←
Tal Brown353 posts
Tal Brown avatar

Tal Brown

@talbrown

Zone7 CEO | Dad | 3rd Culture Kid

◷ Joined May 2009▧ 33 Photos and videos
Tal Brown avatar
Tal Brown@talbrown·26 May 2020

Wow. @Zone7ai covered by @BBCSport discussing how our injury risk insights help leagues and teams make informed decisions. Once again, @GetafeCF's lowest injury rates in #laliga shine bright. #teampride #football

BBC Sport@BBCSport26 May 2020

Based on Project Restart's provisional return date of 20 June, Manchester City players could face 13 games in 49 days. 👀 Premier League players are facing a "25% increased injury risk". Full story 👉 bbc.in/3c0u9rR #bbcfootball #EPL

Post attachment
Tal Brown avatar
Tal Brown@talbrown·3h

Net takeaway: the last mile of judgment and accountability is still human.

Tal Brown avatar
Tal Brown@talbrown·3h

Three or four years ago, earning expert trust was extremely hard it is a little bit easier now with Claude and modern LLMs). Still, the answer is consistency plus a strong user experience where the human expert can share feedback and tweak what actually goes out.

Tal Brown avatar
Tal Brown@talbrown·3h

From my years at Zone7 in pro sports, the biggest challenge was getting experts to feel psychologically safe enough to rely on AI recommendations. An expert in any field, tax included, is inherently skeptical of how much knowledge AI can provide.

Tal Brown avatar
Tal Brown@talbrown·3h

Getting those mechanics right is a technical job, not something a non-technical person can do easily.

Tal Brown avatar
Tal Brown@talbrown·3h

Here's what I'm seeing as we build: 1. The mechanics are still a hard technical endeavor Setting up autonomous workflows still requires a real technical background. Look at managed agents in Claude.

Tal Brown avatar
Tal Brown@talbrown·3h

But this idea of a completely self-sufficient agent running pieces of a business for you, making decisions and taking action on its own, the only place I've seen this is in linkedin posts.

Tal Brown avatar
Tal Brown@talbrown·3h

"I run my entire business on AI agents." I see this all the time in my feed, and it scares me because I feel like I'm not doing enough. I don't doubt that people are using AI, LLMs, and Claude every day. We use them heavily at Gambit Tax.

Tal Brown avatar
Tal Brown@talbrown·Nov 13

Is it easier now? A little. Three or four years ago this was extremely hard. It's a bit easier with Claude and LLMs. But the expert still owns the last step. Curious what others are seeing.

Tal Brown avatar
Tal Brown@talbrown·Nov 13

At Gambit we saw it in sales. AI-built, hyper-personalized tax advice doubled our close rate, literally doubled. But only when it was presented and provided by a human, and done fast, within 48 hours of a new lead coming in.

Tal Brown avatar
Tal Brown@talbrown·Nov 13

The second half is UX, and it's more than an edit button. The expert has to be able to share feedback, push back and tweak what's actually going out there, to the athlete or the client. Once they've shaped it, it's their call, not the machine's.

Tal Brown avatar
Tal Brown@talbrown·Nov 13

Consistency is the first half. One great recommendation doesn't do it. Trust builds when the output holds up week after week and the expert learns what to expect from it.

Tal Brown avatar
Tal Brown@talbrown·Nov 13

A black box does not breed confidence in an expert. If a sport scientist or a tax advisor can't see what went in and why a recommendation came out, they won't rely on it. And honestly they shouldn't.

Tal Brown avatar
Tal Brown@talbrown·Nov 13

Experts in any field, tax included, are inherently concerned and even skeptical about how much knowledge AI can provide. I don't think that's a problem to fix. It's the expert doing their job.