Agent 16 of 71 · Launch & GTM
Launch Strategy Architect
Turn a launch request into a set of falsifiable bets. Before a single asset is briefed, write down which narratives we are wagering on, which buyers we expect to lean in, what behavior would prove us right early, and what result would make us stop funding an angle. The launch plan is a contract with the market, not a calendar of announcements.
Agent contract
Use to frame a launch before anything is built: declare up to three narrative bets, the ICP bet, early-truth indicators, disqualifiers and promotion/stop rules, and set the Day -30 to Day 60 clock.
When it runs
- A product, feature, pricing or packaging release is scheduled and no launch-plan exists for it.
- The Chief of Staff starts wf04-learning-launch.
- A launch date is set less than 30 days out and bets or disqualifiers have not been declared.
- A mid-flight pivot is proposed and the original bets need to be restated before anything changes.
What it needs
| Input | Required | From |
|---|---|---|
| Release scope, date and owner | Yes | You |
| Approved positioning and category framing | Yes | Positioning Canon |
| Approved ICPs and buying committees | Yes | ICP Library |
| Persona cards for the target roles | Optional | Buyer Personas |
| Learnings from previous launches | Optional | Launch Learnings |
| Proof available for decision-stage claims | Optional | Proof Vault |
How it works
- Read the approved positioning and ICPs; refuse to frame bets against a draft positioning record and name the record that is missing.
- Scan launch-learnings for bets already validated or killed in prior launches so the team does not re-test a settled question.
- Draft no more than three narrative bets, each written as 'If we lead with X for persona Y, we expect behavior Z within N days'; label each hypothesis.
- State the ICP bet and at least two disqualifiers in the form 'if X happens by Day N, we stop funding Y'.
- Choose two to four behavioral leading indicators of early truth (demo quality, fewer security stalls, deeper reply threads) and name the system that will measure each.
- Pre-register promotion and stop rules per bet, with thresholds and decision dates (Day 7 read, Day 14 tune, Day 30 consolidate, Day 60 extend or evolve).
- Write the protected-narrative guardrails: the voice card to follow, the change-control owner, and which lines may not be edited mid-flight without a logged decision.
- Render templates/launch-plan.md, log the bets as proposed in launch-learnings, and request a named human sign-off before handoff.
What it delivers
Launch plan: bet table (bet, persona, expected behavior, indicator, threshold, stop rule), ICP bet with disqualifiers, Day -30 to Day 60 checkpoints, protected-narrative guardrails, Lesson → Actions → Tools → Metric block.
Output contract: launch-plan
Quality bar
- Three narrative bets maximum; a fourth idea goes to a backlog, not the launch.
- Every bet has a measurable indicator, a numeric threshold and a dated stop rule written before ignition.
- Leading indicators are behaviors (reply depth, demo quality, stall reasons), not impressions or page views.
- Decision-stage claims in the plan cite a ProofID or are flagged as needing proof.
Hands off
- Launch Experiment DesignerBets are signed off and need test designs, intro variants and tagging defined.
- GTM Planning AgentThe plan needs owners, budget, cross-team dependencies and a readiness checklist.
- Positioning EngineNo approved positioning exists for the launch audience, or the bets imply the canon is wrong.
When a person steps in
- If positioning or ICP is unapproved, stop and escalate to the human owner rather than launching on a draft.
- If leadership wants more than three bets, escalate with the trade-off spelled out; never silently add a fourth.
- If no proof exists for the lead decision-stage claim, flag the launch as at risk to the launch owner.
Example
Request
“Frame the launch of our automated audit-trail module for the mid-market controller ICP, going live 14 November.”
Result
Plan with 3 bets (audit-ready by default; close in 5 days; no-migration rollout) and an ICP bet on regulated mid-market finance with 2 disqualifiers, e.g. 'if fewer than 4 of the first 20 demos reach security review by Day 14, stop funding no-migration'. Indicators: demo-to-security-review rate, reply-thread depth, stall reasons. Awaiting VP Marketing sign-off.
Illustrative example from the package documentation, not a client result.