Business Plan

VibeHub is AI-native version control. Git versions text files, but modern AI-assisted software is bigger than that: the spec, the prompts, the agent transcript, the model configuration, the code diff, the eval result, and the deployed artifact. VibeHub captures all of it as a single forkable, replayable unit we call a vibe — and hosts a community where builders publish, discover, fork, and replay each other's work.

The product has two halves: the vb command-line tool, which captures vibes automatically as you build with AI, and the VibeHub platform, which stores, renders, and socializes them. The core product is free; metered actions (agent turns, transcript-captured commits, remote pushes, and eval runs) are paid for with credit packs starting at $4.99.

VibeHub is built and operated by AW3 Technology, Inc., headquartered in New York, NY. The company develops the vb CLI, the VibeHub.co platform, and the surrounding developer surface (REST API, TypeScript SDK, and MCP server).

Mission

Make AI-assisted software reproducible, reviewable, and shareable — so that anyone can build on anyone else's AI work instead of starting from a blank prompt.

Vision

A world where every meaningful AI build is published as a vibe: forkable like a repo, replayable like a recipe, and verifiable like a test suite. VibeHub aims to be the default home for that work, the way GitHub became the default home for source code.

Operating principles

• Ship in public. The alpha, the changelog, the roadmap — and this business plan — are open.
• Builders own their work. Code, prompts, specs, and vibes belong to the people who make them.
• Capture, don’t interrupt. Versioning should be a side effect of building, never a second job.
• Charge for metered value, not for access. The core product stays free.

AI now writes a large share of new code, but the tooling for managing that work hasn't caught up. Git was designed to version text files written by humans; it has no concept of the prompt that produced a change, the agent transcript that explains it, or the eval that proves it works.

What breaks today

• Context evaporates. Prompts and agent transcripts live in chat histories and terminal scrollback. A week later, nobody can say why the code looks the way it does.
• Nothing reproduces. Without the model, parameters, and seed that generated a change, “it worked on my machine” becomes “it worked on my model.”
• Review is blind. Teams review AI-generated diffs without the reasoning trail behind them, so review quality degrades exactly as code volume explodes.
• Forking strips the “why.” Cloning a repo copies the output of an AI build but none of the intent — the spec, prompts, and evals that made it good.
• Evals are an afterthought. There is no standard place to attach “did it actually work?” to a unit of AI-generated work.

These aren't edge cases — they're the daily experience of every developer who builds with Claude, Cursor, Copilot, or any coding agent. The gap between how software is made and how it's versioned is the opportunity.

A vibe bundles everything that defines a unit of AI-assisted work into one versioned, forkable object:

The vb CLI

The vb CLI wraps the builder's existing workflow. It captures transcripts at commit time, pushes bundles to VibeHub with vibehub-remote, and replays published vibes locally — same spec, same prompts, same config — so a fork starts from the full context, not just the code.

The platform

VibeHub.co hosts, renders, and socializes vibes: rich bundle views, public and private visibility, forking, replay, follows, and discovery. A built-in vibe agent answers questions about any bundle and helps builders work inside the dashboard.

Developer surface

• REST API for programmatic access to vibes, builders, and evals.
• TypeScript SDK for integrating vibe capture into existing tools.
• MCP server so coding agents can read and publish vibes natively.
• Integrations with the editors and agents builders already use.

Infrastructure

The platform runs on Vercel with Neon serverless Postgres and blob storage for bundle artifacts; agent features are powered by frontier language models. The stack is deliberately serverless so infrastructure cost scales with usage — which matters because revenue does too (see Section 07).

There are roughly 30 million professional developers worldwide, and surveys from GitHub, Stack Overflow, and JetBrains consistently show that a large majority now use AI tools in their daily work — the fastest tooling adoption curve in the history of the industry. Beyond professionals, AI is minting a new population of builders who never used Git at all: designers, founders, and operators shipping real software through prompts.

Why now

• Agent-written code is becoming the majority of new code in AI-forward teams, and none of its context is versioned anywhere.
• Reproducibility and AI-governance pressure is rising — teams increasingly must show how AI-generated changes came to be.
• A new generation of builders is forming its tool habits right now, before an incumbent owns this workflow.

