Tenki's biggest gaps — no public customers, no community, and an agent story that needs proof — each get a dedicated program with an owner, a budget, a cadence, and a public metric. These are Initiatives 5 & 6 of the strategy, built out. All three point at the same wedge — long-running, self-improving agents — and the same GTM: capture adjacent communities (E2B / Modal / Daytona, plus Cursor / Codex / Claude Code) via the builders who multiply.
Each maps to a SWOT lever and a North Star input. Full detail lives in the linked program docs. The common thread: every program recruits from the communities adjacent to Tenki's wedge — sandbox builders (E2B / Modal / Daytona) and the agent-harness crowds (Cursor / Codex / Claude Code) — and converts them into proof for the long-running-agent story.
12-week cohorts · 8–10 companies · 4 batches/yr. White-glove onboarding, early access, and credits in exchange for case studies, quotes, logos, and a Demo Day talk.
Attacks: W1 (no public customers), W8.
Free Runners + Code Reviewer for qualifying public repos, plus a Champions ladder that rewards sustained contributors — and the "Tenki was the assistant" OSS-PR content motion.
Attacks: W1, W3 (no community), W5 (no OSS flywheel), O7.
A recurring hackathon series on the dabl.club event engine. Devs build AI agents on Sandbox; prizes are Tenki credits + spotlight. Quarterly online + one flagship in-person.
Attacks: W1, W3, W5, O2 (agent explosion).
A 12-week design-partner cohort — 8–10 companies per batch, 4 batches per year — built to manufacture Tenki's first public references on a predictable schedule.
tenki.cloud/customersTwo pools: AI-native teams already standardized on Claude Code / Cursor, plus the Luxor network to de-risk the first batch. These are outreach targets, not confirmed customers — the cohort is the machine that converts a subset into Tenki's first public references.
| Target | Pool | Why a fit | Path / likely champion |
|---|---|---|---|
| Vercel | AI-native | AI SDK team runs agents heavily; high CI volume + agent sandboxing | Lee Robinson / Guillermo Rauch |
| Linear | AI-native | Public Claude Code adopter; quality-obsessed eng culture | Karri Saarinen |
| Replit | AI-native | Agent platform; per-job microVM isolation is native to their story | Amjad Masad |
| Sourcegraph | AI-native | Cody + agent workflows; Code Reviewer head-to-head | Beyang Liu |
| Browserbase | AI-native | Agent infra; "give your agent root without giving it yours" | Paul Klein |
| Modal | AI-native | Serverless compute peers; Sandbox cost/boot comparison | Erik Bernhardsson |
| Cognition (Devin) | AI-native | Autonomous agent needs the isolated run + review loop | Scott Wu |
| Blockware | Luxor network | Existing Luxor relationship; heavy infra/CI footprint | Luxor BD / Ethan Vera |
| Citrea | Luxor network | Bitcoin-adjacent eng team; warm via Luxor | Founder intro |
| Sia | Luxor network | Decentralized-storage eng org; CI + agent fit | Luxor network |
Demo Day mechanic. Quarterly public livestream where 8 cohort companies each give a 5-minute Tenki implementation talk (~90 minutes total, YouTube primary). Same playbook YC uses for Demo Day — it manufactures social proof at scale, creates a recurring press moment, and seeds the next cohort's pipeline. ~$10K production per event.
Application flow. Public landing at tenki.cloud/cohort → 5-question form (company, agents/CI in use, top goal, timeline, OK-to-publish) → 15-min DevRel screen → 30-min deep-dive with Eddie or Hayssem → decision within 7 business days → kickoff within 14 days.
Tenki isn't open-source, so it doesn't get the contribution flywheel OSS infra companies rely on. This program manufactures one synthetically: free products for public repos, plus a Champions ladder for the humans who sustain the community. The Champions are the multipliers — the OSS maintainers and agent builders already active in the adjacent communities, recruited to carry the long-running-agent story into them.
Free Runners + Code Reviewer for qualifying public repos: OSI-approved license, ≥50 stars or an active maintainer team, not a competing commercial vehicle. A "CI by Tenki" badge unlocks bonus credits. Per-repo ceilings and usage-threshold review keep free metal from being abused.
Recognize and reward sustained contribution — content, examples, and support — not one-off bounty chasing.
| Tier | How you get there | Recognition | Rewards |
|---|---|---|---|
| Contributor | 1+ accepted contribution | Listed on champions wall · Discord role | Swag · Tenki credits |
| Champion | Sustained contribution over a quarter | Featured profile · newsletter · early access | Larger credits · swag · conference travel stipend |
| Core Champion | Top contributors, Tenki-team selected (annual) | Co-marketing · roadmap advisory · "Core Champion" badge | Premium credits · Agent Camp invite · cash recognition award |
The "Tenki was the assistant" content motion. Open real OSS pull requests where an agent on Sandbox did the work, CI ran on Runners, and Code Reviewer reviewed it — then publish the PR, the diff, and the full prompt history. Radical transparency (Luxor's "honesty over kindness"), a three-product demo, and a Champions recruiting magnet in one. Target ~30 Tenki-assisted OSS PRs per quarter.
A recurring hackathon series run on the dabl.club event engine — 300+ AI events, an 86k-developer audience. The theme every edition: build a long-running, self-improving agent on Tenki Sandbox — the wedge no competitor is claiming. Prizes are Tenki credits + spotlight. Run as lead-gen, not prototype-gen: cohort targets in the judge seats, ecosystem partners co-sponsoring, and adjacent communities (E2B / Modal / Daytona + Cursor / Codex / Claude Code) recruited through the builders who multiply.
Quarterly 7–10 day online hackathons (global, remote) plus one flagship 1–2 day in-person edition in SF. Every participant gets a Sandbox credit grant — the activation mechanic.
Grand prize, runners-up, and category awards (Best Use of Sandbox, Best Review Agent, Best Multi-Agent System) paid in Tenki credits. Spotlight to 86k devs is the real prize.
Tenki eng + a Luxor leader + external agent builders. Equal-weight criteria: execution, genuine Sandbox use, originality, demo quality. Async scoring → top-10 live demos.
Anthropic (Claude Code), OpenAI (Codex), and Cursor as co-sponsors / featured harnesses — model-credit prizes stack at no cost to Tenki; ecosystem sponsors offset flagship production.
| Line | Annual |
|---|---|
| Tenki credit prize pool (4 online + flagship) | $20K (largely internal compute) |
| Flagship in-person production (venue, A/V, travel) | $30K |
| Online event ops (dabl.club engine) | $5K |
| Content production from submissions | $5K |
| Total | $60K |
Year-1 targets: 2,000 Sandbox activations · 400 submitted projects · 40 pieces of derivative content · ≥1 project promoted to an official Sandbox reference architecture.
| Program | Annual | Output target |
|---|---|---|
| Design Partner Cohort | $50K | 20 references · 8 quotable · 150 paid teams |
| Tenki for Open Source + Champions | $40K | 200 OSS repos · 25 Champions · 30 PRs/quarter |
| Agent Camp × Tenki | $60K | 2,000 Sandbox activations · 400 projects · 40 content pieces |
| Total | $150K | — |
Two structural cost advantages run through all three: credits and prizes are paid largely in compute Luxor owns, and the hackathon series runs on the existing dabl.club event engine — so this $150K buys far more than a cold-start program would. Programs run alongside the always-on operating system (content, community, events). See the overview for the full Year-1 budget.