON) anchor to an external standard, so disputes usually resolve by pointing at the cited source; if the source itself is contested, we surface the alternative in the interpretation guidance.

Who decides what is in Rogue?

I'mBoard maintains the registry. Industry-backed KPIs (tier: "published" in the JSON) require an external-standard anchor. I'mBoard-authored KPIs (tier: "editorial") cover board-critical metrics where no public standard exists yet; they're flagged in the JSON and re-anchored to a published source when one catches up. Both are versioned.

One ontology. Every board metric.

Stop debating definitions. Start citing a registry that humans and agents both read the same way.