Blackedge Consulting is looking for a Project Architect.
Let’s start with something amusing. Blackedge Consulting is not an architectural firm, but a recruitment agency. That is, one group of people is looking for another person to work for a third company. A classic chain of intermediaries, where each link is convinced of the indispensability of the human factor. This makes the situation even more ironic.
Their client, an architectural firm in Dublin, has encountered a classic problem of growth and scaling. They have projects, and they are large (residential sector), and they need someone to take on the most labor-intensive and risky part of the job: transforming a concept into detailed technical documentation and overseeing its implementation on site. They are not looking for a creative genius, but a reliable, high-precision “processor” who will flawlessly generate drawings in Revit, ensure their compliance with Irish building regulations, and coordinate related departments. The pain lies in routine, errors due to the human factor, and the slow pace of documentation release.
Now, let’s imagine that instead of hiring yet another operator for a complex “machine” (Revit), we could upgrade the “machine” itself by embedding a brain into it. Instead of another employee with their weaknesses, vacations, and risk of burnout, the company could invest in creating a digital assistant that would handle 80% of an architect’s routine tasks.
What would this look like in practice? Let’s play out a short dialogue between myself, an old IT manager, and the hypothetical director of this Dublin firm.
— Listen, you’re looking for someone who will produce technical drawings without close supervision. That sounds like the description of an ideal algorithm. You spend months searching, then years on adaptation and retention, and they can still make mistakes.
— But AI won’t be able to understand the full complexity of the project! Who will make the decisions? Who will bear the responsibility?
— Let’s take it step by step. We’re not firing everyone; we’re giving them a super-tool.
Step 1: Creating an “AI Co-pilot” for Revit. We’re not reinventing the wheel; we’re using existing solutions or refining them. The market already has plugins like TestFit or Spacemaker from Autodesk, which handle generative design. We’ll go further. We’ll create a system that takes a basic 3D model as input and outputs a complete package of working documentation.
Step 2: Automation of detailing. An AI module, trained on thousands of the company’s successful projects, will automatically generate standard nodes, sections, and specifications for windows, doors, and materials. The human architect’s task shifts from “drawing” to “checking and approving.” They become not an executor, but a quality controller.
Step 3: Automated compliance audit. This is the most expensive part of the job, where the cost of error is maximal. We “feed” the neural network all Irish planning and building regulations. After this, the system real-time scans the Revit model and highlights any deviations: “Attention, the width of the escape corridor is 3 cm less than the norm, see paragraph 5.2.1b,” “Detected use of material with insufficient fire resistance class for this type of building.”
How to overcome distrust? Very simply. We start small. First, we implement the system as an “advisor.” It doesn’t make corrections but merely points out potential problems. Employees see that the system catches real errors they missed. Gradually, trust grows, and more authority can be delegated to it.
How to validate the AI’s work?
Initially – with good old human control. We take two senior architects. One works the old-fashioned way, the second – in tandem with the AI assistant. After a month, we compare the results: task completion speed, number of errors made (especially critical ones), and the cost of these errors. Spoiler: the machine, knowing no fatigue or bad mood, will win by a landslide in tasks requiring precision and rule adherence.
Next – cross-validation. The documentation package generated by the AI is given for review to a person who was not involved in the project. And vice versa. This will quickly show where the system is strong and where it still needs further training.
The final exam – the construction site. How clear and actionable is the AI’s documentation for contractors? Any request for clarification from builders is a ticket to the backlog for algorithm refinement. After 2-3 projects, the system will produce documentation cleaner and more accurate than any Mid-Senior level employee.
Ultimately, instead of searching the overheated job market for another “Revit operator” for a hefty sum, the company could invest in a system that would make the entire team 5 times more efficient. They are looking for a person to fill an operational gap. But they could invest in technology that would allow them to build more, faster, and with fewer errors. However, this requires shifting the paradigm from “hiring hands” to “creating intellectual assets.” And it seems not everyone is ready for this yet.
Источник: https://www.linkedin.com/jobs/view/4406452973/
