Most AI engagements fail in the same room they were sold in. A vendor walks an executive team through a slick demo, a budget gets approved, a project manager is assigned, and six months later there is a working pilot that nobody uses. The model was fine. The integration was fine. What was missing was someone in the room with the authority to change how the work actually gets done.
We have a hard rule at Dissel: a founder, CEO, or owner-operator is in the room for every engagement. Not as a stakeholder. Not as a sponsor. As a builder. If we cannot get that person at the table, we walk away.
Why agentic work is different
A SaaS rollout asks a company to adopt a tool. An agentic engagement asks a company to rewire a workflow - who owns which decision, on what cadence, with what hand-offs. That is a redesign of the operating model, not a procurement decision. Middle managers cannot greenlight that. Heads of function rarely can either. The authority sits with the founder, the CEO, or whoever is allowed to say "we stop doing it that way."
Without that authority in the room, every decision becomes a memo. A spec gets drafted, sent up the chain, watered down, sent back, and by the time the agent ships it is wrapped in so many human approvals that the throughput gain is gone. The pilot "works" and the business does not move.
“If the only people in the room are the people who manage the old workflow, you will build a faster version of the old workflow.”
- Derk Disselhoff
What changes when the founder is at the table
Three things, immediately.
- 01Scope sharpens. A founder will cut a six-month roadmap to one workflow that matters by Friday. That is the engagement we want to run.
- 02Trade-offs get resolved in the room. "Should the agent send this without a human review?" stops being a six-week debate and becomes a yes or no in the same meeting.
- 03The operating model moves with the agent. Roles, hand-offs, and KPIs are redrawn as the agent is built - not bolted on six months later when nobody remembers why.
The output is a system, not a pilot. The team running the workflow tomorrow looks different from the team running it today, and everyone in the room signed up for that on day one.
What we ask before we start
Three questions, before we write a line of code:
- 01Who is allowed to say "we stop doing it that way"? That person joins the weekly working session.
- 02Which workflow do we rebuild first? One, named, measurable. Not a portfolio.
- 03What does the team look like on the other side? If nobody can answer, we are not ready to build.
The uncomfortable part
This filters out a lot of work. Plenty of organisations want the appearance of AI transformation without the disruption of redesigning anything. They want a pilot they can present at the next board meeting. We are not the right partner for that.
The engagements that compound - the ones where six months in the company is genuinely running differently - all share the same property. The founder was there. Not approving slides. Building.
“AI does not redesign your company. The founder in the room does. The agent is just what the redesign runs on.”
- Derk Disselhoff
The invitation
If you are the person who can change the operating model, we want to build with you. Bring one workflow that matters. We will bring the agent, the platform, and a team that knows how to ship into production. The rest we draw together, at the same table.
