Nocmet Book a call

Multi-agent AI systems for companies · Talks

I design systems where AI does the work and humans decide.

For companies, I build teams of AI agents that deliver the greatest possible productivity gains, governed by deterministic checks and human approvals that no agent can bypass. And I talk about it on stage. My motto: “Make it work, then make it better”.

Portrait of Aleksander

Aleksander Multi-agent AI systems architect, speaker

Book a call See the value created LinkedIn

Diagram of a multi-agent system A request reaches an orchestrator, which delegates to specialized agents with isolated contexts. Their work goes through deterministic checks, then human approval: NO-GO, it goes back to the orchestrator; GO, the deliverable is produced. An idea, a request Orchestrator delegates, never produces spec design code review Deterministic checks scripts · hooks · tests Human approval GO / NO-GO NO-GO GO Traceable deliverable
AI agents, isolated contexts scripted checks human decision

What I do

Two crafts, one subject.

01 · Build

Multi-agent AI systems for your company

I design and deploy, remotely, teams of specialized AI agents, connected to your tools and processes, so that your teams gain as much productivity as possible while staying in control.

  • Identify the repetitive work worth handing over to agents.
  • Design the agents, their checks and the human approval points.
  • Go live, test on real cases, hand over to your teams.

Discuss your project

02 · Speak

Talks on AI, agents and work

I explain what AI agents really change in an organization, as seen from the workshop floor: what works, what breaks, and what must remain a human decision.

  • Concrete content, drawn from the systems I build.
  • No jargon: for executives, teams or a general audience.
  • On stage, in English or French.

See the topics

Before focusing on AI agents: years of software development, training and science communication.

How I work

From need to production, in a loop.

  1. Listening

    20-min video call
    transcript

  2. Scoping

    specification
    estimate

  3. Design

    orchestrator
    specialized agents

  4. Build

    code
    automated checks

  5. Approval

    human decision

  6. Production

    go-live

  7. Improvement

    feedback
    iterations

  • Humans decide
  • Code does the math, not the model
  • Everything is traceable
  • Your data stays with you
  • Proven on real cases
  • Your teams understand it

Cases · the value created

What it changes, in practice.

Clients and their projects remain confidential. What can be told: the work no longer done by hand, what is now traceable, and what remains a human decision.

A services company

From client meeting to quote, with no retyping

Before
After each meeting, the requirements document, the proposal and the quote were written by hand.
What was built
Agents that start from the meeting transcript and prepare all three documents; the amounts are calculated by scripts.
What it changes
Every line of the quote traces back to what the client actually said, and nothing goes out without the manager's approval.
A consulting firm

A sorted inbox, data that never leaves

Before
The inbox was sorted by hand: its contents were too sensitive for an online service.
What was built
A sorting assistant whose AI model runs entirely on the computer, with encrypted storage.
What it changes
No message is sent to a third party. Confidentiality comes from the design, not from a promise.
A young company

Regulatory answers checked at the source

Before
A general-purpose AI quotes legal texts from memory: impossible to rely on for legal obligations.
What was built
Agents specialized by field, whose every citation is checked by script against the official sources.
What it changes
Every reference can be verified, deadlines are calculated, and a legal professional signs off before any commitment.
A marketing team

One source, a whole campaign ready to review

Before
Every video or document was adapted by hand, channel by channel.
What was built
A tool that turns one source into angles, an outline, a schedule and copy ready to review.
What it changes
The team no longer starts from a blank page: it reviews and approves block by block. Nothing is published without it.
A social-sector organization

Regulatory paperwork in a single tool

Before
Case tracking and mandatory documents meant entering the same data again and again.
What was built
A business application that tracks each case and generates declarations and certificates.
What it changes
Data is entered only once, the documents follow from it, and access is partitioned by role.
A business unit

From idea to tool, without the long wait

Before
An idea for an internal tool would wait for a budget and a team to become available.
What was built
A workshop of agents that specifies, prototypes, builds and tests, with human sign-off at every step.
What it changes
A prototype to approve very early, then a tested, documented tool, with every decision on record.

Talks

AI in business, told by someone who puts it to work.

Topics I tailor to your audience and your format.

Topic

AI agents, in practice

What an agent is, what a team of agents can do today, and what it must never do on its own.

Topic

Putting AI to work in your company

Where to start, what to hand over to agents, what to keep in human hands.

Topic

Trust and control

Traceability, deterministic checks, human approval: how to trust what an AI produces.

Topic

Sensitive data and AI

Local AI, GDPR and the AI Act explained without jargon, and what they mean for system design.

Topic

From idea to tool

What changes when a team of agents specifies, builds and tests, and what that asks of the humans.

Pricing

Prices stated upfront.

Multi-agent systems: a four-step path

  1. To get acquainted

    Intro call

    20-minute video callFree

    You describe your need, and I tell you whether AI agents can meet it.

    Book a call

  2. Starting point

    Scoping and assessment

    Fixed price€4,500 excl. VATDelivered in 2 weeks

    • Use cases ranked by priority
    • Target architecture
    • Risks identified
    • Firm quote

    Amount deducted if the project goes ahead.

  3. Delivery

    Multi-agent system in production

    From€25,000 excl. VAT

    Designed, built, tested, approved by you, then put into production.

  4. Over time

    Monthly support

    From€3,500 excl. VATper month

    Enhancements, maintenance and monitoring of the system.

Consulting and expertise by the day: from €1,600 excl. VAT per day.

Talks

Keynote 45 to 60 minutes, tailor-made
in France €4,500 excl. VAToutside France €7,500 excl. VAT
Panel discussion
in France €1,800 excl. VAToutside France €3,000 excl. VAT

Travel expenses not included.

Prices exclude VAT. Running costs (AI models, hosting) are billed separately, at cost.

Where I work

Anywhere on Earth. For now.

I build your systems remotely, wherever you are, and I give my talks on stage, anywhere on the planet, in English and French. The Moon will follow as soon as a base opens there; Mars, once transport has improved.

  1. Earthtoday
  2. Moontomorrow
  3. Marsthe day after

Contact

A system to build, an audience to win over?

An AI agent project, or a talk to prepare: it all starts with a free 20-minute video call. You pick the slot, you describe your need or your event, and I tell you what is possible.

Book a call

Intro call · 20-min video call

The booking calendar opens in a new tab.

Choose a slot

Booking is handled by Google Calendar. Open in a new tab