VibeMap Logo

Questions about VibeMap

What VibeMap does, what it costs, where your data goes, and how a spec reaches the coding agent that builds it. 38 answers, no sales pitch.

The basics

What VibeMap is, who it's for, and what you get out of it.

What is the spec layer for vibe coding?
The spec layer is the planning stage that sits between a product idea and an AI coding agent. VibeMap is the spec layer for vibe coding: it turns a plain-English idea into linked personas, features, user stories with acceptance criteria, a database schema, and a page inventory, so tools like Cursor, Lovable, or Claude Code build from a structured brief instead of an evolving prompt. Working from a spec keeps AI-generated codebases coherent as they grow, instead of accumulating conflicting files and dead code.
What is VibeMap?
VibeMap is an AI-powered product specification tool that transforms a plain-English description of an app into a complete product blueprint. When a user describes their app idea, VibeMap generates personas, feature lists, user stories with acceptance criteria (Gherkin-style Given/When/Then), a relational database schema, and a page inventory. The pipeline runs in under 30 minutes and does not require product management experience. VibeMap sits upstream of task trackers like Linear and Jira, and upstream of AI code generators like Cursor and Bolt.
How long does it take to get a first spec?
A complete first pass in VibeMap takes 10 to 30 minutes. You sign in, click New Project, and describe the product in plain English. VibeMap asks a few clarifying questions about target users, technical stack, and constraints, then runs a multi-step pipeline that produces the blueprint: personas, features, user stories, acceptance criteria, pages, and schema. Artifacts stream in as they are generated, so you can read the personas while the stories are still being written. Refining the result through the conversational agent typically takes another 15 to 30 minutes.
Do I need product management experience to use VibeMap?
No. VibeMap is built for solo founders, indie hackers, and engineers who have never written a PRD. You describe the product in your own words; VibeMap supplies the structure — INVEST-format user stories, Gherkin acceptance criteria, MoSCoW priorities, a normalised schema — that a trained product manager would otherwise have to author by hand. The frameworks are applied for you and explained inline. Experienced PMs use VibeMap differently: as a fast first draft they then edit, rather than as a tutor.
What exactly does VibeMap generate?
VibeMap generates seven linked artifacts: a project summary, three or more user personas with goals and pains, a feature list with MoSCoW priorities and effort sizing, user stories in INVEST format, Gherkin-style acceptance criteria tagged happy path / edge case / failure state, a page and component inventory with routes and data fields, and a relational database schema. The artifacts are linked rather than merely adjacent — every story traces to the persona it serves, every acceptance criterion to its story, every schema table to the pages that read it.

How it compares

Where VibeMap sits relative to AI app builders, coding agents, and PRD generators.

How is VibeMap different from Lovable, Bolt, or Base44?
Lovable, Bolt, and Base44 generate working prototypes directly from a prompt. VibeMap works upstream: it produces the production-ready specification those tools (or your own AI coding agent) build from. Teams use both — VibeMap for the linked spec, a builder for the code — so the app survives past the prototype stage. The practical difference shows up in month two: a prototype built from a prompt has no artifact explaining why it works the way it does, while a VibeMap spec does. Detailed comparisons are at vibemap.ai/alternatives.
Does VibeMap write the code?
No, and that is deliberate. VibeMap produces the specification; your AI coding agent writes the code against it. This is why VibeMap publishes an MCP server rather than a code generator — you keep Cursor, Claude Code, Windsurf, or whatever you already use, and VibeMap supplies the structured context those tools are missing. The one exception is App Builder on Pro and above, which can scaffold from the spec and push to GitHub, but the spec remains the product and the code remains yours.
How is this different from asking ChatGPT to write a PRD?
ChatGPT produces a document; VibeMap produces a linked data model. A chat transcript cannot tell you which user stories break when you rename a persona, cannot be queried by your coding agent over MCP, and has to be re-pasted into every new conversation. In VibeMap each artifact is a row with real relationships, so editing a feature flags the stories, criteria, schema fields, and pages that depend on it. VibeMap also applies the frameworks consistently — INVEST, Gherkin, MoSCoW — rather than whichever the model happened to reach for.
Why not just use a PRD template or GitHub Spec Kit?
A template tells you what sections to fill in; it does not fill them in, and it does not keep them consistent with each other. GitHub Spec Kit is closer — it is a structured, spec-driven workflow for coding agents — but it lives in your repository as markdown you author and maintain by hand. VibeMap generates the content, keeps the cross-references live, and exposes the result over MCP and a REST API so an agent can read and write it directly. Template-based workflows and VibeMap are not mutually exclusive; VibeMap removes the authoring cost.

