VALUE FIRST

The ladder

LevelQuestionSession
1. UnderstandWhat does AI change about value and trust?When the Metrics Go Up and the Team Goes Quiet
2. ApplyHow do you organize collaboration and governance?The 3 Collaboration Models · Your Agent Architecture Is Your Org Chart
3. RedesignWhat work do you delegate, what do you protect?The Work That Remains · Designing Agents for Your Value Stream

AI for leaders & AI for practitioners

Four sessions are ready as pre-conference workshops — two tracks, two formats each. For leaders:Leading When Agents Do the Work (full day) and The Work That Remains — extended (half day). For practitioners: Agentic Development, Hands-On (full day) and Designing Agents for Your Value Stream (full day, design-focused). Every conference workshop is also the demo for an in-house program.

Sessions

Keynote / talk · 45'

When the Metrics Go Up and the Team Goes Quiet

AI tools speed up delivery while quietly eroding candour, ownership, and trust — the conditions teams actually run on. Three erosion patterns already visible in real teams, and the values check that catches them before they become irreversible.

Participants can:

  • Name the tension between AI-driven speed and team trust in your own context.
  • Apply a values check to a running AI initiative.
  • Formulate three concrete agreements that protect psychological safety around AI use.

Agile2025, Denver (earlier version under the title "AI Harmony").

Talk · 45'

AI Development Governance: The 3 Collaboration Models

Three concrete models for human-AI collaboration in software delivery — who decides, who reviews, who ships — and what AI changes about team collaboration. Includes a live hands-on demo using the open-source BMAD method, showing AI agents organized in agile roles. Leaders leave knowing which model fits which risk profile.

Participants can:

  • Distinguish the three models and map them to risk profiles.
  • Determine, for their own delivery process, who decides, who reviews, and who ships.
  • Explain what an agent-structured workflow (à la BMAD) changes about roles.

Dutch AI Conference, Amsterdam. Audience feedback: the hands-on demos and the concepts on how AI changes team collaboration were rated most valuable.

Keynote / talk · 45'

Your Agent Architecture Is Your Org Chart

Conway's Law applied to AI agents: your agent design will mirror your organizational design, whether you plan it or not. Patterns from Lean, Agile, and consent-based decision making.

Participants can:

  • Explain how organizational design seeps into agent design (Conway's Law).
  • Apply two design patterns from adaptive ways of working to an agent architecture.
  • Recognize the risks of agent structures that copy existing silos.

Interactive talk · 90'

The Work That Remains

When AI does the building — what is left for the team? Participants build the Delegate / Redesign / Protect model themselves through structured exercises.

Participants can:

  • Apply the Delegate / Redesign / Protect distinction to their own workload.
  • Identify which team work deserves protection and why.
  • Formulate one redesign experiment for their team.

No preparation needed.

Full-day workshop · max 20 · no laptops

Leading When Agents Do the Work

One day through the full ladder: understand what AI changes about value and trust, apply the 3 Collaboration Models and the AI Strategy Canvas to your own organization, and make your first redesign decisions with Delegate / Redesign / Protect. For leadership teams and executives. No laptops needed — this is decision work, not tool work.

Participants can:

  • Map one of their own AI initiatives on the AI Strategy Canvas.
  • Choose a collaboration model per workflow, matched to risk profile.
  • Leave with three explicit redesign decisions, each with an owner.

For leadership teams; come with one real AI initiative in mind.

Full-day workshop · max 20 · laptops required

Agentic Development, Hands-On

From first agent to governed workflow in one day, using the open-source BMAD method. Set up agents in agile roles, run a real story through the workflow, hit the failure modes, and add the quality gates that catch them. For developers, architects, and tech leads. Bring a laptop; leave with a running setup and your governance choices documented.

Participants can:

  • Configure an agent workflow with roles and hand-offs.
  • Identify where the workflow fails without human review, and place quality gates there.
  • Take home a working configuration plus a decision log.

Laptop with terminal access required; basic development experience assumed.

Full-day workshop

Designing Agents for Your Value Stream

Cross-functional teams bring their own value stream as working material and leave with a concrete agent design canvas. Design workshop, no laptops — the hands-on coding counterpart is Agentic Development, Hands-On. Together they work as a two-day track.

Participants can:

  • Map their own value stream as the basis for agent design.
  • Deliver a completed agent design canvas for one concrete step.
  • Explicitly lock down governance choices (who checks what).

Come as a team if you can; bring knowledge of one real value stream.

All sessions are hands-on. My facilitation style uses Liberating Structures and Training from the Back of the Room — the room works, I don't lecture.

For your program page

Copy-ready descriptions for conference and event program pages.

Leading When Agents Do the Work
AI changes who does the work, who reviews it, and who is accountable. In this full-day workshop, leadership teams apply three practical models to their own organization and leave with explicit redesign decisions. Decision work, not tool work — no laptops needed.
Agentic Development, Hands-On
Build a governed agent workflow in one day with the open-source BMAD method. Configure agents in agile roles, find the failure modes, place the quality gates. For developers, architects, and tech leads. Laptops required.
Designing Agents for Your Value Stream
Bring your own value stream as working material and leave with a completed agent design canvas, including who checks what. A design workshop, not a coding lab. For cross-functional teams.
The Work That Remains — extended
When AI does the building, what is left for the team? Participants build the Delegate / Redesign / Protect model themselves through structured exercises, applied to their own workload. Half day, highly interactive.

