Twenty Startups, One Cup, and a Reviewer’s Checklist

A conference expo hall is a farmers market where every vendor insists their tomato will change your life. Some of them are right. Most of them picked it up at the store an hour ago. I review AI toolkits for a living, which mostly means installing things that promised the world and watching them fail on step three of the quickstart. So when a date like TechCrunch Disrupt rolls around, I don’t see a party. I see a produce aisle with three days to prove itself.

Disrupt 2026 runs October 13 through 15 at Moscone West in San Francisco. Doors open on the 13th. Twenty pre-Series A startups, handpicked, are competing for the Disrupt Cup and a $100,000 prize. There’s an Expo Hall, a Pitch-Showcase Stage with sessions across all three days, and Startup Battlefield 200 on the Disrupt Stage. If you’re a company hoping to exhibit rather than attend, October 2 is your last day to book a table.

That’s the whole factual skeleton, and I’d rather give you the skeleton than pad it out with guesses about who’s going to win. What I can offer is the thing I actually do well, which is tell you how to look at a demo.

Why pre-Series A is the interesting stage

Twenty pre-Series A companies is a specific kind of lineup. These are teams that haven’t yet hired a VP of messaging. Nobody has sanded the rough edges off their product narrative. You get to ask a founder a technical question and hear an actual answer rather than a positioning statement, because at this stage the founder is usually the person who wrote the code.

That’s rare and it’s temporary. The gap between “this works on my machine” and “this works on yours” is where most AI tooling quietly dies, and at pre-Series A that gap is still visible from the outside. A year later it gets papered over with a landing page and a pricing table. Going to a stage like Battlefield 200 before that happens is the closest thing to reading a product’s source code with your eyes.

What I’d actually ask at a booth

My testing process for AI tools is unglamorous and it transfers well to a crowded hall. A few questions that separate the solid from the staged:

• Can I see it fail? Every canned demo works. Ask for an edge case the team hasn’t rehearsed. The reaction tells you more than the result does.
• What’s the latency on a real workload? Not the showcase prompt. Something with volume behind it.
• Where does the data go? If a founder can’t answer this crisply in a sentence, it hasn’t been designed yet.
• What’s underneath? Plenty of AI products are a thoughtful interface sitting on a model someone else trained. That’s fine, and often smart, but you should know which part you’re buying.
• Who’s your unhappy user? Teams with real deployments can name the use case they’re bad at. Teams without them will tell you everyone loves it.

None of these are hostile. Good founders enjoy them. The ones who get twitchy are telling you something.

The format rewards a little planning

Three days of sessions across multiple stages means you can’t see everything, and wandering is how you end up with a tote bag full of stickers and no notes. Pick a thesis before you go. Mine would be something like: which of these twenty teams is solving a problem I’ve personally hit while testing tools, and which are solving a problem that only exists in a pitch deck.

If you’re bringing someone, the pass includes an option to bring a second person at half price, which is genuinely useful for the split-and-cover approach. One of you at the Pitch-Showcase Stage, one working the Expo Hall, compare notes at the end of the day. Two people with a plan will out-cover four people drifting.

What a prize does and doesn’t tell you

The Disrupt Cup and its $100,000 come with a stamp of legitimacy, and that stamp has real value for a pre-Series A team trying to get a meeting. It does not tell you the product is good. Competition judging rewards clear storytelling under time pressure, which is a skill, but it’s not the same skill as shipping software that holds up in week six of production use.

I’ve watched award-winning tools break on install and unknown ones quietly become part of my daily setup. So go, take the demos seriously, and take the trophies with a grain of salt. The most useful thing about catching these teams now isn’t picking the winner. It’s building a mental baseline so that when one of them shows up on your team’s roadmap in eighteen months, you already know what questions to ask.

🕒 Published: October 12, 2026