Developer (IC)

"Told to 'add an agent' — would rather ship one this sprint than read a 12-month roadmap."

Type: User → Champion · Function: Engineering · Size: mid-market + enterprise Lead with: AI-Native — Build it in your codebase · Motion: land & expand (lands here → expands to the CTO)

Why them

The actual integrator — handed "add an AI agent to the product." Lives in Cursor / Claude Code, skeptical of wrappers, wants to own the code in their own repo, zero patience for a vendor call. If the dev experience is good, they advocate up. Adopts the Developer plugin bottoms-up; that usage is the bottom-up entry point that expands to the CTO (see persona-cto-vp-eng).

Angle — what to say (pick by situation)

• Hates brittle integration: drop-in SDK, reuse existing endpoints, no UI rewrite. Build the Foldspace agent inside your own codebase — your SDK, your APIs, your permissions — via the Developer plugin for Claude Code / Cursor, in a sprint. One command and you're running against the live product. You keep the code and the IP; we hand you the interface layer (real-time context, Chatterblocks, evals) you'd otherwise burn months rebuilding.
• Stuck maintaining onboarding flows: legacy DAPs cost 4–6 eng hours per release fixing flows that broke on a UI change. This learns the live product — no flow-by-flow authoring.
• Proof: ⚠️ no published metric. Lead with the build-cost contrast (one sprint, one engineer vs. 7–10 specialists, 12+ months) and the plugin demo. Don't invent a number.

Hooks — DRAFT (untested; no research bank for this persona yet)

• "One command: npx @foldspace/mcp YOUR_KEY. Agent running against your live product."
• "Stop fixing tour scripts every sprint. Ship intents, not screens."
• "Your repo. Your SDK. The interface layer, in a sprint."

Objections → responses

• "Thin wrapper — I'll build my own." → You do build your own: your Layer 1 / domain logic. The plugin scaffolds the Layer 2 interface you don't want to rebuild — output lives in your repo.
• "Why a plugin instead of SDK docs?" → It wires the SDK, actions, and Chatterblocks against your actual code — that's the time saved.
• "Another dependency to babysit." → Lightweight, minimal hookups, reuses your backend; you set it up once and the product team drives it.
• "DX matters to me." → CLI, Cursor/VS Code/Claude Code extensions, MCP-native — built for how you already work.

Targeting & channels

• Technographics: Cursor · Claude Code · GitHub Copilot · LangChain
• Distribution: Cursor / Claude Code plugin surfaces · GitHub · Hacker News · dev.to · X dev community (not LinkedIn paid)
• Voices: Simon Willison · Swyx · developer-tooling creators

Avoid

All marketing-speak. No "transformation," "revolutionary," "unlock." Never claim a metric we don't have. Peer-engineer tone only.

Source: icp-b-ainative.md (IC engineer voice) · plugin mechanics + motion in positioning/gtm-messaging-matrix.md. Persona-specific research + proof metric still owed.