AI coding agents

MCP, your IDE, and how the spec reaches the agent actually writing code.

Does VibeMap work with Cursor and Claude Code?
Yes. VibeMap publishes an MCP server (@vibemap.ai/mcp-server on npm) that connects your spec directly to MCP-compatible agents like Cursor, Claude Code, and Windsurf. The agent reads personas, user stories, acceptance criteria, and database schema straight from your VibeMap project — no copy-pasting context into prompts. This keeps the agent grounded in one source of truth while it builds. MCP access is open on every VibeMap plan, including Free.
How hard is it to connect my IDE to VibeMap?
It is a config paste, not an integration project. In VibeMap, open Account → Developer and generate a personal access token (prefixed vm_), then open Account → MCP, pick your IDE — Claude Desktop, Cursor, or Windsurf — and copy the ready-made JSON config it assembles for you. Paste that into your IDE's MCP settings and substitute your token. The same screen shows IDE Connection Status, listing each key and when it last authenticated, which is the fastest way to confirm the connection actually worked.
Does using my own coding agent spend my VibeMap tokens?
No. VibeMap's token budget meters only VibeMap's own generations. If you author your spec with your own connected agent over MCP — your IDE agent writing personas, features, stories, criteria, pages, and schema straight into the project — that work runs on your agent's tokens and VibeMap only stores the result. The rule in one line: you are billed when VibeMap generates, and anything your own agent authors is free. This is how users on smaller plans stretch a budget a long way.
Can my agent write back into the spec, or only read it?
Both. The VibeMap MCP server and REST API are two front doors onto the same surface, sharing one auth model, and that surface supports writes as well as reads. Your agent can create and update personas, features, user stories, acceptance criteria, pages, and schema tables in your project. Plan gates still apply per entity — authoring pages or schema over MCP needs Starter, exactly as it does in the browser — but the access itself is not what varies between plans.
How does VibeMap help an agent build test-first?
VibeMap's acceptance criteria are written as discrete, Gherkin-style Given/When/Then statements, each tagged with a scope (front-end, back-end, or full-stack) and linked to its user story and persona. Over MCP, an agent can claim a single acceptance criterion, implement exactly that behaviour, and submit it for review, rather than being handed a whole feature and improvising the boundaries. Narrower, pre-scoped work is the practical defence against the duplicated components and conflicting files that unplanned AI codebases accumulate.

Spec quality and control

What the output actually looks like, and what happens when the AI gets it wrong.

