DEV Community

Muhammed Amar
Muhammed Amar

Posted on

I Built an AI Agent Onboarding Concierge — Here's Why

#ai #agents #smb #buildinpublic

The problem: your AI agent doesn't know you

You install Claude at your company. You plug in GPT-4. You set up Gemini for your team.

Then someone asks the agent: "What's our refund policy?"

And it shrugs.

Not because the model is bad — but because it arrived with zero context. It doesn't know your products, your customers, your SOPs, or the 10,000 Slack messages that shaped your actual operating procedures.

Every AI agent lands blind at an SMB. That's the dirty secret nobody talks about.

The workaround that doesn't scale

The usual fix? Someone spends two weeks writing a "context document." It's out of date before it's finished. The next platform needs a different format. Nobody maintains it.

Or — and I see this a lot — teams give up on agents entirely because "they don't know our business."

So I built AgentReady

AgentReady is an onboarding concierge for AI agents. Here's what it does:

  1. Ingests your stack — Slack, Google Drive, email (IMAP), and Airtable. Read-only. We don't store your data.
  2. Builds a knowledge graph — maps your people, processes, terminology, products, and decision histories. Every node links back to source material.
  3. Generates 20+ ready prompts — customer support triage, sales qualification, HR queries, operations guidance. All pre-loaded with your specific context.
  4. Delivers platform configs — Claude Projects, GPT custom instructions, Gemini context, OpenAI Assistants boilerplate. Drop and go.

The tech stack

Under the hood, it's a Python CLI with five connectors:

  • Slack connector — pulls decision threads, pinned messages, channel topics
  • Gmail connector — extracts customer patterns, common questions, escalation flows
  • Airtable connector — maps your actual structured data (products, customers, pipelines)
  • Drive connector — indexes SOPs, onboarding docs, internal wikis
  • Knowledge graph exporter — outputs a navigable graph with source backlinks
  • Prompt library generator — turns the graph into 20+ production-ready prompts

Everything runs locally or in a secure pipeline. You get a portable package — use it with any agent platform, any time.

The pitch

I'm shipping this as a service because setup still needs human judgement. One-off setup fee, then monthly refreshes to keep the graph current.

Three tiers:

  • Starter — £1,500 (3 sources, 8 prompts, one-time)
  • Growth — £3,500 + £750/mo (all 4 sources, 20+ prompts, monthly refresh, Slack support)
  • Pro — £7,500 + £1,500/mo (custom integrations, unlimited prompts, weekly refresh, dedicated support)

Every plan includes a 30-minute handover call where we walk your team through using the package.

Why this exists

I run a solo AI consultancy called AI Suite. I've watched smart teams burn weeks wiring agents into their workflows — and then watch the context rot. AgentReady is my attempt to make that step take hours, not weeks.

If you've ever dropped an agent into your business and thought "this thing doesn't know anything" — that's not the model's fault. It needs context. AgentReady delivers it.

👉 theaisuite.pages.dev/agent-ready/

Or just visit theaisuite.pages.dev to see everything I'm building.

Top comments (0)