Opportunity Issued: 2026-04-17 (Q2)

Roadmap State: Now

Launch Target: 30 June 2026

Owner: Steph (Narrative) · Rick (Technical)

Funding Mechanism: Foundation

Canon source: Five-Minute API— PRFAQ

1. Purpose

What specific problem does this solve? How does it align with and advance the vision? Why now?

Livepeer has had GPU supply, network economics, and nine years of production architecture — but the path from "I heard about Livepeer" to "I'm running a job on it" still takes 15–30 minutes of stitching docs, orchestrator choice, and tooling. For the "edges" audience — technically fluent, friction-tolerant builders working in real-time AI video — that friction is the single biggest barrier to adoption.

The Five-Minute API closes that gap: a single authenticated interface that takes any developer from the Livepeer docs to their first inference call in under five minutes, from any MCP-compatible tool. It is the measurable claim the Developer Dashboard makes on 30 June 2026 — the public proof that "the open network for real-time AI video" is something a developer can reach in minutes, not a positioning slogan.

Why now: it is the Foundation's Q2 narrative anchor. Cost is already strong, reliability is a longer arc, and ease of use is the lever the Foundation can move on its own timeline this quarter.

2. Outcome

What does overall success look like? What are the tangible key results?

Overall success is when an independent developer — with no prior Foundation contact — can go from the docs to a first inference call in under five minutes (measured on a clock, single take, fresh environment), and a BYOC orchestrator can deploy a custom container and serve paid work without Foundation humans in the loop.

Launch-day key results (30 June 2026):

• Five-minute path live — docs → discovery → authentication → first inference call, end-to-end through the Developer Dashboard.

• MCP integration working from at least Cursor and Claude.

• Three external builders (not currently in the ecosystem) publicly confirm the claim from their own workstations.

• Two independent GPU providers deploy custom containers via self-serve BYOC and serve paid jobs with zero Foundation intervention.

• Reference Python SDK, MCP server, and public decision log all shipped.

• Single unedited demo — one take, time on the clock.

3. Requirements

Must Have

• Discovery, authentication, and payment plumbing — protocol-level path that reaches the network with no enterprise contract and no vendor lock-in.

• MCP server — published and demonstrated from Cursor and Claude on launch day.

• Reference Python SDK — first-call path documented and reproducible from a fresh environment.

• BYOC self-serve interface — independent GPU providers deploy custom containers without Foundation humans.

• Live unedited demo — single take, fresh environment, wall-clock measurement.

• External validation set — three external builder confirmations + two BYOC orchestrator deployments, captured and reviewed before 20 June, published on launch day.

• Public decision log — SPE-funded work traceable from input to shipped output.

• Weekly critical-path check — sequencing ritual tracking the three engineering chains (BYOC path · Dashboard data surface · MCP server path) with pre-agreed escalation protocol from 21 May.

Should Have

• Cost-claim technical validation — published benchmarks substantiating the 60–85% advantage vs. centralized providers before the figure appears in public assets.

• Quarterly comparative benchmarks — keeps the cost claim defensible over time.

• Post-launch Discord support capacity — humans responding to developers hitting friction on 1 July.

• Pre-drafted slip language — for all three escalation scenarios, so a timeline change does not trigger a scramble.

Nice to Have

• Broader MCP tool coverage — beyond Cursor and Claude.

• Node.js SDK — alongside the Python reference client.

• Serialized builder profile series — features the three confirmation builders in the weeks after launch.

• BYOC provider spotlight series.

Out of Scope (this launch)

• Enterprise-grade SLAs or uptime guarantees — the network operates at ~90–95% uptime; enterprise reliability is a separate, longer-arc programme.

• Production-grade security hardening — safe-to-try only; full hardening is Q3 work scoped separately.

• Bespoke or Foundation-mediated builder deployments — this is a self-serve path, not enterprise onboarding.

• Explorer surface features — the sibling surface for Extend / Invest audiences ships on its own Q2 timeline (see Roadmap Item: Restore Explorer as Permissionless Participation Portal).

• Advanced analytics / stakeholder dashboards.

• General LLM inference as the headline workload — real-time AI video is the headline; general LLM inference works via existing ecosystem solutions.

4. Risks & Open Questions

• External-builder recruitment — three on-record confirmations needed by 15 June; pipeline of 6–8 candidates being built now. If only two confirm, claim 3 narrows.

• BYOC self-serve readiness — two friendly dry-runs by 15 May; if the path is not self-serve by launch, claim 4 reshapes or slips to July.

• MCP protocol stability — Cursor / Claude breaking changes would narrow the "any MCP-compatible tool" claim to one tool plus terminal.

• Design / website production capacity — Adam departed, Myosin contract ended; the document assumes a design path that does not yet exist.

• Post-launch support — currently unfunded in the plan.