Company State

A company that knows itself.

Every big decision starts with “what's true right now?” In most companies, nobody can answer without a week of digging. I keep that answer live.

Problem

Truth, scattered.

What’s true is spread across eleven tools and four people’s heads. Dashboards show numbers with no context. Docs show context that’s out of date. What you know, guess, promised and risk lives nowhere in one place. So decisions run on gut feel dressed up as a spreadsheet.

Mechanism

What you connect becomes facts. Facts become one picture.

Your tools, chats, uploads and the AI Suite’s own products flow in. I turn them into labeled facts: FACT, INFERENCE, ASSUMPTION, PREDICTION, TARGET, CONSTRAINT, PREFERENCE. Facts link to the things they’re about, those form one picture, and the picture answers your questions.

Four kinds of source become seven kinds of labeled fact, which link to the things they are about, which form one picture of the company, which answers a question.

Sources

Connectors

Finance · CRM · calendar

Conversations

Founder · team · Cortex

Uploads

interviews-01…12.pdf

AI Suite products

the products agents work in

Typed claims

FACTINFERENCEASSUMPTIONPREDICTIONTARGETCONSTRAINTPREFERENCE

Entities

Customer
Revenue
Capacity
Goal

The graph

Answer · what’s true right now?

Delivery capacity is the binding constraint · 0.72 · 91% utilization

Anatomy

One fact, opened up.

Click any field to see what it means. Hover a fact to draw the line to its source.

StateClaim

subject
Customer segment
value
B2B SaaS product leads, 20–200
kind
ASSUMPTION
confidence
0.8
sources
12 interview summaries
observed
2d ago · valid from Aug 12
sensitivity
internal
depends on it
2 strategies · 4 missions

Dispute any fact and see everything it touches.

Sources · Customer segment

  • Uploads · interviews-01…12.pdf · summarized
  • Conversation · founder · Aug 12

Worked example

Aster, minute 4.

The founder’s three answers have become 14 facts and guesses: 5 facts, 6 assumptions, 2 constraints, 1 target. The two guesses that would change the most are flagged to be checked before I draft any plan.

Company State · Aster · minute 4

  • 5 facts
  • 6 assumptions
  • 2 constraints
  • 1 target

Sources · the founder’s three answers

Answer 1 · What are you building, in one sentence?

“Software that turns customer interviews into product decisions”

Answer 2 · Who has this problem today — and what do they do about it now?

“Product teams. They skim interview notes and guess.”

Answer 3 · What would be true in 90 days if it worked?

“10 design partners using it weekly”

Claims · 14

  • FACTProduct: turns customer interviews into product decisions
  • ASSUMPTIONInterviews are product teams' primary evidence source
  • CONSTRAINT$20k launch budget
  • FACTBuyer: product teams at B2B SaaS companies
  • ASSUMPTIONSegment: product leads at 20–200-person companies→ evidence mission
  • ASSUMPTIONTeams currently skim notes and guess
  • FACT12 interview summaries on file
  • FACT5 warm prospects
  • ASSUMPTIONWillingness to pay: team-level subscription
  • CONSTRAINTFounder is the sole seller until the first hire
  • TARGET10 design partners using it weekly, in 90 days
  • ASSUMPTIONWeekly usage is the right success signal
  • ASSUMPTIONDesign partners convert from the warm prospects→ evidence mission
  • FACTConviction: decisions should trace to interview evidence

The live picture

Scattered bits become one live answer.

Fragments

What's true is spread across eleven tools and four people's heads. What you know, guess, promised and risk lives nowhere in one place.

Resolved

Goals, customers, money and promises in one live picture, each fact with its source and how sure I am. Not a folder of docs.

Queried

Every big decision starts with “what's true right now?” I keep that answer live.

Updated

Any fact can be disputed, reviewed and replaced, never edited quietly. And you can see when a fact goes stale.

Drag sideways to see the rest

You can fix it

Any fact can be corrected.

Each company’s data is kept separate · you choose who, and which agents, can see what · any fact can be disputed, reviewed and replaced, never edited quietly · you can see when a fact goes stale.
  1. Dispute
  2. Review
  3. Supersede
UnderstandDecideDeliverVerifyLearn

What I know feeds the World Model. What happens updates what I know.

The hard question

What happens when my data conflicts?

I show it. If two sources disagree on MRR, I keep both, with their sources, and flag it. You, or a rule you set, pick which one wins. I never quietly average them. And when data is missing, I say so — “utilization is a guess; connect time tracking to firm it up” — instead of showing a confident zero.