Colleagues, today I have another gem from LinkedIn on my desk. The Spanish company APR Salud is looking for a technical architect to join its team. The vacancy is compelling, the tasks are serious. Let’s analyze it from our favorite angle: what if, instead of a human, we hired an algorithm?
**APR Salud is looking for an Arquitecto Técnico o Arquitecto Dpto. Proyectos.**
**Briefly about the company:** APR Salud are serious players involved in the installation and maintenance of high-tech medical equipment. We’re talking about giants like MRI, CT scanners, and other diagnostic machines. They prepare “homes” for them – special rooms in clinics that must comply with hundreds of requirements.
**What pain are they trying to solve:** The company is clearly on the rise. More clients, more projects. And every project is a headache. You need to take a hospital plan, a specification for, say, a new Siemens tomograph, consider Spanish building norms (normativa), radiation protection requirements, calculate floor loads, create a project in CAD/BIM, write a ton of accompanying documentation for the tender, and then also supervise the construction to ensure contractors don’t mess things up.
Currently, all this is handled by one (or several) people. They are the bottleneck. They get tired, can make mistakes, get sick. Their experience is unique, and if they leave, the company loses expertise. APR Salud is looking for another such “brain” to scale. A classic path. But perhaps an outdated one.
**How the task could be solved with AI:** Instead of searching for another specialist who will manually draw, calculate, and write, you could create a system that takes on 80% of the routine. The idea is not to fire everyone, but for one experienced engineer, with the help of an AI assistant, to be able to manage not two, but ten projects simultaneously, with fewer errors.
—
Let’s imagine a dialogue in APR Salud’s meeting room. Carlos, the head of the projects department, sits opposite me. He’s skeptical.
**Me:** Carlos, you’ve been looking for an architect for two months. Good specialists are in high demand. What if I told you that your ideal candidate isn’t human?
**Carlos:** (Smirks) Are you suggesting we hire a robot to run around the construction site with a tape measure? Our clients are hospitals; here, the cost of error is human life, not just shifted deadlines.
**Me:** Not a robot. A system. Let’s go step by step.
**Step 1: Creating “The Single Source of Truth.”**
Currently, knowledge is scattered: technical manuals for hundreds of equipment models are in PDFs, building codes are on government websites, and past project experience is in your engineers’ heads. We create a unified knowledge base. We upload everything: equipment specifications, building codes (Spanish, of course), data from your past 50 projects – drawings, budgets, problems encountered, and their solutions. This will be the brain of our system.
**Step 2: Generative Room Design.**
Your new architect receives a task: “We need to fit a GE Signa Pioneer 3.0T machine into a 5×6 meter room on the second floor of an old hospital wing.” What do they do? They open AutoCAD or Revit and start drawing, recalling similar projects.
But our system does this:
* It has already “read” the 300-page installation manual for this machine, including requirements for shielding (Faraday cage), ventilation, floor loads, and service access.
* It knows all local building and sanitary norms.
* You simply provide it with a 3D scan of the room and select the equipment model.
In 5 minutes, the AI offers not one, but 10 layout options, each with a detailed breakdown: here’s where the machine goes, here’s the technical room, here are the cable routes. For each option, a preliminary estimate, timeline, and potential risks are immediately calculated (e.g., “Attention, this option will require floor reinforcement, increasing the budget by 15%”). Your engineer is no longer a drafter; they are a strategist who chooses the best of the proposed options. For this, a combination of Autodesk Revit with plugins like Dynamo or specialized generative design platforms can be used.
**Step 3: Documentation Automation.**
The most tedious part of the job is writing technical reports (memorias técnicas) and specifications for tenders. Carlos, does your architect enjoy doing that?
**Carlos:** (Gloomily) Nobody enjoys it. It takes up a lot of time.
**Me:** Exactly. Our system, having chosen a layout option, generates a complete package of documents with one click, based on templates. All data – areas, materials, specifications – are automatically inserted. For this, we configure a large language model (LLM), fine-tuned on your technical documentation. It will write texts in your company’s style, without errors, and strictly to the point.
**Step 4: On-site Supervisor with Augmented Reality.**
I agree, AI cannot walk around the site by itself. But it can give superpowers to whoever is there. Your foreman or junior engineer puts on AR glasses (e.g., Microsoft HoloLens). And what do they see? Directly overlaid on the real construction site, they see the 3D model of the project. They can approach a wall and instantly see where cables should run, if the ventilation is correctly installed, and if the actual situation matches the plan. The system itself will highlight deviations in red. This is precisely the “compliance with specifications check” you mention in the job description.
**Reducing Distrust:** We are not proposing to fire everyone tomorrow and hand the keys to Skynet. We start with a pilot project. We take one real site and run it in parallel: the old way and with the help of an AI assistant. We compare speed, accuracy, and budget. Your engineers will see that AI is not a competitor, but a powerful tool that frees them from routine and allows them to focus on complex, creative tasks.
**How to Validate AI Results?**
Very simple. What we get at the output is not a ready-made order for execution, but options. The final decision is always up to a human. The role of your senior architect changes. They are no longer a “one-man band”; they are a **human validator**. Their task is to check the AI-proposed solutions, ask the right questions, and make the final, informed decision. They use their experience not for routine drafting, but for critical analysis. You’ll agree, this is a much more effective use of a costly specialist’s talent and time.
So, Carlos, perhaps you’re not just looking for another pair of hands and another head. Perhaps you’re looking for an opportunity to make every head on your team ten times more productive.
Источник: https://www.linkedin.com/jobs/view/4406149628/