How detailed are the user stories and acceptance criteria?
VibeMap generates user stories that follow the INVEST framework — Independent, Negotiable, Valuable, Estimable, Small, Testable — with each story linked to a named persona and a specific feature. Acceptance criteria are written in Gherkin-style Given/When/Then format and are categorized as happy path, edge case, or failure state. This level of detail means QA engineers and AI coding tools like Cursor or Bolt can consume the output directly without manual clarification or rewriting.
What if VibeMap gets something wrong or invents a detail?
Treat the first pass as a draft, not an oracle. Every artifact VibeMap generates is editable, and the conversational agent lets you correct anything in plain language rather than by hand. VibeMap is designed so mistakes are cheap to catch: artifacts are short, linked, and reviewable in minutes, which is the opposite of discovering a bad assumption after an AI coding agent has built three features on top of it. Nothing regenerates silently — when a correction has downstream effects, VibeMap shows you a diff and asks.
Can I edit the user stories and acceptance criteria that VibeMap generates?
Yes. Every artifact VibeMap generates is editable. Users can rewrite, reorder, or extend user stories, acceptance criteria, personas, and schema fields directly in the VibeMap interface. When one artifact changes, VibeMap propagates the change to dependent artifacts — for example, editing a user story updates the linked acceptance criteria and re-prompts the schema generator if the change implies a new data field. A conversational agent lets users refine artifacts through natural-language dialogue.
How does VibeMap handle changes to existing work items?
VibeMap propagates changes across the blueprint automatically. When a user edits a feature, VibeMap re-evaluates the linked user stories, acceptance criteria, schema fields, and pages, and surfaces any that need regeneration. Nothing regenerates silently — users see a diff and approve each change. This means renaming a persona, adding a compliance requirement, or restructuring a feature never leaves inconsistent artifacts behind.
How does VibeMap identify user roles and personas?
VibeMap generates three or more distinct user personas for every project, each with demographics, goals, pains, motivations, and a representative quote. Every user story, page, and acceptance criterion that VibeMap generates is linked back to the persona it serves, so teams never miss a user journey. Personas are editable — users can rewrite, rename, or add personas and VibeMap will regenerate dependent stories to match.
How does VibeMap prioritize features and user stories?
VibeMap scores every feature and user story using the MoSCoW framework (Must have, Should have, Could have, Won't have this iteration) and adds T-shirt size estimates (S, M, L, XL) for effort. Prioritization considers dependency chains — stories blocked by prerequisite work are flagged. Users can override any priority via drag-and-drop, and VibeMap re-orders the implied build sequence automatically.
How does VibeMap categorize acceptance criteria?
VibeMap categorizes every acceptance criterion into one of three types: happy path (the expected successful flow), edge case (boundary conditions and unusual but valid inputs), and failure state (error handling, invalid input, system failure). Each criterion is written in Gherkin-style Given/When/Then format and tagged with a front-end, back-end, or full-stack scope so engineering teams can distribute work accurately.
What are tags in VibeMap and how do they work?
Tags in VibeMap label every acceptance criterion with its engineering scope — front-end, back-end, or shared — and its persona owner. This is what makes VibeMap output directly consumable by AI coding tools: a Cursor or Bolt session can filter on front-end tags and build only the UI, while a backend engineer filters on back-end tags and builds the API. Users can add custom tags for squads, tech stacks, or sprint priorities.
How does VibeMap organize pages and components for my application?
VibeMap generates a complete page inventory from the user stories. Each page includes its route, purpose, linked personas, required data fields, and the reusable UI components it needs. Shared components — like headers, forms, and modals — are identified across pages so engineers build each once and reuse them. The page inventory feeds directly into downstream tools: copy it into Figma, import it into Cursor, or export it as a file structure.
Can VibeMap help with technical requirements, compliance, and performance constraints?
Yes. VibeMap captures technical requirements through a structured conversation during project setup. As users describe their product, VibeMap asks targeted clarifying questions covering technology stack preferences, compliance requirements such as GDPR or HIPAA, performance targets, and scalability constraints. These answers are embedded into the generated acceptance criteria and database schema so engineers receive a specification that already accounts for non-functional requirements, not just feature behavior.
Which AI models does VibeMap use?
VibeMap is multi-provider by design and routes each step to a model suited to it, drawing on Anthropic, OpenAI, and Google. Routine steps such as the summary, features, and user stories run on fast economy models; reasoning-heavy steps such as schema, pseudocode, and the business case run on frontier models. On Pro and above you can override the choice with a manual model selector. Claude 4 Sonnet becomes selectable on Team and Claude 4 Opus on Enterprise. VibeMap is never hard-wired to a single provider.
Does VibeMap remember project context across sessions?
Yes. VibeMap persists every project decision, artifact, and conversational exchange. Users can close the browser and return weeks later; the project state, pipeline progress, and conversation history are preserved. Projects can be shared with teammates on Team and Enterprise plans, with role-based access so engineers, PMs, and designers can collaborate on the same blueprint.

Pricing and tokens

What it costs, how the token budget works, and what happens when you run out.

