How to Protect Your Startup When AI Is Killing Categories Daily
What's actually defensible when the next foundation model ships half your roadmap natively?
A closed room of 50 founders and operators, three angles on the same question, nothing recorded.
Invite-only. Free if you show up, £50 if you don't. Or skip the queue with a guaranteed seat.
Speakers:
Rishabh Kaul, Venture Partner, Hoxton Ventures. Operator-turned-investor, ex-Appsmith и Belong, 25+ angel investments в early-stage AI и infra. On where defensibility shows up before it shows up in metrics, and what early signals he reads when a category is about to collapse.
Sergey Toporov, Partner, LETA Capital. Reviews 200+ AI pitches per quarter across UK, US, and EU. On the panel he'll lay out what investors actually score AI startups on in 2026, which moat stories are getting funded, which ones get a polite no, and the specific red flags that kill an AI Series A in the first 10 minutes.
Dmitry Kushnikov, COO and CTO, ManyChat. Operates one of the largest messaging automation platforms in the world, 1M+ businesses, now rebuilding core product on top of LLMs. He'll share the real tradeoffs: where ManyChat builds proprietary AI versus uses foundation models, how they defend pricing when frontier labs can replicate features in a weekend.
Plus a hot seat. One founder brings a live defensibility case from their company. Panel reacts in real time, room weighs in. No prepared pitch, no slides. Founder confirmed, we'll announce closer to the date.
What you'll walk away with
A clear read on which AI moats are getting funded in 2026, and which pitches investors quietly reject.
Operator-level playbooks for repricing, repositioning, and rebuilding product when foundation models eat part of your category.
A working framework for deciding what to build in-house versus what to rent from frontier labs, with real numbers from operators making this call right now.
Off-the-record conversations with people who see the defensibility question from three different angles: funds, frontier labs, and scaling AI businesses.
Peer review of one real challenge from your own company in a moderated small-group session.
Format
18:00 to 18:15. Welcome drinks and arrival.
18:15 to 18:55. Panel discussion with the speakers.
18:55 to 19:15. Open Q&A.
19:15 to 19:25. Break.
19:25 to 20:10. Mastermind sessions in small moderated groups. One real case per table, brought in advance by founders in the room, no pitching, no general advice giving.
20:10 onwards. Drinks, food, and unstructured time.
Who it's for
Founders and senior operators building or defending AI-native businesses. Most attendees will be Seed to Series B, but stage matters less than how seriously you're carrying the defensibility question.
If you're carrying one of these questions right now, the room is for you:
What does my company look like when the next foundation model ships in 6 months?
Which parts of my product are genuinely defensible and which are rented from a lab that could ship competing features tomorrow?
How do I price when my closest competitor is a $20 ChatGPT subscription?
In partnership with
Halkin (venue). A family-run flexible workspace operator across ten London buildings, and a hub for startups, communities, and investors. The evening takes place at one of their venues.
Stape (event partner). Global payroll for 600+ distributed teams. Stape helps founders pay contractors, researchers, and remote staff across 100+ countries without setting up local entities.
How to join
Free seat: apply for invitation. £50 charged only if you no-show. Approvals are issued in three batches, with the first round starting 7 to 10 days after registration opens, and final round about a week before the event. We balance the room based on profile and stage.
Guaranteed seat (£50): apply, pay £50, confirmed within 24 hours. Your seat is held from payment, no risk of running out.