# Entelech

> Entelech is an AI-native deploy agency based in the Benelux (Netherlands/Belgium). Senior forward deployed engineers join a client's team, find the work that is ready for AI, build the AI system inside the client's own stack and data, measure it weekly against a baseline, and hand over full ownership. If it doesn't beat today's numbers, it doesn't ship.

Tagline: "It's already in your business. We carve it out." The name comes from Aristotle's *entelechy*: the moment a potential becomes actual.

## What Entelech does
- Puts AI pilots and agent workflows into production, typically in about 8–12 weeks for a first narrow system.
- Builds inside the client's cloud, repository and tools; model-agnostic; client data is never used for model training.
- Client owns all code, prompts, evals, pipelines and documentation at the end.
- Typical use cases: claims and document intake, customer support resolution, finance operations and reconciliation, sales operations, internal knowledge search and copilots, evals/monitoring/LLMOps.

## How it works (five steps, each ends with evidence)
1. Find (weeks 1–2): map processes, data and owners; rank use cases by value and readiness. Deliverable: opportunity map and business case.
2. Design (weeks 2–3): architecture, data flows, guardrails, human checkpoints. Deliverable: system drawing and eval plan.
3. Build (weeks 3–8): built in the client's repo with the client's team; weekly demos on real data. Deliverable: working system in staging.
4. Measure (continuous): evals, shadow runs and cost tracking against the baseline. Deliverable: weekly eval ledger.
5. Deploy (weeks 8–10): live with named owners, monitoring and a runbook, then handover. Deliverable: production system and runbook.

## Engagements
- Scoping sprint — 2 weeks, fixed fee: process and data map, ranked use cases, system drawing, eval plan, go/no-go.
- Deploy — 8–12 weeks, fixed scope: two to three engineers inside the client team, one system from design to production, weekly demos and eval ledger, runbook, training and handover.
- Run and extend — monthly: monitoring, evals and model updates, new use cases, on-call, quarterly value review.
A good fit: a real process with volume, an owner who will run it, and access to the data.

## Clients
- Burgstate

## Partnerships
- Anthropic Partner Network
- OpenAI Partner Network
Entelech deploys Claude (Anthropic) and OpenAI models, as well as open and cloud-provider models where they fit better.

## Team
- Sebastian Ziemianski — forward deployed engineer
- Bram Berden — forward deployed engineer
- Melih Bardacki — forward deployed engineer

## FAQ
- What does "forward deployed" mean? Engineers work inside the client organisation, on its data and systems, and are measured on a system running in production, not on hours or slides.
- How fast can something be live? A narrow first system is typically live in eight to twelve weeks; the two-week scoping sprint confirms whether that is realistic.
- Where does our data go? Nowhere new: built in the client's cloud and accounts, providers configured so data is not used for training, decisions and model versions logged for risk and EU AI Act reviews.
- Who owns what at the end? The client owns everything: code, prompts, evals, pipelines, documentation.
- What if the numbers don't hold? Then it doesn't ship; every phase has a go/no-go against the baseline.

## Careers — The Bottega
A six-month apprenticeship for students and recent graduates (CS, AI, engineering, maths, physics) who want to put AI into production: Apprentice (learn) → Journeyman (shadow a client deployment) → Engineer (ship a system). Entelech also runs campus build nights, guest lectures, thesis projects and "coffee with an engineer". Details: https://entelech.ai/careers

## Contact
- Email: hello@entelech.ai
- WhatsApp: +31 6 25636095
- Website: https://entelech.ai
- Brand book: https://brand.entelech.ai