Most questions people bring to aVenture follow one of a few shapes. Turn a website into a company's name, logo, and tags. Match a spreadsheet of domains to companies. Size a market. Read what a company raised and who wrote the checks.
The calls for each of those already existed, but you had to find and chain them yourself from the API reference. The new Use Cases and Examples page in the aVenture docs does that work for you. It covers eight jobs, and each recipe shows the API request, the aventure CLI command, and the MCP tool call side by side, followed by a real response from the live data.
Every recipe starts from something exact
Each recipe is deterministic: the same input returns the same record. Nothing in the recipes guesses a record from a bare name. They start from a website, a profile URL, a slug, or an id, and each of those identifies one record.
That matters when the output feeds code. A dashboard that shows a company's logo, or a pipeline that joins your CRM to aVenture, cannot act on "probably this one." When a name is all you have, Lookup handles the judgment call separately and says when it cannot decide.
A company's brand kit from its website
Send a website you hold to GET /v1/entities/brand and read back the company's brand name, legal name, logo, links, and tags in one call:
curl -s -H "Authorization: Bearer $AUTH_TOKEN" \
"https://api.aventure.vc/v1/entities/brand?url=stripe.com" |
jq '{nameBrand: .core.nameBrand, nameLegal: .core.nameLegal,
logoSquare: .core.image.logoSquare,
tag: [.classification.tag[] | {name, isPrimary}]}'
For Stripe, that returns Stripe, Stripe, Inc., the logo path /logos/gfc/gfcNIiXzgUdP6tGupXsKe.png, and the tags Platform and SaaS, with SaaS marked primary.
The lookup ignores the scheme, a leading www., and the case of the host, so stripe.com, https://www.stripe.com, and HTTP://STRIPE.COM all return the same record. A LinkedIn or X profile URL works too.
The logo field is a storage path. For a full image URL you can put straight into an <img> tag, GET /v1/entities/{entityId}/logo returns a cdnUrl. Each brand read counts one brand view per company each month, on an allowance separate from profile views, and a repeat read that month is free. Company brand lookup covers each plan's allowance and what to build with it.
Matching a list of websites to companies
POST /v1/entities/lookup-matches takes an array of websites and answers one row per input, in input order, so the answer joins straight back to a spreadsheet. Each row says MATCHED, AMBIGUOUS, or MISSING:
[
{ "input": "stripe.com", "status": "MATCHED", "detail": { "core": { "nameBrand": "Stripe" } } },
{ "input": "notion.so", "status": "MATCHED", "detail": { "core": { "nameBrand": "Notion" } } },
{ "input": "zzqx-unknown-domain-4417.com", "status": "MISSING", "candidateEntityId": [] }
]
An AMBIGUOUS row lists every record that shares the URL instead of picking one for you. A sibling endpoint, POST /v1/entities/lookup-batch, returns only the records it found and skips the rest. Use it when you want records, not a row-by-row reconciliation.
Target lists, funding, and portfolios
The remaining recipes chain the same pattern: filter a list, open one record, read what is attached to it.
- Target list.
GET /v1/entities?stage=Seed&tag=Fintech&headquartersState=New%20Yorkfinds 111 seed-stage fintech companies headquartered in New York. Filter values are display names, so the state isNew York, notNY. - Funding history. Stripe's 19 rounds come back newest first, and naming a round label such as
Series Ilists its 10 investors. - Investor portfolio. Resolve Sequoia Capital from
sequoiacap.com, then list the 44 companies it backed with a round since January 2025, one row per company. - Founder background. Resolve Patrick Collison from his LinkedIn URL and list every company role, current and past.
- Competitors. Stripe's stored competitors include Block, Wise, and Visa, each with the reason it is listed. This read is free on every plan.
- News monitoring. Ask for articles linked to a company since your last check, using the date of each check as the next
publishedAfter.
Each recipe on the docs page shows the exact request, the CLI command, and the MCP call, plus the response fields worth reading and the parameter details that trip people up.
Where the recipes stop
These recipes read records aVenture already holds. A company that is not in aVenture comes back MISSING or 404, and the recipes do not create it. Use Lookup or request research for that.
Data quality is uneven at the edges. A domain can belong to two records, such as a firm and its fund, and some rounds have no amount or no round label. The recipes say how to handle each case, but they cannot fill in data that is not there.
Try a recipe
Every recipe needs an aVenture account. Create one, pick your surface in the quickstart, and start with the brand kit recipe. If you are building something the page does not cover yet, tell us through support.
