Perplexity just crossed the Atlantic with its most ambitious marketing play yet. At London Tech Week 2026, CEO Aravind Srinivas announced the Billion Pound Build -- a UK edition of the company's Billion Dollar Build competition that ran in the US earlier this year. The pitch: use Perplexity Computer to build a company from scratch, and compete for a share of £1M in platform credits. The prize is credits-only (no cash), but the real story is what Perplexity is actually betting on.

The Setup: Two Phases, Twelve Weeks

The competition is structured around a 4-week pitch phase followed by an 8-week build phase, with up to £1M in Perplexity Computer Credits shared among up to 3 winners. Here is how the timeline breaks down:

1. Phase 1 -- Pitch (open now): Free to enter, open to all UK residents 18+. Pitch in any format -- written, document, video, Computer-built dashboard, website, or anything else. Pitches close 6 July 2026.
2. Phase 2 -- Build (invite-only): Up to 20 finalists are invited to Phase 2 on 20 July 2026. Each finalist receives a complimentary 8-week Perplexity Max subscription for the duration of the build. Ship product, get users, show traction.
3. Live Reveal: Finalists pitch a panel of industry leaders at the Live Reveal on 23 September 2026. Winners announced live, in-call, streaming live.

The judging panel may award one, two, three, or no prizes at its sole discretion. The prize is Computer Credits only -- no cash alternative. Credits expire 18 months after award.

What Perplexity Computer Actually Is

To understand why this competition matters, you need to understand what Perplexity is asking teams to build with. On February 25, 2026, Perplexity launched Computer, a multi-model AI agent that orchestrates 19 different AI models to complete complex, long-running tasks on your behalf. This is not a chatbot. Perplexity Computer is proactive and agentic -- you set a goal, walk away, and come back to a finished deliverable. It is less "chat assistant" and more "autonomous digital employee."

The system runs entirely in the cloud, breaking goals into subtasks and routing each to whichever model -- Claude, Gemini, GPT, Grok, or others -- was best suited for the job. It can even create subagents to handle specific parts of a workflow in parallel. Real use cases include reviewing documents, planning marketing campaigns, adjusting ad spend, generating tax filings, and booking travel. The system handles cognitive work that previously required a human at every step.