Durable advantages

• The bundle format: capturing all seven artifacts as one unit is the product — partial copies miss the point.
• Replay: a fork that re-runs the original spec and prompts is qualitatively different from a code clone.
• Network effects: every public vibe makes the platform more valuable to the next builder, compounding like early GitHub.
• Tool neutrality: VibeHub works alongside whatever agent or editor wins, rather than competing with it.

VibeHub is free at its core — hosting, browsing, forking, and publishing public vibes cost nothing. Revenue comes from credits, a single metered currency spent on the actions that consume real compute:

Credits are sold in packs, priced for the hobbyist through the heavy team:

Why credits

• Revenue scales with usage, and so does our serverless cost base — margins stay healthy at every volume.
• No seat-gating means no friction to adoption: the free tier is genuinely useful, and payment starts exactly when value does.
• Credits are priced above the underlying compute and model cost, targeting 80%+ blended gross margin.

Expansion revenue

The roadmap adds team workspaces with shared credit pools and admin controls (2027), followed by enterprise plans — SSO, audit trails, private replay infrastructure, and compliance reporting for organizations that need to govern AI-generated code. Longer term, a marketplace for premium vibes and evals gives top builders a revenue share and VibeHub a take rate.

Phase 1 — Alpha (now)

Invite-driven waitlist, building in public. Every changelog entry, blog post, and public vibe is content marketing. The goal of this phase isn't revenue — it's a core of builders who publish vibes worth forking.

Phase 2 — Community-led growth

Every public vibe is an acquisition channel: it ranks in search, gets shared in the communities where vibe coders already live, and ends with a fork button. The viral loop is built into the product — you can read a vibe without an account, but forking and replaying it pulls you onto the platform and into the vb CLI.

Phase 3 — Teams

Individual builders bring VibeHub to work, the same bottom-up motion that built GitHub, Figma, and Notion. Team workspaces convert that organic usage into recurring, multi-seat revenue without an outbound sales force.

Channels

• Public vibes and SEO: every published bundle is an indexed, shareable artifact.
• Developer content: docs, changelog, and engineering blog — already live and indexed.
• Integrations: meeting builders inside the agents and editors they already use, via the SDK and MCP server.
• Community: X, GitHub, and Discord, where the vibe-coding conversation is already happening.

We treat the roadmap as a sequencing commitment, not a date commitment — in an ecosystem moving this fast, order matters more than quarter.

VibeHub is built by a deliberately small, senior team at AW3 Technology that uses its own product daily — the platform's features ship as vibes. Operating AI-natively keeps headcount, and therefore burn, far below what this product surface historically required.

How we stay lean

• Serverless everything: Vercel compute, Neon Postgres, and blob storage mean zero idle infrastructure and no ops team.
• AI-assisted development with human review — measured in our own evals, shipped through our own CLI.
• Community-led growth in place of a paid acquisition budget during alpha and beta.
• Hiring follows revenue: the first dedicated hires land in developer relations and platform engineering as beta usage grows.

Key operating costs

• Model inference for the vibe agent and eval runs — the dominant variable cost, recovered through credit pricing.
• Compute, storage, and bandwidth for bundle hosting and replay.
• A small fixed base: tooling, compliance, and (eventually) salaries.

The model below is illustrative — it exists to show the shape of the business, not to promise the numbers. The drivers are registered builders, the share who buy credits, and average credit spend per paying builder.

Unit economics

• Gross margin: credits are priced above metered compute and model cost, targeting 80%+ blended margin.
• CAC: near zero during community-led phases; every public vibe is unpaid acquisition.
• Payback: immediate — credit packs are prepaid, so revenue arrives before the costs it covers.

Funding

The company is founder-funded through alpha. The serverless cost base and prepaid revenue model keep the burn floor low enough that outside capital is a choice about speed — accelerating team workspaces and enterprise — rather than a requirement for survival.

Questions about this plan, partnerships, or investment? We'd love to talk.

• Email: will.schulz@aw3.tech
• X: @aw3technology
• GitHub: aw3-technology