An outside-in plan for Tenki — the developer platform for AI code review, drop-in GitHub Actions runners, and disposable microVMs for AI agents (a product of Luxor Technology). Built by a developer-relations leader who shaped the developer program at Cisco Meraki, led DevRel at MetaMask, and brings an 86k-developer community — tuned to Tenki's wedge and the agent era.
Three products shipping weekly on one Firecracker substrate, on compute Luxor owns. But almost no public customers, no community, no content engine, an empty blog. That gap is the entire DevRel opportunity — and the timer is running.
Why DevRel is the lever. Tenki doesn't need more engineering hours — it needs a developer brand, public trust (especially on the precision/honesty front), named references, and a defining agent-era narrative before better-funded incumbents (Blacksmith, Depot) and the AI-review pack (CodeRabbit, Cursor) define the category for it. We attack those gaps directly — and bring an 86k-developer audience to seed the funnel on day one.
"We started Tenki on the belief that compute is the new digital oil. As global demand for compute accelerates, scalable and tradable compute is set to become one of the most valuable markets of our lifetime."
— Tenki, About
Tenki was born inside Luxor — a profitable, SOC-2 Bitcoin-mining infrastructure company — as it extended its "compute as a commodity" thesis from mining into generalized, AI-era compute. The dev products (runners, review, sandbox) are the consumer wedge; Luxor's owned compute and GPU marketplace are the supply.
This is the through-line for every post, talk, benchmark, and pitch: the agent era runs on compute you can trust — Tenki is where it runs.
Runners, Code Reviewer, and Sandbox all run on the same Firecracker microVM substrate. In the agent era they converge into one thing: where an autonomous agent writes, runs, and gets its code reviewed in isolation. No competitor owns the whole loop.
CI + review + agent sandboxes on the same per-job-isolated microVMs. The convergence story is the differentiation pure-plays can't match.
Tenki rides Luxor's captive bare-metal (mining + the new GPU/AI build-out) instead of paying retail cloud margins like every VC-funded competitor.
Code Reviewer leads recall (68.9%) but trails on precision (29.9%). The honest gap is the most interesting unsolved problem in the category — and our content engine.
dabl.club's audience (36–40% open rate) becomes an attributed acquisition channel feeding Tenki-owned community, benchmarks, and events.
Field notes from hands-on use and a read of the live sandbox market — threads I'd pressure-test with the team in week one.
E2B, Modal, and Daytona optimize short-lived sandboxes — spin up, run, tear down. The 2026 frontier is long-running, self-improving agents. Tenki's persistent volumes + snapshot/restore + microVMs fit that — a category no competitor is claiming.
I installed the Sandbox ADE — the native-app approach is a real edge over hours of self-hosting Daytona — but hit a setup snag on first deploy. The highest-leverage first move: a First-Five-Minutes audit to get time-to-first-successful-sandbox under five minutes.
Most "sandboxes" are afterthoughts of a cloud business; Daytona is great at visibility but expensive and hard to self-host. Tenki wins on a sandbox-first brand, easy native UX, and owned-compute price.
And the GTM to match: capture adjacent communities (E2B/Modal/Daytona, plus the Cursor / Codex / Claude Code crowds) via multipliers; run hackathons as lead-gen, not prototype-gen — customer targets in the judge seats, partners co-sponsoring, and a branded event hub that spins off content.
A proven 7-initiative framework, tuned to Tenki's wedge. Every program ties to a Strength to amplify, Weakness to close, Opportunity to capture, or Threat to neutralize.
The flagship is TenkiBench — the agent × open-model matrix (every coding agent × every open model, sliced both ways), run end-to-end on the stack: Runners orchestrate, Sandbox runs each pairing in parallel, Luxor compute serves the models, plus bring-your-own-evals. Alongside it: reproducible AI-review, runner, and sandbox benchmarks — published even where Tenki loses (honesty is the credibility). Leaderboard at tenki.cloud/benchmarks.
Official, maintained references: Claude Code + Sandbox (co-marketed with Anthropic), Codex + Sandbox (OpenAI), Cursor / Devin / Cline. Each ships a repo + 5-min video + guide + cost calculator. "Give your agent root without giving it yours."
Migration content + Migration Wizard storytelling, honest comparison guides (vs GitHub Actions, Depot, Blacksmith, Namespace), a public cost calculator, and a first-five-minutes onboarding overhaul that crushes time-to-first-green-run.
Discord (per-product + per-agent channels), weekly office hours, weekly build-in-public stream, biweekly newsletter, "How I ship with Tenki," coordinated HN/Reddit/X launches, podcast guesting — seeded from dabl.club's 86k developers.
12-week cohorts · 8–10 companies · 4 batches/year. White-glove onboarding in exchange for case studies with quantified ROI, quotes, co-presented talks, and logos. Quarterly public Demo Day livestream.
Free Runners + Code Reviewer for public repos; a Champions program for sustained contributors; an "Agent Camp × Tenki" hackathon series (leveraging the dabl.club event engine) where devs build on Sandbox; plus OSS PRs where Tenki was the assistant, diff + prompt history attached.
GitHub Marketplace listing; Anthropic (Claude Code) + OpenAI (Codex) + Cursor co-marketing via the Sandbox references; AI-agent registry placements; the Luxor compute/enterprise tie-in; alliances across the AI-coding ecosystem.
Instrument the funnel + DevRel scorecard · first-five-minutes overhaul · Discord + weekly office hours + weekly build-in-public stream · Benchmark Drop #1 (the AI-reviewer teardown) · migration comparison guides + cost calculator · Cohort #1 opens · dabl.club cross-pollination begins.
Milestone: community seeded (500+ engineers), funnel instrumented, 1 benchmark live, time-to-first-run improving.
Sandbox reference architectures (Claude Code, Codex, Cursor) · Anthropic/OpenAI co-marketing · Agent Camp × Tenki hackathon #1 · Cohort #1 Demo Day livestream · Benchmark Drop #2 · Tenki-for-Open-Source launch.
Milestone: 1,500 weekly active workspaces, 8 case studies in flight, agent-era narrative landed.
Flagship event presence (AI Engineer World's Fair, GitHub Universe, KubeCon) · enterprise references + the Luxor compute narrative · Champions program · Cohorts #2–3 · Benchmarks #3–4 · first analyst attention.
Milestone: 3,000 weekly active workspaces, 20 named references, 150 paid Team/Enterprise accounts.
Tenki keeps engineering focus on the product. I bring the playbook, the tools, an 86k-developer audience, and the cadence.
Substance is identical in all three. Only exclusivity and ramp differ. Nebius conflict disclosed + wound down; location handled up front.
I've built developer programs from zero and led DevRel at scale — and I bring an 86k-developer audience with me. Tenki gets the operating system, the network, and a builder who ships, on day one.
Docs, webinars, a partner ecosystem, and the reusable DevRel playbook I still run today.
Led developer relations through multiple L2 launches across the crypto ecosystem.
My own developer community (36–40% open rate), pivoted hard into AI agents in early 2025 — a day-one distribution channel for Tenki.
This whole strategy and site is the proof — built before the first conversation. I'd rather file the PR than write the press release.
If full-time fits: a ~60-day Nebius wind-down and I join all-in. To de-risk first: a 90-day paid pilot (~$155K) with defined OKRs and a conversion gate — Phase 1 of this plan, with hard metrics. Or fractional at $35–50K/mo. Same playbook in all three: Benchmark Drop #1, weekly office hours, Cohort #1, Discord, OpenClaw, and dabl.club distribution from day one.