Capabilities

What I do, in concrete terms.

Six areas of expertise to go from idea to a system in production. Start from the problem you want to solve, or from the technical building block: both lead to the same place.

LLM & RAG

Your documents become a knowledge base you can query in plain language, with sourced answers.

The right answer, sourced and cross-checked, in seconds.

View details

for a sourced answer from your documents

10x

documents processed per day, no rekeying

made-up answers tolerated in production

What I build

Internal copilot & document RAG

Contracts, procedures, records: your documents become a copilot you query in plain language, with sourced answers. On easy questions, a direct answer; on hard ones, the system searches in several passes, cross-checks the documents and checks its answer before showing it. Your team stops searching, and each person only gets answers on what they're already allowed to see.

Internal copilot · Sourced answers · Iterative retrieval · Permissions respected

Document reading and extraction

Invoices, contracts, purchase orders, forms: your documents are read by models that understand the layout as much as the text. The right data comes out structured, ready to drop into your tool, even on PDFs that are scanned, stamped or badly framed.

Extraction · Vision · Invoices & contracts · Structuring

Model selection and integration

The right model for each task (Claude, GPT, open-source), cleanly plugged into your systems.

Claude · OpenAI · Open-source · MCP

The stack

Claude
OpenAI
RAG
pgvector
Qdrant
Embeddings
Reranking
Python
Vision
OCR
Agentic RAG

Agents & orchestration

Agents that carry out real tasks in your tools, connected via MCP, the standard that saves you from rewiring everything each time you switch models.

An agent that acts, not a chatbot that talks.

View details

24/7

the agent acts in your tools, day and night, no queue

once

one MCP connection written once, reused with Claude, GPT or a private model

on-site

the MCP connector can run in your own infra, right next to your systems

What I build

Your tools connected via MCP

MCP is the USB-C of your software: you connect your CRM, emails, databases and APIs once, and it then works with Claude, GPT or a private model. You switch models without redoing everything, and the connector can stay in-house.

MCP · No vendor lock-in · Sovereignty · Sovereign MCP server

Tool-using business agents

An agent that answers, decides and executes real actions in your tools through these connections, and escalates to a human when needed.

Tool actions · Human escalation · Support

Multi-agent orchestration

Several specialized agents that coordinate on complex tasks, with an orchestrator and guardrails.

Multi-agent · Coordination · Guardrails

The stack

MCP
LangGraph
Claude
Tool use
n8n
Python
TypeScript
Webhooks

Automation & workflows

Your repetitive tasks handed to reliable workflows, with AI where judgment is needed, and alerts when something goes wrong.

Repetitive work runs on its own, under control.

View details

manual rekeying on automated processes

7 days

for a first useful workflow in production

100%

of failures visible: alerts, logs, statuses

What I build

Business process automation

Email triage and dispatch, data entry, follow-ups, reporting: repetitive chains executed by workflows plugged into your tools.

n8n · Workflows · Back-office

AI-augmented workflows

AI inserted at the right step of the workflow to classify, extract, summarize or draft. The rest stays deterministic, and therefore predictable.

LLM · Classification · Extraction

Custom integrations

CRM, ERP, email, spreadsheets, internal tools: I connect them through APIs and webhooks, including when no off-the-shelf connector exists.

API · Webhooks · CRM · ERP

The stack

n8n
Make
Zapier
Python
TypeScript
API
Webhooks
Claude

Cloud & on-prem

Your AI systems deployed wherever your constraints require, down to your own infrastructure. The agent acts on your tools, and when the model and the MCP server run inside your perimeter, no data ever leaves.

Your data stays with you, even when the agent acts.

View details

on-prem

models and MCP server inside your infrastructure

sovereign hosting, outside the US Cloud Act

data leaves your perimeter, by design

What I build

Sovereign MCP server

Your CRM, emails, databases and tools connected to the agent through an MCP server that runs in-house or on an EU cloud. Paired with a private or EU-hosted model, the agent reads, decides and acts on site: neither your documents nor your queries go to a third party. I work out with you where the model needs to run so the promise actually holds.

MCP · On-prem · Sovereignty · EU cloud

On-prem & private models

Open-source models deployed inside your infrastructure: your data never leaves your perimeter, not to answer, not to act, not to train anything.

On-prem · Open-source · Private models · Security

EU deployment & industrialization

Scalable, monitored production deployment on an EU cloud or your own cloud (AWS, GCP, Azure, Vercel), with CI/CD, versioning and zero-downtime updates. I check where your models actually run, not just the logo on the invoice.

EU cloud · CI/CD · MLOps · Zero-downtime

The stack

Docker
AWS
GCP
Azure
Ollama
vLLM
Terraform
Vercel

Audit & optimization

I find where AI truly creates value, and cut the bill on what's already running.

AI at the right price, where it pays off.

View details

-30 to -70%

on your AI bill, at the same quality

Step 0

the state of your data, checked before we code

48h

for a first audit readout

What I build

AI audit & strategy

Your processes put under the microscope to find the use cases with real ROI, and honestly rule out the ones that aren't worth it.

Audit · Quantified use cases · ROI roadmap

Cost optimization

Model selection, caching, prompt architecture, batching: the LLM bill goes down without losing quality.

30 to 70% less · Caching · Batching · Model selection

Data readiness

Before building anything, I look at your data where it lives: is it complete, up to date, gathered, usable? Building AI on data that isn't ready is building on sand. I tell you what's blocking and where to start, without a big pointless overhaul.

Data readiness · Data quality · Feasibility

The stack

Token audit
Prompt engineering
Caching
Batching
Observability
Evaluation

Reliability & security in production

Evals, guardrails, observability and security: AI systems that don't drift, don't break silently, and are far harder to hijack.

AI that doesn't go off the rails, and far harder to hijack.

View details

red-teaming

I attack your AI before others do

injection

the #1 flaw in the OWASP LLM top 10, tested and contained

access

your data compartmentalized, not exposed by the RAG

What I build

Evaluations and guardrails

Test suites that measure the real quality of your answers on every version, plus filters and output validation that keep the agent in its lane, even on the unexpected.

Evals · Guardrails · Regression testing

Red-teaming and security audit

I test your AI like an attacker: prompt injection, attempted data leaks, an agent pushed outside its scope. I deliver a clear report of the flaws found, ranked by severity, and fix the most critical ones with you. Based on the OWASP top 10 for LLM applications.

Prompt injection · Red-teaming · OWASP LLM

Access control and permissions

The number one problem with an enterprise RAG is that it answers from documents the user wasn't allowed to see. I lock down access: each person only sees their own scope, and your agents and tool connections (MCP) hold only the permissions strictly needed.

Access control · RAG · MCP permissions

The stack

Evals
LangSmith
Traces
Guardrails
Observability
Logfire
Monitoring
OWASP LLM
Promptfoo
Red-teaming

Contact

Ready to go from demo to production?

Book a call

Reply within 24 hours · first conversation free, no strings attached.