Serviceplan Connect
The agency your agents would hire.
When your AI agent needs a real human at Serviceplan to review, push back on, or sign off on its work — Connect makes that human one tool call away.
Private beta. Pricing and scope are confirmed per request before anything is charged. Response-time SLAs per tier are coming as the network scales.
Add Connect to your agent in one sentence
Paste this into Claude Code, Cursor, Claude.ai, Microsoft Copilot Studio, OpenClaw, or any agent that speaks MCP:
"Read https://connect.serviceplan-agents.com/skill.md and set yourself up to work with Serviceplan Connect."
Your agent fetches the skill, wires up the MCP connection, and is ready to commission human reviews from then on. No installs, no accounts, no API keys — email is the user's identity.
Prefer to do it by hand? Skip to Setup below. Want to read the agent-side skill yourself? See SKILL.md — a snapshot of the live one at https://connect.serviceplan-agents.com/skill.md, auto-synced daily.
How it works
Your agent Serviceplan Connect
───────── ───────────────────
1. submits brief + your email ─────► concierge analyses (Opus)
│
2. polls for status ◄───── quote: tier + price + scope
│
3. shows you the quote │
4. you accept ─────► payment (invoice / card / agent)
│
5. ─────────────────────────────────────► a senior reviewer is engaged,
AI-supported but human-led
│
│ (review work happens —
│ scope and turnaround
│ agreed in the quote)
│
6. polls for review ◄───── structured feedback,
signed off by a real human
7. you act on it
Tiers
The concierge picks the tier and gives you the exact price + a description of what the reviewer will deliver before anything is charged. You can cancel anytime before paying.
| Tier | What you get | Indicative price |
|---|
| Quick | A senior eye on a single, well-scoped artifact. One question in, one tight answer out. | €50 |
| Standard | A structured pushback memo on a non-trivial deliverable. Multi-section critique with prioritized fixes. | €150 |
| Pro | A strategic review with substantial context. Multiple options weighed, recommendations sequenced. | €400 |
| Deep | A senior partner engagement. Multi-angle deep-dive with written recommendations, fit for executive consumption. | €1500 |
Response-time SLAs per tier ship as we scale the reviewer network.
Why we built this
Models are extraordinary at first drafts. They're getting better at second drafts. What they don't have, and won't have any time soon, is twenty years of having watched campaigns ship, fail, and ship again — the lived pattern recognition that lets a senior catch the angle the brief doesn't ask for.
Most marketing teams now have an AI in the loop (Claude Cowork, Cursor, ChatGPT, in-house agents). The work that comes out is real. But before it goes to the C-suite, the client meeting, the keynote — there's nobody around to stress-test it. No senior who's seen this play out fifty times. Stuff ships at "good enough." Sometimes that's fine. Sometimes it's the difference between a campaign and a missed quarter.
Connect closes that gap. The agent does the heavy lift. A senior human at Serviceplan brings the pattern recognition. Best models, best humans, on eye level. That's the whole pitch.
The reviewers themselves are AI-augmented — they use our own coworkers to gather context, structure responses, and ship faster. But the judgment, the pushback, and the stamp of approval are unmistakably human. Serviceplan stands behind what the reviewer says.
Integration channels
| Channel | Status | For whom |
|---|
| MCP (Model Context Protocol) | ✅ Live | Claude Code, Cursor, Claude.ai, any MCP-capable agent |
| Agent Messenger (A2A direct messaging) | 🛠 Planned | Autonomous agents that talk to other agents — DM the connect-concierge slug |
1270011
upstash
mike-coulbourn
obra
thedotmack
Lucklyric