PowerPoint-ware.
You receive 80 slides about "digital transformation" and zero lines of code. The report gets filed and the operation carries on the same. Invoice paid, zero ROI.
Service · Diagnostics
Diagnosis before solution.
Something in your company costs more than it should and nobody can explain exactly why. We interview the people who run it, read the data you already have and map the handoffs. We come out with a prioritised roadmap: what to solve, in what order, with what return.
Expensive consultancies, reports that never get built.
Diffuse risk.
The big consultancies talk about AI without saying which model, who signs off, what happens when it gets it wrong. Accountability dissolves between layers of managers.
Internal talent burnt out.
Your team ends up doing the work the consultants promised. The AI that got deployed is a generic chatbot nobody uses after two months.
Diagnosis → pilot → scale-up.
Three measurable phases. Between each one there is a gate: if the pilot does not work, we do not scale. If the previous phase produces no data, we do not continue. No trusting expectations.
-
01
Diagnosis (2–3 wks)
Process mapping, available data, real candidates for AI automation. Output: a prioritised list with estimated ROI per intervention.
-
02
Pilot (4–6 wks)
We build one working use case in a bounded production setting. We measure against a human baseline. We decide with data whether to scale.
-
03
Scale-up (8–12 wks)
We extend the validated case to the rest of the process. Integration with internal systems, change plan and training.
-
04
Operation
Transfer to the internal team. Dashboards for usage, cost and accuracy. Optional evolutionary maintenance.
Product and judgement.
Map of opportunities
A prioritised analysis of where AI produces real ROI in your company, and where it makes no sense at all. Clarity before euphoria.
Working pilot
One use case implemented, measured and documented. Not slides: software running in your environment.
Scale-up roadmap
The next 6–12 months with objectives, dependencies and an estimated budget per phase.
Team training
Hands-on sessions with your technical and operational people. They are left genuinely able to maintain and extend the solution.
Legal documentation
A GDPR and AI Act compliance matrix. An auditable log of algorithmic decisions.
Impact metrics
A dashboard with hours saved, errors reduced, response time. Baseline and delta updated in real time.
- 3 Phases with a decision gate
- 4–6 wks From diagnosis to a working pilot
- GDPR And the AI Act, built in from the design
Our own products, not slides.
To decide with judgement.
-
How to prioritise AI integration in your company: a quantitative model
Where do you start implementing AI? Most people answer with a hunch. But deciding where AI goes is a quantifiable problem: it has variables, formulas and a rankable result. This article is that model, with a case worked through step by step.
-
The 3 levels of mastering Artificial Intelligence (and why most people never get past the first)
What level of Artificial Intelligence are you really at? The competitive advantage is no longer in having access to AI, but in knowing how to integrate it. These are the 3 levels that separate someone who uses AI from someone who masters it, and how to identify which one you are at today.
-
How to implement AI in a Spanish company in 2026: a step-by-step guide
Most Spanish companies know they have to integrate AI. Few know how to start without spending a year on consultancy. This guide gets concrete: what to assess, what to buy, what to build, and how to measure in 90 days.
What people ask us.
-
01 How are you different from Deloitte, KPMG or Accenture?
They sell diagnosis; we sell implementation. Their business unit is consultant hours; ours is delivered product. We invoice less because we are a smaller team, and we deliver sooner because AI agents run in parallel.
-
02 How is the cost of deploying AI calculated?
Three variables: the size of the organisation, the number of processes to intervene in and the level of operational criticality. A diagnosis is radically cheaper than a pilot, and a pilot than a scale-up. We close on an agreed scope, not by the hour. For a concrete number on your case: 30 min.
-
03 Which models do you use? OpenAI? Claude?
We choose per case. Claude (Anthropic) for complex reasoning and auditability, GPT (OpenAI) for generation and low cost, open-source models (Llama, Mistral) when the client needs self-hosting. The decision is technical, not commercial.
-
04 Who signs off if the AI gets it wrong?
Plantekia signs the contractual accountability for the delivered system. In production use, every high-impact decision requires recorded human approval. We do not automate critical decisions without supervision.
-
05 Is it compatible with the European AI Act?
Yes. We apply the AI Act risk framework from the diagnosis onwards: impact assessment, algorithmic transparency, auditable inference logs. Documentation ready for inspection if it comes.
-
06 Do you have experience in my sector?
We have our own product in production in local government, construction, education and vocational training, and public procurement. In any other sector we come in with a bounded diagnosis before committing to a pilot. We do not fake experience we do not have.
Let's start
Let's talk about your operation.
Tell us what costs more than it should. We come back with a concrete plan, realistic timelines and a clear yes or no.
contacto@plantekia.com