◦ averian labs · ai-native venture studio
orbit. find ideas. build products. ship code. without you typing a prompt.
a team of specialist agents that runs the venture lifecycle. olivia finds the ideas. drew architects the products. mason plans the builds. cody writes the code. quinn reviews the pull requests. lana launches them. you merge the prs.
◦ the venture loop
find → validate → build → review → launch → learn → repeat.
nine phases. one continuous loop. each phase owned by a specialized agent.
◦ find
ideas
olivia surfaces opportunities aligned to your founder context and audience reach.
◦ validate
evaluate
olivia scores ideas across nine dimensions. only worth-building moves forward.
◦ architect
products
drew turns validated ideas into product outlines. positioning, modules, two-week mvv scope.
◦ plan
builds
mason sequences technical work into sprints with risks and dependencies surfaced.
◦ build
code
cody hands off build plans to claude code. prs land in your inbox for review.
◦ review
quality
quinn reads every PR claude code opens, compares it against mason's plan and cody's handoff, and tells you exactly where to look first.
◦ launch
go to market
lana drafts pricing, the first-10-sales plan, and day-one launch copy.
◦ learn
coming sooncustomer signal
paige (coming soon) watches revenue, churn, and customer health continuously.
◦ improve
coming sooniterate
the team feeds back into earlier phases. ideas refine. products improve. the loop closes.
◦ the team
seven specialists. and the meta-agents that build them.
each specialist owns a phase of the venture loop. the meta-agents are the platform layer that builds and coordinates the team.
specialists
the team that finds, builds, ships, and refines
Olivia
Idea Analyst
Evaluates business opportunities across nine dimensions. Worth-building, iterate-first, or move-on — calibrated to your founder context and audience reach.
Drew
Product Architect
Turns validated ideas into product outlines. Positioning, modules, two-week MVV scope, and the buyer-recommendation narrative for each persona.
Mason
Build Planner
Sequences technical work into sprints. Per-task implementation specs, risk register, external dependency map, and a build verdict.
Cody
Code Executor
Bridges build plans to Claude Code. Generates handoffs with prerequisites, founder tasks, and verdicts for safe auto-execution.
Quinn
QA Reviewer
Reviews Claude Code's pull requests. Compares the diff against Mason's plan and Cody's handoff, surfaces concerns, kudos, and a recommended review focus so the founder's final gate is sharp and fast.
Lana
Launch Strategist
Pricing, positioning, the first-10-sales plan, and day-one launch copy. Channel-specific recommendations for each product.
Indigo
Brand and Design Director
Turns a launch-ready product outline into a brand system: positioning, voice, color, typography, and Tailwind-ready design tokens — one idea translated into a coherent identity.
meta-agents
build and coordinate the specialist team
Forge
Builds new agents
Turns a founder brief into a complete, buildable slice spec + Claude Code handoff. The meta-agent that grows the team itself — Cody executes what Forge designs.
Helios
Orchestrates the team
Reads every product's lifecycle state and recommends the next action for each, so nothing stalls. The founder approves each step — Helios never fires an agent on its own.
◦ personality
brief. specific. honestly calibrated.
each agent is tuned to its phase. specific numbers. real verdicts. no corporate hedging.
should i build TradeReply for HVAC contractors?
Your network includes friends who own HVAC and plumbing companies. Founder context is strong. Pain intensity is high — missed calls cost $80-$300 per lost job. Score: 23/27. Worth building. Verify 10DLC carrier registration before promising a two-week ship.
how long will sprint 1 take?
22 hours of code across 14 tasks. One critical risk: Twilio 10DLC approval takes 10-15 business days. Don't promise pilot contractors a two-week launch unless you pre-register Twilio brand today.
generate the handoff for sprint 1
Done. Four prerequisites: Supabase project, OpenAI key, paid Twilio account, Vercel project. Three founder tasks: review PR, complete 10DLC registration, run the live phone test. Verdict: review_first. Ready for your approval.
◦ faq
questions i keep getting asked.
A multi-agent venture lifecycle platform. Each agent owns a phase of the loop — olivia finds ideas, drew architects products, mason plans builds, cody writes the code through claude code, quinn reviews the pull requests, lana launches them. The founder approves and merges. Built for solo founders or small teams running a portfolio of small experimental products.
Currently: averian labs (heather and craig manuel). Soon: founders who want to operate as a portfolio without becoming a ten-person team. If you build small products solo or with a co-founder and want infrastructure that compounds, orbit is for you.
Not yet. Orbit is currently proprietary while we validate the lifecycle and the agent calibration. We may open-source components, playbooks, or the agent runtime in the future.
Next.js App Router on Vercel. Supabase Postgres with Row Level Security. Claude Sonnet 4.6 via the Anthropic API for agent reasoning. Claude Code (via subscription) for code execution. Three.js for the orbital visualization. TypeScript throughout.
Orbit is currently single-tenant for averian labs while we validate the loop. If you want to follow along as we open it up, reach out via the footer.
Heather and Craig Manuel of Averian Labs. Built in partnership with Claude (Anthropic) — Cowork mode for strategic work and Claude Code for implementation.