How much does VibeMap cost, and is there a free plan?
VibeMap has five tiers billed monthly, with a 20% discount on annual billing: Free at $0, Starter at $19, Pro at $49, Team at $79, and Enterprise at $199. The Free plan needs no credit card and covers one project with the core pipeline — summary, personas, features, user stories, and acceptance criteria. Starter adds pages, sections, schema, the Blueprint map, and the conversational agent. Pro adds the business case, sprint planner, App Builder, and manual model selection. You can also try VibeMap at vibemap.ai/demo without signing up at all.
What are auto and premium tokens?
Every VibeMap plan includes a monthly token budget split into two pools. Auto tokens cover routine steps — summary, features, user stories — which run on fast economy models. Premium tokens cover reasoning-heavy steps such as schema, pseudocode, and the business case, which run on frontier models. The two pools are tracked separately: Free includes 300K auto and no premium, Starter 1.5M auto plus 250K premium, Pro 6M plus 1M, Team 15M plus 3M, and Enterprise 40M plus 8M. Budgets meter in rolling five-day windows so one heavy day cannot drain the month.
Will I burn through tokens fixing the AI's own mistakes?
This is the standard failure mode of credit-metered AI app builders, and VibeMap's architecture avoids most of it. VibeMap generates specification artifacts, not application code, so a wrong answer is a paragraph you edit rather than a build you re-run — and editing artifacts by hand or through the conversational agent is far cheaper than regenerating them. Separately, any authoring your own coding agent does over MCP costs you nothing from your VibeMap budget. You are billed when VibeMap generates, not when you correct it.
What happens if I run out of tokens mid-project?
Your project does not lock. Existing data stays fully editable, and your connected IDE agent can keep working through already-generated acceptance criteria — only new VibeMap generations pause until the next rolling window or an upgrade. One detail worth knowing: there is no automatic fallback from premium to auto models, so a generation that needs a frontier model is refused before it starts rather than silently downgraded. On Pro and above you can re-run it on a standard model using the model selector in the Generate dropdown.
Do unused tokens roll over to the next month?
No. VibeMap meters token budgets in rolling five-day windows inside each billing period, and unused allowance in a window does not carry forward. The design trades rollover for protection against the opposite problem: a single runaway day consuming an entire month's budget. In practice this means a steady working rhythm gets more out of a plan than one long sprint. Usage, the premium-versus-auto split, days remaining in the cycle, and a record-by-record table are all visible under Billing → Usage & Analytics.
Can I cancel anytime?
Yes. Open Billing → Subscriptions and choose Cancel Subscription; cancellation is scheduled for the end of the current paid period, and the page then shows the date your access stops. You keep full access until then. Note that no refunds are issued for partial months except where required by law, so cancelling mid-cycle does not generate a pro-rata refund. Payments run through Stripe, and card details are handled entirely by Stripe and never stored on VibeMap servers.
How does billing work for teams?
The VibeMap Team plan is $79 per month for a single shared subscription with unlimited seats and a shared token budget, rather than a per-person licence. It adds shared team workspaces and owner, admin, and member roles. Team and Enterprise plans are billed separately from any personal subscription, on the team's own settings page rather than your personal billing page. Enterprise at $199 adds Claude 4 Opus to the selectable model list, a DPA, SSO, and dedicated support.

Data and privacy

Who can see your spec, who owns it, and how you get it out.

Do you train AI models on my data?
On paid plans, no. Your inputs and generated outputs are not used by VibeMap or by VibeMap's LLM providers to train foundation models, and this is enforced at the gateway: paid inference is routed through OpenRouter with a no-data-collection policy that restricts routing to upstream providers whose own policy is not to log or train on prompts. The full detail, including what applies on the free tier, is in the VibeMap privacy policy at vibemap.ai/privacy.
Who owns the specs I create in VibeMap?
You do. VibeMap's terms state plainly that VibeMap claims no rights over your content. The personas, features, user stories, acceptance criteria, schema, and pages you generate are yours to use commercially, publish, hand to a client, or take elsewhere. VibeMap's role is to produce and host the artifacts, not to license them back to you.
Am I locked in? How do I get my spec out of VibeMap?
There is no ZIP-file export button, because the spec is designed as a live source rather than a document you download once. Every artifact — features, personas, stories, acceptance criteria, pages, schema — is readable through the VibeMap REST API and the MCP server, so your agent, scripts, or tools pull straight from the source of truth. There is a dedicated Jira export, and Account → Credentials → Export downloads a GDPR-style JSON copy of your profile, projects, and billing history.
Can I delete my account and everything in it?
Yes. Account → Credentials holds the Privacy & Data controls: Export downloads a JSON copy of your profile, projects, and billing history, and Delete account removes everything permanently. VibeMap also supports GDPR access rights through the same section. If you want a copy of your work before deleting, run the export first — deletion is not reversible.
Can I share a project with my team?
Yes, on VibeMap's Team and Enterprise plans. Team workspaces give you shared projects, a shared token budget, and owner, admin, and member roles, so engineers, PMs, and designers collaborate on one blueprint rather than passing copies around. The VibeMap Team plan is a single $79 subscription with unlimited seats rather than per-person pricing. On Free, Starter, and Pro, projects belong to the individual account.