The three models

The concepts behind the sessions above, in more depth — reference material you can use yourself, whether or not you book a session.

MODEL 01

The 3 Collaboration Models

Most teams adopting AI tools are making one of three choices about who's in control — without realizing they're making a choice at all. At one end: AI-assisted development, where a human reviews and decides on every step. At the other: fully agent-driven workflows, where the system acts and a human reviews after the fact. Each model carries a different risk profile, a different definition of "done," and a different governance requirement.

Getting the model wrong is how most AI adoption fails — not because the tools don't work, but because nobody decided who's accountable for what.

Apply this when your team is choosing — or has quietly inherited — an AI collaboration model for a specific workflow. Map the task to a risk profile first, then pick the model, not the other way around.

MODEL 02

Agent Architecture Is Your Org Chart

Conway's Law says systems mirror the communication structure of the organizations that build them. It was true for software architecture, and it's true for agent architecture: the way you split work between agents will mirror the way your organization already splits work between teams — whether you design it deliberately or not.

That's a problem when your organization has silos you don't want to repeat. An agent architecture copied straight from an org chart with hand-off delays and unclear ownership will reproduce exactly those delays and that unclear ownership, just faster.

Apply this when you're designing — or inheriting — a multi-agent system. Borrow patterns from Lean and consent-based decision making before you borrow your existing reporting lines, and check, deliberately, whether the agent boundaries you're drawing solve a real problem or just copy an old one.

MODEL 03

Delegate / Redesign / Protect

When agents take over the building, the obvious question is what's left for the team — and the wrong answer is "whatever's left over." Some work should be delegated outright: agents do it, humans check outcomes, not process. Some work needs redesigning: the task still needs a human, but the way it's done has to change now that agents handle the mechanical parts. And some work needs protecting: the parts of the job where the friction itself is the point — the debate in planning, the disagreement in a design review — because removing that friction removes what made the team good.

Apply this per task, not per role. Run it across a real backlog, task by task, and all three categories show up — usually in proportions that surprise the team doing the sorting.

Formats

FormatDurationMaxUse
Keynote45'UnlimitedMain program
Interactive talk / breakout90'~60Parallel session, with an exercise
Half-day workshop3.5 h (incl. breaks)24Pre-conference or in-house
Full-day workshop7 h (incl. breaks)20Pre-conference or in-house

Above 20 participants I work with a co-facilitator — quality over volume.

Room requirements

FormatRoom setupMaterialsOther
Keynote (45')Theater seating, stage, screen visible from every angleProjector + screen, wireless handheld or lapel mic, countdown timer visible to the speaker, speaker screen to see slides and slide notes, reliable Wi-FiLive survey — the audience uses their own phones
Interactive talk / breakout (90')Theater or cabaret seating, with room to turn into pairs or small groupsProjector + screen, 1 flipchart, markers, reliable Wi-FiA short paired exercise works without fixed tables
Half-day workshop (3.5 h, incl. breaks)Cabaret seating — round or rectangular work tables, max 4-6 participants per table. No theater seating.One flipchart per table (4-6), markers, sticky notes, projector + screen, wall space for flipchart sheetsNatural light and room to move between tables preferred; breaks are built into the 3.5 h
Full-day workshop (7 h, incl. breaks)Same as half-day, plus extra wall space for cumulative exercises over the daySame as half-day + extra flipchart sheets per session blockCatering breaks (lunch + 2× coffee) — logistics with the organizer; a separate corner for break conversations is welcome

Booking bundles

A

Keynote

45' + Q&A or an exercise.

B

Keynote + pre-conference workshop

The best value as a package — the most attractive for organizers (ticket revenue) and for double visibility.

C

Workshop standalone

Half day or full day.

D

Interactive talk

90' parallel session, with an exercise.

Fees on request — a fixed fee or revenue share on workshop tickets, depending on what works for your event.

The demo

Live demos use, among other things, the open-source BMAD method — AI agents in agile roles: PM, architect, developer, Scrum Master. It makes the "Agent Architecture = Org Chart" thesis literally visible: the audience sees an agent team organized like an agile team.

In-house

The same keynotes, workshops, and training programs, delivered on location. Flagship: AI leadership development — a multi-day training track for leadership teams along the same ladder (Understand → Apply → Redesign). Leading When Agents Do the Work is the single-day conference version of this same track. Every workshop above is also available as an in-house program.

Availability

Available for conferences worldwide. All sessions in English and Dutch.

Frequently asked questions

How quickly do you respond?
Within two business days, with a proposal tailored to your event.
How far in advance should we book?
Most conferences book 3 to 6 months ahead.
What does it cost?
Commercial conferences pay a speaking fee. Community and non-profit events are considered case by case.
Can sessions run virtually?
Yes, for both keynotes and workshops — in person is preferred, but virtual works.
Can the format be shorter or longer?
Keynotes scale from 20 to 75 minutes; workshops run from half a day to multiple days.
Do leadership audiences need a technical background?
No — leadership keynotes focus on governance and decision-making, not code.
Back to top ↑