£𝟮𝗺 𝘁𝗼 𝗴𝗲𝘁 𝘁𝗼 𝘀𝗲𝗲𝗱. 𝗧𝗵𝗮𝘁 𝘄𝗮𝘀 𝘁𝗵𝗲 𝗻𝘂𝗺𝗯𝗲𝗿.

It's now £200k.

That was the message from the AI-native startups at this week’s LettsTalk-Tech event. The economics have shifted so fast it's hard to keep up.

Months of work are being compressed into days. Teams past Series A are capping out at five people - not because founders are cutting corners, but because agents are doing the work that used to require twenty.

One startup was rebuilding the reinsurance industry from scratch. Another was dismantling the £50B+ Bitcoin lending market. A third was dismantling a $9 billion sales engagement industry with two founders and an agent.

None of them looked like startups from five years ago.

A few things stood out:

𝗘𝘅𝗲𝗰𝘂𝘁𝗶𝗼𝗻 𝘄𝗮𝘀 𝗮𝗹𝘄𝗮𝘆𝘀 𝘁𝗵𝗲 𝗽𝗿𝗼𝗯𝗹𝗲𝗺. The number one reason startups failed wasn't the idea. It wasn't the market. It was execution - fragmented tools, ad hoc processes, too much dependency on who the founder happened to know. AI is systematically removing that bottleneck.

𝗧𝗵𝗲 𝘀𝘂𝗽𝗽𝗼𝗿𝘁 𝘀𝘁𝗿𝘂𝗰𝘁𝘂𝗿𝗲 𝗶𝘀 𝗯𝗲𝗶𝗻𝗴 𝗿𝗲𝗯𝘂𝗶𝗹𝘁. The old model - advisors, mentors, accelerators - was slow, expensive, and inconsistent. The new model is step-by-step, AI-guided, available at 2am, and doesn't charge you a percentage of your round.

𝗧𝗵𝗲 𝘀𝗲𝗹𝗳-𝗱𝗿𝗶𝘃𝗶𝗻𝗴 𝗰𝗼𝗺𝗽𝗮𝗻𝘆 𝗶𝘀 𝗮𝗿𝗿𝗶𝘃𝗶𝗻𝗴. Agents that autonomously execute strategy, file for IP protection, run legal checks - not as a future concept, but running live in production today.

The metrics VCs have used for decades - runway, headcount, time-to-market - are being rewritten in real time.

What does this mean for established businesses?

The structural advantages incumbents relied on - capital, talent density, execution infrastructure - are being commoditised faster than most boards are meeting to discuss it.

Miss the event? Recording here.