CAPIVA INTAKE
Your idea, turned into context.
Capiva Intake is a free tool that does what a senior consultant does in the first five minutes of a project: it reads your raw idea, extracts what is already defined, flags what is missing, and surfaces what is contradictory — without making anything up. The result is structured context: a domain glossary and an initial spec, each item carrying its source.
This is how every AI-first project should start: context before code. That context is what seeds a harness — the working environment that makes AI produce useful code instead of generic output. Describe your idea below and watch the intake happen.
How it works
You describe the idea
In plain language, the way you would explain it to a colleague. No rigid form, no mandatory jargon.
The tool interrogates
It returns questions specific to your domain — with options already proposed — and shows the tensions it found in your material.
You get the context
Domain glossary, requirements, stakeholders, contradictions and open questions — ready to seed your harness.
Start with the intake
- 1Idea
- 2Questions
- 3Context
Write in natural language, the way you would explain it to a colleague. The more concrete, the better the intake.
What you get
At the end the tool hands you two Markdown documents, with every item tagged by its source (from the idea, from your answer, or to be defined):
- CONTEXT.md: the domain glossary: terms, definition, name in code, terms to avoid, and domain rules.
- INTAKE-summary.md: the initial spec: objective, functional and non-functional requirements, stakeholders, constraints, contradictions found, and open questions.
You see both on screen. To download the .md files, you leave your email.
How the tool works under the hood
- It makes two calls to the Gemini 2.5 Flash model, from Capiva’s server: the first analyzes the idea (what is already defined, what is missing, what contradicts itself) and writes the questions; the second turns the idea and your answers into the two documents.
- The idea can be up to 8,000 characters. There is a per-IP usage limit and a global daily limit.
- Your idea is not stored on Capiva’s servers. Even so, do not send personal or sensitive data.
- The documents come out in the language the idea was written in.
Where the intake fits in a project
The intake is step 1 of Capiva’s method: context before code. CONTEXT.md and INTAKE-summary.md are the two artifacts that seed a capivaOS harness, the environment around the agent that makes the AI work with your domain’s vocabulary and rules.
Frequently asked questions
Is Capiva Intake free?
Yes. Analyzing the idea, answering the questions and seeing the generated context costs nothing. Downloading the .md files asks for an email.
Which AI model does the tool use?
Google’s Gemini 2.5 Flash, called from Capiva’s server. The API key never reaches the browser.
Does Capiva keep my idea?
No. The idea is processed to produce the answer and is not stored on Capiva’s servers. Do not send personal or sensitive data.
What do I do with the generated files?
Put CONTEXT.md and INTAKE-summary.md in the project repository. They are the starting point of a harness, such as capivaOS, so the AI agent works with your domain’s terms and requirements.
Why is the address /prompt-machine?
The page lives at the address of Capiva’s former Prompt Machine. The Intake does not return a ready-made prompt: it returns context (glossary, requirements, contradictions and open questions), which is what an agent needs to write useful code.