Tom Gersic (@tomgersic) header
Tom Gersic (@tomgersic) profile picture

Founder and CEO https://teapawt.org/t.co/wavCl9mKHM

Chicago
Joined December 2010
Photo by Tom Gersic (@tomgersic)Photo by Tom Gersic (@tomgersic)Photo by Tom Gersic (@tomgersic)Photo by Tom Gersic (@tomgersic)Photo by Tom Gersic (@tomgersic)Photo by Tom Gersic (@tomgersic)Photo by Tom Gersic (@tomgersic)Photo by Tom Gersic (@tomgersic)Photo by Tom Gersic (@tomgersic)Photo by Tom Gersic (@tomgersic)
Faster research is the start. Verified, reproducible research is the point. youex.ai/blog/healthcare-sal…
1
39
Sellers don't ignore lead scores because they're stubborn. They ignore them because nobody can explain them.
5
AI adoption can look great on the leadership dashboard and very different at the desk. youex.ai/blog/healthcare-sal…
1
5
The measure of an AI insight: did it help someone make a better decision? youex.ai/blog/healthcare-sal…
5
"VP" in a title is a reason to look closer. It isn't a reason to skip the queue.
12
FAA SMART in one diagram: inputs feed shared data, compute predicts, simulates and optimizes, then three layers of humans validate, route and decide. The software recommends. People act. youex.ai/blog/inside-the-faa…
23
Tom Gersic retweeted
DOT says Friday's ground stop didn't involve SMART. It still makes the point: predictive software is only as good as the data and infrastructure under it. Our technical look at the FAA's new air traffic planning system: youex.ai/blog/inside-the-faa…
1
45
The FAA's SMART system pulls ~200 data streams to predict congestion and suggest reroutes. Humans keep the final call. Early FAA numbers: 180 flights rerouted, ~10 min and ~80 gal of fuel saved per flight. How it works, and where it can break: youex.ai/blog/inside-the-faa…
26
Efficiency is nice. Revenue that wouldn't have happened otherwise is the test.
41
AI can amplify a strong sales process. It can also accelerate confusion.
29
The most eager lead in your inbox might be a student writing a paper. Intent isn't fit.
13
Best AI metric I saw this week: the time from signal to decision.
25
The question isn't which model is best. It's which model handles each step, and whether the handoffs make the result better.
24
Behavior tests don't prove a model hasn't been altered. If it can use tools, treat the weights like code, then contain what it can do.
28
Six hours saved per seller is the start of the measurement, not the end. Where did the hours go?
40
Marketing: "Sales won't follow up." Sales: "Marketing sends junk." Both are right. The fix is evidence, not effort.
9
A sales agent doesn't need to write an essay to decide an account needs human review. Some agent decisions should stay small, scored and inspectable.
26
Best AI boundary I've seen this week: DePaul's agent prepares the financial-aid counselor. Fifteen minutes of prep becomes about one, and that time goes back to the student. Sales works the same way. AI does the research. You have the conversation.
12
Google's new benchmark grades finished files: formulas, factual accuracy, visual quality. That's the right direction. Evaluate the account plan, quote package, or service report someone will actually use. The chat response is an intermediate step.
16