Construction

Four jobs construction firms are actually giving AI

Vendor pricing once meant calling 5 to 10 suppliers. Annija Eizenarma names the construction workflows where AI is working now, and one that isn't yet.

Four workflows are where AI is earning its place in construction and adjacent legacy industries: cross-company communication with subcontractors, vendor price comparison, technical drawing and design automation, and — not yet, but coming — partnership intelligence. Annija Eizenarma, a founder, investor and adviser with seven years in legacy industries, describes the old baseline for pricing alone: "people used to call around 5, 10 vendors to just see what they can get."

Three of those four she describes as already working in practice. The fourth she frames explicitly as where she sees the future going, not as adoption she has observed.

What are construction firms actually using AI for?

Asked which tasks legacy businesses have used AI for in a way that makes sense, Eizenarma named four areas.

Cross-company communication. In construction, "what is important is the interconnectivity with all of the subcontractors and all of the construction companies that are working for them as well" — with the goal of uniting that communication in one place. Critically, she described the working model as AI "building and plugging in their existing platforms that they use," not replacing them.
Price and vendor comparison. "What else works is of course comparing prices and understanding where they can get a better deal." Her read is that opportunity remains, with one condition — "if you can be narrow enough to focus on specific industry."
Technical drawing and design automation. "They would do a lot of drawings and a lot of the things that are needed for the construction project... many aspects in that flow are also now being automated." She described prompt-to-plan generation as available today, including electrical wiring, "where the engineer only needs to make some of the adjustments according to their preferences."
Partnership and collaboration intelligence. Forward-looking rather than observed. Her framing is "how I see where the automation really can help" and "that's where I see the future going."

Her summary: "The list is endless of where the tech can really help these traditional businesses."

Should the software replace your existing systems or sit on top of them?

On top, in the pattern Eizenarma describes as working. The cross-company communication example is AI plugging into the platforms a firm already runs, and she identifies attempts to reinvent established workflows as the less successful approach.

She also argued that modern, license-free tools can compete directly with incumbent systems on cost, so the evidence is not one-directional. The consistent principle is about the work rather than the software: automate the process you already have before you agree to change it.

Is buying a competitor a faster way to modernise?

It is a strategy Eizenarma actively recommends, and the least obvious thing on her list of what works is not software at all.

"What often is the best for the company to implement any innovation in the business is to expand their presence through mergers and acquisition."

The mechanic runs two ways. An established operator can buy a more advanced competitor and inherit the modernisation:

"You can get a bigger market share, if you're an existing company in the legacy industry, by acquiring other companies that are maybe more tech advanced, are a little bit more further ahead in adopting software solutions that you might need, and learning the best practices from their teams and implementing it to your business as well."

Or a buyer can acquire a business and install the technology afterward. In a market where a vendor relationship takes years to build, buying the operator that already adopted the software is a legitimate shortcut, and it is being used.

What is coming that nobody sells yet?

Collaboration between independents, under consolidation pressure. Eizenarma's read on the direction of these industries is horizontal:

"How I see where the automation really can help is understanding how to establish strategic partnerships, how to collaborate with other independents, other construction companies ... to ensure bigger, better projects. You can do more together with partners than just sitting alone in your corner."

She was blunt that the pressure driving it is not optional: "we want it or don't, the acquisition and the consolidation is happening and it will happen even more and more every year," with national chains and private equity acquiring independent operators. Independents who can pool bids and partnerships can compete for work consolidated players would otherwise take. Artificial Reality, the AI buyer-evidence podcast hosted by Galina Fendikevich, has catalogued no product serving a coalition of independents acting together.

What should you ask a vendor before you buy?

Ask what they know about your day. The diagnostic Eizenarma's evidence supports is whether the vendor can describe your workflow before you explain it — and whether their product does one thing you can point at, rather than requiring you to change how you run.

Then ask for one delivered result before you commit to anything wider. Her advice on proving one thing well is aimed at builders; Artificial Reality's inference is that it inverts cleanly into a buyer-side test.

Pick the narrowest of the four jobs above that costs you real hours this month — for most firms it is still the vendor call-around — and buy for that one task only. If a vendor cannot show you a result on it inside a few weeks, the breadth of the rest of their pitch is not evidence in their favour.