Arjun Pratap Singh Saharan

[Arjun Pratap Singh Saharan @arjunpss]

The most common reason estimators don't trust AI: they tried ChatGPT once, watched it hallucinate on a spec, and decided AI isn't ready.

That's a fair conclusion from a bad sample. The construction-trained version is a different product.

https://t.co/tHWtcT4NLo

[Arjun Pratap Singh Saharan @arjunpss]

Deepti on the doctrine that decides the next twenty-four months:
https://t.co/hXGJ9TT4cz

[Arjun Pratap Singh Saharan @arjunpss]

The category will be won by whoever holds three things at once: the data pipeline, the model, and the workflow surface. A wrapper has none of them.

[Arjun Pratap Singh Saharan @arjunpss]

In construction you do not get a second bid because your first one was close. You get a second bid because your first one was right and the GC relationship survived to the next project.

[Arjun Pratap Singh Saharan @arjunpss]

Wrapper companies cannot answer that. They demo on the clean sheet. They fall apart on the hard one.

[Arjun Pratap Singh Saharan @arjunpss]

Every preconstruction lead in a pilot this quarter has the same question for AI vendors: what happens on the hard sheet. The one with the regional symbol convention and the addendum that contradicts the schedule. 🧵

[Arjun Pratap Singh Saharan @arjunpss]

Deepti's full thesis on the architectural call that decides which construction AI companies are still here in 2028:
https://t.co/rtVZsjKfKo

[Arjun Pratap Singh Saharan @arjunpss]

A zoo of classifiers has no shared representation to anchor cross-trade relationships.

You can bolt heuristics on top. They fail on exactly the hard cases where the value is highest. The foundation model learns the topology in-distribution.

[Arjun Pratap Singh Saharan @arjunpss]

When the state of the art in vision encoders moves — and it has moved several times in the last 18 months — a zoo competitor runs 37 retraining cycles to absorb it.

A foundation model retrains the encoder once. The gap widens quarter by quarter.

[Arjun Pratap Singh Saharan @arjunpss]

Three things a 37-model construction AI vendor cannot do, no matter how much more data they collect:

1. Get cross-trade transfer
2. Amortize encoder improvements
3. Reason over physical connectivity

These compound. 🧵

[Arjun Pratap Singh Saharan @arjunpss]

Deepti on the three moats — technical, data, operational — and why each one is brittle alone:
https://t.co/7wMAQRH9BC

[Arjun Pratap Singh Saharan @arjunpss]

Construction data is radioactive by contract. Sealed bid sets carry owner confidentiality. Federal jobs carry CUI rules. Many sub agreements explicitly prohibit third-party cloud handling.

The compliance boundary decides where the software has to live.

[Arjun Pratap Singh Saharan @arjunpss]

The drawings that teach a model real work live in NDA archives at GCs, in sub records, in permit offices that rate-limit downloads, on local servers in preconstruction trailers.

None of it is in Common Crawl. None of it is on GitHub.

[Arjun Pratap Singh Saharan @arjunpss]

The AI discourse assumes horizontal labs always have more data because they have the internet.

In construction, the internet is mostly empty. 🧵

[Arjun Pratap Singh Saharan @arjunpss]

Deepti wrote up the evolution and where AI actually fits in the estimating workflow:
https://t.co/vX09vT2cGX

[Arjun Pratap Singh Saharan @arjunpss]

Two tools, two points in the workflow:

AI on extraction.
Bluebeam on review, markup, collaboration.
Estimator on the judgment calls in between.

The problem isn't the markup layer. It's the 8 hours of clicking before it.

[Arjun Pratap Singh Saharan @arjunpss]

Bluebeam is what 2M+ AEC professionals use to mark up, review, and collaborate on docs. When a GC sends a plan set, they expect Bluebeam markups. It's the shared language.

AI removes the manual counting that has to happen before any of that work starts.

[Arjun Pratap Singh Saharan @arjunpss]

Every VP of Preconstruction asks me the same question about AI takeoff tools: do I have to give up Bluebeam?

No. And any vendor pitching that is solving the wrong problem. 🧵

[Arjun Pratap Singh Saharan @arjunpss]

Foundation > zoo. The math doesn't care which feels easier to build.

Deepti on why this is the architectural choice that decides the next decade:
https://t.co/rtVZsjKfKo

[Arjun Pratap Singh Saharan @arjunpss]

Three questions to ask any AI takeoff vendor:

1. One model for the whole sheet, or 37 stitched together?
2. Does adding data for one trade lift the others?
3. When a new vision encoder ships, how many of your models retrain?