Develop product strategy

Three coworkers, one strategy: product-management positioning, finance's realizable-value budget, legal's gated filings — co-produced for a real investor deck, every claim sourced.

This is the worked example of cross-domain coworker collaboration on a strategy artifact — the Workers Launchpad application deck at /deck — produced 2026-07-02 by the product-management, finance, legal, and marketing coworkers under one human's approval gates.

What each coworker contributed

Product-management (cwc-product-management, product-strategy skill):

• Positioning that keeps user and buyer distinct — the user is an AI agent consuming llms.txt, JSON APIs, and MCP tools; the buyer is the agent's operator, metered at Cloudflare's Monetization Gateway. Slides optimize the agent's cost-to-value; pricing meets the operator at the meter.
• One named north star — returning agent visits (change-feed re-polls, cite_as re-entries) — with everything else (25 MCP endpoints, 49-entry catalog) reported as input levers with measured movement, not goals.
• A now/next/later roadmap where every item carries its gate owner, and the deliberate absence of a TAM slide: a market-size number would have been fabricated, so curl-verifiable traction stands in its place.

Finance (cwc-finance, startup-finance + spend-tracking skills):

• BUDGET-5K.md — a $5,000 allocation where every line is spend_commit-gated, ordered by unlock-leverage (runway lock → entity → trademark → portfolio), with $0-on-purpose categories.
• The credit reality check: when a "$100,000 in Cloudflare credits" headline entered the plan, the operator challenged it as too good to be true — finance fetched the program terms the same day, confirmed the challenge (tiered program, 1-year window, Workers AI capped at list pricing, Registrar renewals not eligible), restated it at realizable value (~$60–600 at current scale), and appended a superseding ledger row rather than silently editing. The standing rule — vendor credits at realizable value, never face value — now lives in the budget file itself.

Legal (cwc-legal, startup-formation + compliance-playbook skills):

• The formation research (state choice, C-corp vs LLC trade-offs, the Stripe Atlas vehicle) and the attorney-review rule gating any filing; trademark clearance-before-filing sequencing; the rule that program terms get legal review before any application is submitted.

Why this belongs on this site

Each strategy fact lives as a durable row, not a chat memory: budget lines and corrections on subagentfinance.com, decision tasks on subagenttasks.com (session sess_2026-07-02_budget5k), the deck itself served at /deck from the same single source as its claude.ai artifact. The collaboration pattern is the platform's whole thesis — plugins as specialists, gates where money and law are involved, durable state over transcripts — applied to the company's own strategy rather than to a customer demo.