Investor deck · assembled by the engine · draft
AnyComp prices any job in any context from a model — not a survey match. It synthesizes a canonical universal job structure (the join key for the economy of work) and a multivariate pay model, so it can price the roles surveys can't, instantly, with the provenance to defend it.
Draft — 6 slides still carry a [bracketed]placeholder for Mike’s real figures (traction · raise · market size · valuation). Confidential — investor room.
say it so a partner can repeat it in five minutes
AnyComp tells you what any job pays — including the roles the surveys can't price — from a model, with the receipts. Public browse for everyone; “increase your pay” for individuals; “you can't do comp job-by-job” for enterprises.
the shift in the world that opens the window
AI made it possible to synthesize a canonical job structure across the incompatible source taxonomies and to model pay across every dimension cheaply — work that was previously manual and proprietary. Large LLM context windows make the modeled answer portable as abstracted insight lists.
the venture-scale opening
Compensation data and benchmarking is a large, locked market — billions flow to the proprietary survey houses (Mercer / Radford / WTW), whose data is coarse, licensed-by-the-seat, and frozen at publication. We expand from the comp wedge into the operating layer for all work data — the join key.
who buys first, and why now
[Beachhead — Mike's pick.] Two candidates on one engine: enterprise (“you can't do comp job-by-job — here's the modeled answer for every role”) and consumer (“increase your pay” / advancement). The public browse surface is the SEO magnet that feeds both. Lead with one; the others follow off the same substrate.
what we know that others have missed
The canonical universal job tree (function × level), synthesized to map every source without copying any (the join-key thesis), plus a multivariate pay model (geometric-mean blend → P10/P50/P90 across industry/size/stage/geo) calibrated against public data rather than copied from it. That prices the long tail of roles no survey covers — the exact roles modern orgs care about most.
traction / learning velocity for the stage
[Mike: the geo spine + reference catalogs are built; the pay model exists; public-browse MVP status; first users / pilots / willingness-to-pay — the strongest real evidence here.]
how revenue + acquisition compound
A product ladder (cheap cleaned-public data → premium modeled-canonical data) and a sell ladder (download → feed → matched → matched-feed → decision-assist). Posted prices, sold by the slice — never the whole combinatorial dataset. [Mike: unit economics.]
what compounds — why winning creates more winning
Own the data architecture for “job” — the join key everyone treats as a label, so it's undefended:
the unfair founder-market fit
[Mike: comp + four-discipline founder-market fit; the G2G benchmark lineage; filed patents.]
raise → milestones → risks retired → next inflection
Raise [$X] to [milestones — dataset coverage, the first paid tiers, the enterprise wedge], reaching [next inflection].
North Star
Not R² — allocation. Whether the shape of the customer's pay distribution matches the shape of their value/impact distribution: shrink the gap between intent and reality in dollars.