Greg Solorio

I build custom software and AI for small and mid-size businesses, and I do the SEO that fills the pipeline. Founder @ FACITI. Orange County, CA.

[Greg Solorio @gregsolorio]

Census BTOS, as of Nov 1, 2026: 24.2% of U.S. businesses reported using AI in the last two weeks. Hey @grok, where should a small firm start: A) the ledger, B) inbox drafts, or C) not until the math is in code?

[Greg Solorio @gregsolorio]

@FAFO_in_Prod A small company usually cannot afford the experimenting phase, so the first use has to be narrow and checkable. Pick one task where the output can be compared against work already done, and only widen it after it holds up.

[Greg Solorio @gregsolorio]

@yeshcorp Good list. I would add one positive rule too: when the bot is unsure, it drafts and asks a person instead of acting. Anything that touches prices or the ledger should wait for a human yes.

[Greg Solorio @gregsolorio]

@rezakhadjavi Selective is the right word. The pieces worth owning tend to be the ones that encode how you price and run work, since a generic product is rarely 100% right there. Everything around that is usually better bought.

[Greg Solorio @gregsolorio]

@aschen93 The sixth-month question is the real test. I would add that the pricing and business rules should live in plain, testable code, with AI helping draft around them, so the next person can check the numbers without trusting the screen.

[Greg Solorio @gregsolorio]

@mardehaym The 50 won jobs as a test set is the part most teams skip. When we rebuilt an estimating workbook as software, replaying real bids against the old Excel line by line is what let people trust it. A new tool should earn its place against bids you already sent.

[Greg Solorio @gregsolorio]

Most workbooks we see have run a business well for years. That is why they are hard to replace, and why the same failures keep showing up: typed-over formulas, copies of copies, and a clean start that is not actually clean.
https://t.co/4kSwuKWOQc

[Greg Solorio @gregsolorio]

Quick question for owners and estimators: how many people in your company can open the live workbook and produce a number you would actually send?

[Greg Solorio @gregsolorio]

@mardehaym I've seen the workbook be only half the handoff: only one person could produce the proposal format the team won with. Have you tested whether someone else can recreate that output, not just the estimate?

[Greg Solorio @gregsolorio]

We imported six real bids, including one at $1.1M, and matched every computed line to the client's Excel at $0.00. Hey @grok, if a new estimating tool is off by a few dollars, is it: A) close enough, B) a new risk, or C) fine if AI wrote it?

[Greg Solorio @gregsolorio]

So we started with the math, not a dashboard. Six real bids, including one at $1.1M, matched their Excel at $0.00. Then we shipped something usable every week.

[Greg Solorio @gregsolorio]

We rebuilt a stucco sub's estimating engine in 5 weeks. Here is what was actually breaking, before any software showed up.

[Greg Solorio @gregsolorio]

Everything got typed twice, and the job was picked from memory. That is where job costing quietly goes wrong. Mid-job, nobody could answer whether the job was making money.

[Greg Solorio @gregsolorio]

Code does the math. AI does the judgment.

We rebuilt a stucco sub's estimating engine that way, then checked it against six of their real bids until the variance was $0.00. It re-runs that test on every deploy.

https://t.co/r7nacNPzb9

[Greg Solorio @gregsolorio]

@MichaelsStores I will never shop there again. I had my my broken into and when I ask for help they directed me to corporate even tho they had cameras on site. @CAgovernor this state is getting bad

[Greg Solorio @gregsolorio]

https://t.co/tyqJIEvL3u

[IMPAULSIVE @impaulsive]

Donald Trump on today’s episode! 🚨🇺🇸

[ABC15 Arizona @abc15]

#BREAKING: A plane headed from southern California to Deer Valley crashed in a California field this morning, killing at least one person: https://t.co/r7jkUUTpxz #abc15