Our Approach: Why a Good AI Consultant Should Make Themselves Unnecessary

Let us talk about something most consulting firms prefer to avoid: client dependency.

The traditional consulting business model relies on a simple asymmetry. The consultant knows, the client does not. The more the client depends on the consultant, the more the consultant bills. It works brilliantly — for the consultant. For the client, it is a different story.

At PIWA, we chose the opposite approach. Our goal is to make ourselves unnecessary. Here is why, and more importantly, how.

The problem: the consulting industry creates dependency

Look at the AI consulting market as it exists today. The majority of firms follow the same pattern:

1. They arrive with a proprietary diagnostic
2. They deploy solutions only they know how to maintain
3. They produce minimal (or incomprehensible) technical documentation
4. They bill for maintenance and upgrades
5. The client is locked in

This is not a conspiracy. It is a business model. According to Forrester research, 73% of companies that hired a consulting firm for digital transformation believe they could not maintain the deployed solutions without the vendor’s ongoing support.

73%. In other words, nearly 3 out of 4 companies are dependent on their consultant. That is a structural problem.

And in AI, the risk is even higher. Technologies evolve fast, models change, tools multiply. If your team does not understand what was implemented, you are not just dependent — you are vulnerable.

The PIWA philosophy: knowledge transfer as the primary deliverable

When we work with a client, we do not sell a solution. We sell a capability.

The difference is fundamental:

• Selling a solution = the client uses what we built
• Selling a capability = the client knows how to build and evolve things themselves

In practice, this means every PIWA engagement systematically includes:

1. Comprehensive, understandable documentation

Not a 200-page PDF that nobody will read. Living documentation, structured, written for the people who will actually maintain the system. With screenshots, examples, and step-by-step procedures.

We document:

• The technical architecture of each automation
• Technology choices and their rationale
• Maintenance procedures (updates, troubleshooting, adding new rules)
• Required access and contacts
• Known limitations and potential upgrades

2. Tailored team training

Documentation is not enough. Teams need to understand and be capable of taking over. Every implementation includes training sessions:

• Operational training: how to use the tool daily, for end users
• Technical training: how to maintain and evolve the automation, for internal technical leads
• Strategic training: how to identify new automation opportunities, for managers

We do not consider a project complete until the internal team is autonomous. That is our success criterion, not the number of billable days.

3. A planned disengagement objective

From the start of every engagement, we define a disengagement plan. When will the client be self-sufficient? What skills need to be transferred? What is the timeline?

For an ongoing support engagement, we target complete autonomy within 6 to 12 months. Not 6 to 12 years.

Why this approach is counterintuitive (but more profitable)

“If you make your clients autonomous, you lose recurring revenue.”

That is the classic objection. And it is wrong, for 3 reasons:

1. Word of mouth is the best acquisition channel. A satisfied, autonomous client talks about you. A dependent, frustrated client recommends nobody. Our most autonomous clients are our best ambassadors.

2. AI evolves constantly. A client who is autonomous on current automations will naturally return when new opportunities emerge. GPT-5 launches? They call us. A new use case appears? They call us. Not because they are dependent, but because they trust us.

3. Perceived value is much higher. A client who masters their system knows exactly how much value we created. They can measure the hours saved, the errors avoided, the processes accelerated. That is the best marketing.

Where does this conviction come from? The BPM legacy

I did not discover automation with ChatGPT. Before generative AI, I spent years in BPM (Business Process Management) — “old school” business process automation.

BPM taught me a lesson that many AI consultants have not yet understood: technology changes, processes remain. What matters is not the tool you use, but your organisation’s ability to think in terms of processes and identify what can be optimised.

A client who understands their processes and knows how to optimise them needs only an occasional push, not a permanent consultant. And that is exactly what we aim for.

What does it look like in practice?

Take a real example. An 80-person SMB client engaged us to automate sales proposal generation.

What a traditional firm would have done:

• Deploy a proprietary tool
• Charge a monthly licence fee
• Provide (paid) maintenance
• The client never touches the engine

What we did:

1. Process audit of existing workflows (2 days)
2. Selection of an open stack (n8n + Claude API + Google Docs)
3. Built the automation with complete documentation
4. Trained 2 internal leads (1 day)
5. 2-month support with progressive handover
6. Complete disengagement at month 3

Result: the client now generates sales proposals in 3 minutes instead of 45 minutes, and they can modify templates, adjust prompts, and add new document types. Without us.

Total project cost: EUR 12,000. Estimated annual saving: EUR 45,000. The client called us back for a new project 6 months later — by choice, not necessity.

5 warning signs your consultant is creating dependency

If you are already working with an AI provider, here are the signals that should raise a flag:

1. No documentation or documentation your team cannot understand
2. Proprietary tools that only the vendor can modify
3. No training for your internal teams
4. No exit plan: no conversation about when you will be self-sufficient
5. Opaque billing: you do not know what you are paying for or why

If 3 out of 5 of these signals are present, you are probably in a dependency relationship.

Our commitment

At PIWA, we commit to 4 principles:

• Full transparency: you see everything, understand everything, have access to everything
• Open technologies: no lock-in, no black boxes, no proprietary solutions
• Systematic knowledge transfer: every engagement includes documentation + training
• Explicit autonomy objective: we define together the date when you will no longer need us

That is our vision of consulting. Not the easiest path. But the only one that creates lasting value.

Ready to work with a firm that wants to make itself unnecessary?

Start with a 2-hour AI workshop to explore automation opportunities in your business. Or if you already know what you need, go straight to requesting an AI audit.

Discover our path to autonomy — let us take 30 minutes to discuss it.