Frequently asked questions

The questions people actually ask us.

Reliability, confidentiality, choosing the right tools, team autonomy, cost. Straight answers, drawn from our conversations with construction-industry leaders and teams.

AI, in practice

Does it actually work?

AI hallucinates, doesn't it? How do you avoid errors?

That's a fair question, and we don't brush it aside. AI can make things up — so we build every work package with guardrails: we require it to cite its sources, we scope the perimeter, and the human always keeps the final say. On a contractual document such as technical specifications (CCTP) or meeting minutes, nothing goes out without review. AI does the heavy lifting; you stay in control of the substance.

Is it magic, one click and done?

No, and that's actually reassuring. A deliverable usually stabilises over a few iterations: we test it on your real documents, adjust the instructions, and refine. That's exactly what we set up and hand over to your teams — a repeatable method, not a magic trick.

Will AI replace my teams?

Our logic is augmentation, not replacement. AI takes on the thankless, repetitive part; your teams refocus on expertise, judgement and client relationships. With junior profiles, we insist on critical thinking: the tool doesn't think for them, it saves them time.

My deliverables are highly standardised (brand guidelines, mandatory format). Can AI handle that?

Yes, provided you frame it. We start from your guidelines and templates, codify them into the work package, and steer the output (length, structure, style). We're upfront about it: getting a clean fit to your template stabilises over a few versions — and that's included in the handover.

Will AI write my bids and technical proposals for me?

No, and that's a deliberate choice. On a technical bid (mémoire technique) or a tender response, what wins for you is your signature, your references, your point of view — not generic text. AI does the upstream work: it breaks down the tender, surfaces the requirements from the technical specifications (CCTP), structures the outline and prepares a first draft based on your material. You keep the pen on the substance. The gain isn't « AI writes for you » — it's « you respond to more tenders, at equal quality, without spending your nights on it ».

Your data & your tools

And what about confidentiality?

Does my data leave the company?

That's a parameter of the framing, not an option. We start on a controlled perimeter, we sign an NDA, and we choose tools whose terms exclude reusing your data to train the models.

  • Controlled perimeter — we often start on a dedicated project, without plugging in your entire information system.
  • Tools with no reuse for training — we favour plans whose terms exclude using your data to train the models.
  • NDA — we sign a confidentiality agreement before touching a single document.

For GDPR / AI Act matters, we provide the contractual elements to your IT department: the enterprise plans of the main vendors (such as Claude Enterprise, OpenAI Enterprise) include a GDPR-compliant data processing agreement, a no-training clause on your data and hosting in an EU region.

Our IT department won't allow new tools to be installed on our workstations. How does that work?

That's a hurdle we anticipate. We build a framing note for IT with you (perimeter, enterprise licence, GDPR compliance) — enough to turn the blocker into a sponsor. Vendors' enterprise plans are designed precisely to clear IT filters: no training on your data, single sign-on (SSO), audit logs. We prepare the IT-department case for you.

Which tool do you use? ChatGPT, Claude, Copilot?

We're tool-agnostic. We choose the model and the tool that serve your use case and your environment (often Microsoft 365 in the construction industry), not the one that's convenient for us. And because we install a method rather than a licence, you're not locked in to a single player.

We already have Copilot — why add anything else?

Copilot stays relevant where it excels: Outlook, Teams, everyday office work. We're not asking you to unplug it. But on heavy production and construction-specific material — contract terms (CCAP) specific to a tender, project-specific documents, technical bids — the quality gap between an off-the-shelf use and a model properly framed around your trade is real. And above all: we don't add yet another licence, we install a method. You're not paying for an extra tool — you're learning to get ten times more out of what you already have.

I don't want to depend on a vendor that will hold me hostage on price.

That's exactly our stance. We don't sell you yet another SaaS. We install a skill inside your company, on your documents and your processes. If tomorrow you want to switch tools, the method stays — and you're the one who masters it.

AI consumes water and energy. Where do you stand?

We own it and we factor it in. Our « staircase » approach is the opposite of the pharaonic project: we install targeted, useful uses where the time saving is real — not AI everywhere for the sake of it. A controlled, lean climb, no waste.

Working with us

How does it work, in concrete terms?

I don't yet know what I expect from AI. Is that a problem?

No — it's even the most common starting point. That's the whole purpose of the diagnostic: we map your uses and your pain points, we prioritise, and we hand you a concrete roadmap — which use cases, in which order, for what gain. You leave with a plan, not with questions.

I've already tried AI on my own — what do you add on top?

That's actually a good sign: if you're already getting by on your own, it's proof that it works. The real issue is moving from individual use to the whole team — what you do intuitively, the second and third person on your team won't be able to reproduce on their own. Our job is to formalise that intuition, document it and install it for the whole team, on your real documents. You move from individual use that depends on you to a shared skill that holds without you.

Do you need to be « technical » to work with you?

Not at all. We talk construction before we talk AI. The handover is built for business teams, not for IT specialists — the goal is for your people to use the tool day to day without us.

How long before we see results?

We move in work packages, precisely so things go fast on a given perimeter. A first use case runs on a short cycle — we frame it, build it, hand it over — before stacking the next one. You don't pay for 18 months before seeing anything.

What guarantees continuity over time?

Two things. First, the method we install belongs to you: your teams keep it running autonomously, without depending on us day to day — that's the whole point of the handover. Second, Ali El Hariri stays the common thread who guarantees consistency, and draws on a network of partners already in production for the technical work packages. You're never left hanging on a single person. More about the collective.

How much does it cost?

We sell consulting and enablement: a diagnostic, a monthly subscription with no minimum term, and work packages bought one at a time. The amounts scale with your size and your perimeter. A first work package is priced in working days — not in 18-month projects. A diagnostic is half a day; the first work package that proves its worth is a week. We frame all this together in a first 30-minute call.

We're a large group / we're an SME. Who do you work with?

Mostly with construction-industry companies of 50 to 500 employees — general contractors, engineering firms, project management, property management — and with large groups, business unit by business unit. We tailor the entry point to your size: a technical work package that proves its worth, or a broad diagnostic.

We already have a provider or a consultant supporting us. Why you?

All the better — a good generalist provider saves you time on the digital side at large. Mister ConTech is different on one point: we only do AI, and we do it in the language of the construction industry — tenders, technical specifications (CCTP), technical bids, work packages. We often step in as a complement, on the specific business use case where a generalist stops. If your provider already covers that, we'll tell you.

What if it doesn't deliver results for us?

That's exactly why we move in short work packages, not in one big project. We start with a use case on your own documents, over a few days — you see the result before going any further. No 18-month tunnel: if a work package proves nothing, we don't stack it.

My teams are already swamped — they won't have time to get into it.

AI gives them time back, it doesn't take it. We install first where it relieves the most — reading large documents, first drafts — and the training happens on their real files, not in a disconnected classroom. The goal isn't to add yet another tool to learn, but to remove repetitive work.

A question that isn't here?

The simplest thing is to talk it through. 30 minutes, no commitment.

Book a call →