The on-site facilities manager for an 1890s office building in an old mill town notices some small chunks of masonry on the sidewalk. Looking up, she sees that a lot of the mortar joints between the bricks have opened up. Alarmingly, part of the parapet appears to be bulging outward—towards the sidewalk.
Concerned about the potential fall hazards these conditions may pose, she pulls out her smartphone, taps the address and payment details into a website, and files the receipt for $599 as backup for the expense. The cost falls well within her routine operations budget so no further approvals or review are required.
Nearby, a drone is automatically dispatched and heads to the building. It’s already been programmed with a flight pattern based on satellite imagery of the building. When it arrives on site it flies a precise pattern and gathers a thousand or so high resolution images of the front facade before returning to its base and uploading them to the cloud.
An artificial intelligence (AI) model trained on hundreds of thousands of images like these immediately begins processing the data. After a short time it generates a report identifying conditions observed including deteriorated mortar, cracked masonry units, and displacement. The report also contains and elevation drawing with the location of all observations and accurate quantity take-offs.
Another model automatically takes this input and, knowing the building’s age, geographic location, weather history, and construction materials generates a statistical analysis of the probability of various outcomes and the most appropriate repair approaches given those parameters.
Next, an AI repair design model (trained on thousands of consultant reports, drawings, and specifications plus readily available resources like BIA Tech Notes and Secretary of the Interior Standards) creates drawings and specifications for recommended repairs.
Within a few hours of submitting the request online, our facilities manager receives an email with a link to download repair documents. Also included with the email are three proposals to perform the work (automatically generated by contractors based on AI models trained on years of estimates combined with current material and labor pricing data).
The facilities manager picks the most attractive proposal and schedules the work via the contractor’s website. Workers show up on the appointed day and the repairs commence.
How far away is that, really?
Does that scenario scare you? Or do you see it as brimming with opportunity?
A lot of the technology that would support what I describe above exists right now.
drones, obviously
all of the web/e-commerce infrastructure
Some is more speculative but I don’t think too fantastical given current tech and trends. None of it seems at all impossible… just a matter of time.
Yeah, but…
Can you poke some holes in that scenario? I’m sure.
What about permitting?! An Engineer or Architect will still need to be involved!
The drawings won’t have all of the details needed!
No exploratory openings?! How will underlying conditions be discovered?
etc…
The question really is this: do you want to hang your hat on those “but what about” objections? I don’t.
Do you remember when Blackberry was safe because no one wanted an on-screen virtual keyboard?
What about when no one would want a boring underpowered electric car?
Who needs a computer in the house anyway? That’s just for folks at MIT…
Results Over Process
The thing to keep in mind here is that what will matter will be results. AI and tech does not need to replicate what humans do. It’s not a matter of human actions, behaviors, and tasks being replaced.
It’s about results. And the more we think about how technology can help deliver results—rather than how it changes the process—the better off we’ll be at taking advantage of those tools rather than being disrupted by them.
Those that will win in this evolving landscape will be those that can leverage these new tools to deliver the same or better results faster or cheaper. This is not new; that’s been the name of the free market game forever.
Good Enough
The results also don't need to be perfect, only good enough.
The best smartphone cameras are not as good as similarly-priced mirrorless or DSLR cameras in many aspects of photography. But they've become good enough that everyone carries a smartphone and few carry a "real" camera.
It will be the same with AI tools. The current trend of poking holes ("Yeah, but it can't do [whatever] yet") will continue, but there will come a point when it's good enough that it won't matter.
news updates
That’s all for now! See you next week.