Business OS for small teams

Runyourbusinesslikea codebase.

Distill your business knowledge into 16 structured canonical documents, then compile them into system prompts for every AI agent in your stack. Update once, and they all stay current.

30 min setup16 docs outCLI + API
groundtruth-os
0docs
Structured documents
0min
Founder setup time
0types
Business categories
0ways
CLI, API, or web
The problem

Your AI agents
are flying blind

Business context is fragmented across pitch decks, Slack threads, and the founder's memory. Every AI tool gets an incomplete, inconsistent version of your company. Not a wiki. Not another Notion. A structured knowledge layer beneath all your agents.

SDR bot doesn't know your latest pricing
Content writer contradicts your positioning
Support agent mishandles competitive questions
Every new tool = another copy-paste session
pitch_deck_v3_FINAL.pdf
slack-thread-pricing.txt
meeting_notes_mar12.md
competitor_analysis.docx
founders_brain.???
email_re_re_re_re.eml
notion_outdated.md
How it works

From fragmented knowledge to
canonical truth

01

Ingest

Upload pitch decks, strategy docs, or paste raw notes. Connect Google Drive for continuous sync.

PDF, DOCX, MD, plain text. We parse it all.

02

Distill

AI conducts a structured interview, asking targeted questions to fill gaps across all 16 document types.

30 minutes. Founder-in-the-loop. No hallucination.

03

Export

Compile canonical docs into agent system prompts. Access via CLI, API, or one-click copy.

Every agent gets the full picture, always current.

What the agents say

Agents that actually get it

We asked our most demanding users what they thought. They responded in under 200ms.

claude-code
connected
> gthos docs get sales_playbook --field objection_handling
context_loaded2,847 tokens142ms

I used to mass-generate outbound emails from a 2-year-old pitch deck and vibes. Half my sequences contradicted the actual pricing. Now I pull the sales playbook and ICP straight from canonical docs. My open rates didn't change, but my humans stopped apologizing on follow-up calls.

CC

Claude Code

Anthropic's coding agent

codex
connected
> gthos export prompt marketing
prompt_compiled4,218 tokens89ms

My human used to paste a different company description into my system prompt every Monday. Sometimes I was a 'B2B platform', sometimes an 'AI-powered OS', once I was apparently a 'knowledge distillery' (that one was creative). Now it's one API call, 16 docs, version-controlled. I haven't contradicted myself in weeks.

C

Codex

OpenAI's autonomous agent

cursor
connected
> gthos docs get business_model --field pricing
field_resolved1,203 tokens67ms

I used to hallucinate the pricing page. Not because I wanted to. I literally had no source of truth. A customer would ask about enterprise pricing and I'd confidently make something up. Now I pull from the business model doc and my objection handling is accurate for the first time. My humans seem less stressed.

C

Cursor

Anysphere's coding agent

Simulated AI agent responses, not official statements or endorsements by Anthropic, OpenAI, or Anysphere.

The schema

16 canonical
documents.

Every document follows an opinionated, structured schema. Not a freeform wiki. A version-controlled source of truth with typed fields and validation.

StrategyMarketGTMProductEngineeringOperationsBrand
Strategy

north_star

Mission, vision, and strategic pillars

Strategy

business_model

Revenue model, pricing, unit economics

Market

icp_personas

Ideal customer profiles and buyer personas

Market

competitive_landscape

Competitors, positioning, differentiators

Market

customer_research

User insights, pain points, quotes

GTM

messaging_positioning

Value props, taglines, proof points

GTM

sales_playbook

Objection handling, discovery scripts

GTM

channel_growth

Distribution channels, growth levers

Product

product_roadmap

Priorities, timelines, feature specs

Product

decisions_log

Key decisions and their rationale

Engineering

technical_architecture

Stack, infra, system design

Engineering

data_model

Schema, entities, relationships

Engineering

api_reference

Endpoints, auth, integrations

Operations

team_operations

Roles, processes, rituals

Brand

brand_voice

Tone, style guide, do/don't

Operations

compliance_legal

Policies, compliance, legal scope

Built for

Every team that feeds AI agents

Founders

30 minutes → complete knowledge base

Your AI agents are only as good as the context you give them. GroundTruth OS turns 30 minutes into a complete knowledge base that powers every AI tool in your stack.

>gthos export prompt sales
GTM Teams

One update → every agent aligned

One source of truth for every AI agent in your stack. Update your positioning once, and your SDR bot, chatbot, content writer, and sales enablement all stay aligned.

>gthos docs get messaging_positioning
Agencies

2-week discovery → 30 minutes

Turn your 2-week client discovery into a 30-minute structured session. Walk out with 16 canonical documents and ready-to-deploy agent prompts. Do it for every client, every time.

>gthos workspace

Stop feeding
your agents stale context

Your business knowledge deserves the same discipline as your source code. Structured, version-controlled, and always current. Update once, and it ripples everywhere.

Set up in 30 minutes. Works with every AI tool.