Open Protocol · v0.2

Strategy breaks in the gap between the deck and the data.

Tell is an open protocol that turns strategic intent into structured, evidence-driven models that your teams and AI agents can actually read.

$ npm install -g @tell-protocol/cli

Install the CLI Read the Spec Get Started

Strategy deserves better than a slide deck.

Every layer of the modern stack has an open standard. Strategy has nothing.

Strategy lives in slide decks

Strategic intent is trapped in presentations, board packs, and the heads of senior leaders. It can't be queried, versioned, or read by any system.

Assumptions go unchecked

Every bet rests on assumptions that nobody tracks. Teams discover they were wrong after the money is spent and the opportunity has passed.

AI agents have no strategic context

Your AI agents can execute at machine speed, but they have no way of knowing whether what they're doing aligns with what the organisation actually cares about.

How Tell works

Three concepts. One living model.

Define Bets

Encode what your organisation is investing in as falsifiable hypotheses — not plans, but bets. Each carries a thesis, confidence score, and lifecycle status.

Track Assumptions

Map the conditions that must hold for each bet to work. Assumptions are shared across bets — so when one breaks, you see the cascade instantly.

Surface Evidence

Humans and AI agents contribute evidence signals that support or weaken assumptions. Append-only and immutable — a living record of what the data says.

Strategy as code.

A Tell model encodes your strategic portfolio in a single structured format. Manage it from the terminal or build integrations on top.

portfolio.tell.json

{
  "tell_version": "0.2",
  "portfolio": {
    "name": "Acme Corp — Strategic Portfolio",
    "organisation": "Acme Corp",
    "version": 7,
    "bets": [
      {
        "id": "bet_ai_support",
        "thesis": "AI-powered support replaces 30%
                   of tier-1 tickets within 6 months",
        "status": "active",
        "stage": "validating",
        "confidence": 68,
        "assumptions": [
          {
            "id": "asm_csat_parity",
            "text": "AI responses match human CSAT",
            "status": "holding"
          }
        ]
      }
    ]
  }
}

terminal — tell REPL

$ tell
  The Tell Protocol CLI — encode strategic intent

tell (acme-corp) > bet add "AI replaces 30% of tickets"
  ✓ Bet created
    ID     bet_ai_support
    Thesis AI replaces 30% of tickets

tell (acme-corp) > assume add bet_ai_support \
    "AI responses match human CSAT"
  ✓ Assumption added: asm_csat_parity

tell (acme-corp) > evidence add asm_csat_parity \
    --signal supports --confidence high \
    "CSAT 4.2/5 for AI vs 3.8 for human"
  ✓ Evidence recorded: ev_kxUMeNosT9Dx

tell (acme-corp) > status
  Acme Corp — v7
  Bets:        1 active
  Assumptions: 1 holding
  Evidence:    1 total records

Built for the people who shape strategy.

Whether you set the direction, build the systems, or deploy the agents — Tell gives you a shared language for strategic intent.

CEOs & CxOs

See your entire strategic portfolio in one living model — not scattered decks. Track confidence and evidence across every bet.

Heads of Strategy

Replace the board pack with evidence-driven, versionable strategy. Know which assumptions are holding and which are breaking.

CTOs & Engineering

Give your teams strategic context, not just tickets. Connect technical decisions back to the bets they serve.

AI Leaders

Ground your agents in what the organisation is actually betting on. An agent with Tell can do the right things, not just things.

Product Leaders

Connect product bets to evidence, not just intuition. Track whether the assumptions behind your roadmap are holding.

AI-native by design.

Tell operations are exposed as MCP tools. Any agent framework that supports MCP can read strategic context and write evidence — no custom integration required.

tell_read_portfolioRead the full strategic model

tell_write_evidenceSubmit a new evidence signal

tell_read_riskIdentify cross-bet risk concentrations

Install the MCP server: npm install -g @tell-protocol/mcp-server

AGENT → MCP → Tell

// Agent reads strategic context
tell_read_portfolio()

// Agent discovers a relevant signal
// and writes evidence back

tell_write_evidence({
assumption_id: "asm_csat_parity",
signal: "supports",
content: "CSAT 4.2/5 for AI responses"
})

Every layer has a protocol.
Now strategy does too.

Tell is an open standard — Apache-2.0 licensed, community-governed, free to implement. Build Tell-compatible tools, dashboards, and integrations.

The canonical Level 3 platform implementation is Apophenic.

Authentication

OAuth

APIs

OpenAPI

Agent Tools

MCP

Version Control

Git

Strategy

Tell

Level 1

Reader

Parse & display

Level 2

Writer

Append evidence

Level 3

Platform

Full intelligence

Start building with Tell.

Read the specification. Install the CLI. Or explore the getting started guide.

Install the CLI Read the Spec Getting Started Guide