The Offer

AI-Native Transformation

I install the agent infrastructure that turns a normal SME into an AI-native company — in 60 to 90 days. One engagement, one operator, one accountable thread.

Why this exists

Most SMEs are bolting AI onto a 2015 org chart

The chatbot demos look great. The pilots stall. Six months later you've spent six figures on consultants and you still have the same manual workflows, the same spreadsheets, and the same data scattered across twelve tools. The problem isn't the model — the problem is that your company has no place for an agent to plug in.

An AI-native company isn't one that uses ChatGPT. It's one whose data, tools, identity, and workflows are designed so software agents can do real work safely, observably, and at scale. That's an infrastructure problem, not a chatbot problem. I solve the infrastructure problem.

Phase 1 · The Build · 60–90 days

The AI-Native Build

A fixed-scope, fixed-price engagement that installs the full agent stack inside your business. By the end you own a production-grade platform on which your team — and I — can deploy new agents in days, not quarters.

What gets installed:

  • Agent runtime & orchestration — Claude Agent SDK / LangGraph / n8n, hosted in your cloud or mine, with versioned workflows and rollback.
  • Internal tool hub (MCP) — secure, governed connectors to your CRM, ERP, accounting, Drive, email, calendar, and databases. Agents call tools, not screens.
  • Private knowledge layer — a RAG index over your documents, contracts, SOPs, and tickets, with an evaluation suite so you can trust what it retrieves.
  • Identity, auth & cost controls — every agent runs under a real identity with permissions, rate limits, and per-task budgets. No surprise invoices, no rogue actions.
  • Observability & audit trail — every agent decision is logged, replayable, and auditable. Governable AI, not vibes.
  • 2–3 starter agents shipped — pick from the back office: an SDR / inbound triage agent, a finance reporting agent, an ops back-office agent, a support assistant. Production from day one.
  • Team enablement — your people learn to brief, supervise, and improve agents. The infra outlives the engagement.
Timeline: 60–90 days · Fixed fee: €25,000–€60,000 · Outcome: a platform you own, plus 2–3 agents already paying for themselves
Scope a Build
Phase 2 · AgentOps · Monthly retainer

AgentOps Retainer

Agents aren't software you ship and forget — they're systems that need operators. The retainer is how the platform stays sharp and how new agents keep getting deployed.

What's included each month:

  • New agents on demand — bring me a workflow, get a working agent in days. No new contract per use case.
  • Eval & guardrail tuning — when models drift, prompts age, or the business changes, the eval suite catches it and we fix it.
  • Cost & performance monitoring — token spend, latency, success rate, and ROI tracked per agent. You see what's working.
  • New tool connectors — every system you adopt gets wired into the MCP hub so every existing agent can use it.
  • On-call ops — when an agent breaks, escalates, or needs a human, I'm the human. SLA-backed.
  • Quarterly review — what shipped, what's next, what to retire. ROI on the table.
Monthly retainer: €2,000–€6,000 · Cancel anytime · Typical client: 5–15 agents in production within 12 months
Discuss AgentOps
Under the hood

The Stack

I'm vendor-pragmatic — the platform is built from production-grade open and commercial pieces, chosen for your data residency, scale, and existing tooling. Typical defaults:

  • Models: Claude (Anthropic) primary, GPT / Gemini / open-weight (Llama, Qwen) where they win on cost or latency.
  • Agent frameworks: Claude Agent SDK, LangGraph, n8n for low-code workflows.
  • Tooling protocol: MCP (Model Context Protocol) — the open standard for agent ↔ tool integration.
  • Vector / knowledge: pgvector, Qdrant, or your existing data warehouse.
  • Hosting: AWS, GCP, Azure, Hetzner, or on-prem. EU data residency by default.
  • Observability: Langfuse / OpenTelemetry, with custom dashboards for the business owner, not just the engineer.
Pricing & Guarantee

The Build is fixed-fee and fixed-scope — you know the number before you sign. The AgentOps retainer is value-pegged: I show you the ROI of every agent in production each quarter, and the retainer is roughly 20–30% of the value those agents generate. If the first agent shipped during the Build doesn't pay for itself within 12 months of operation, I refund the retainer until it does.

Start with a Free Audit
Not sure yet?

The Audit (free, 1 week)

Before you commit to a Build, we run a one-week audit. I map your top 10 manual workflows, score each on agent-readiness and ROI, and hand back a prioritized roadmap with hard numbers. If we're not a fit, you keep the roadmap. No hours billed, no obligation.

Book the Audit