# What is a data corpus for AI? > What is a data corpus for AI? A body of data treated as one set. In micro1's program it means a company's whole archive across its tools. Self-reported. By Matt Burns, Vaultfish. Canonical: https://vaultfish.com/answers/what-is-a-data-corpus-for-ai Published: 2026-07-26 Updated: 2026-10-13 ## The short answer A data corpus for AI is data treated as one set. In micro1's program it is a company's whole archive across its tools. micro1's page lists $100K+ to $1M+ tiers and 30+ employees, and it states no data-size threshold. The word sounds academic, and it is. Companies now use it for something closer to a filing cabinet. ## Where does the word come from? The plural is corpora. In natural language processing a text corpus is a collection of language data. Researchers often annotate it, for example by tagging each word's part of speech, so a study or a model can use it ([Wikipedia](https://en.wikipedia.org/wiki/Text_corpus)). That sense is a collection built for a purpose. The business sense keeps the idea of one body of data. ## What does corpus mean in micro1's program? In an undated recorded forum talk, David Revland, who works on human data at [micro1](/buyers/micro1), used "corpus" for everything a company has stored: Google Workspace, Slack, project management tools and GitHub. The corpus deal is the track where that whole body moves in one transfer, under a partnership in which micro1 says you keep ownership of the underlying data. micro1's pages state no data-size threshold. Its current public criteria list 30+ employees, mature operations and primarily English documentation, with US companies prioritized. The undated talk gave earlier size, staff and language figures, which are history. See [micro1's requirements](/answers/micro1-data-partnership-requirements) for both. ## How is a corpus different from a workflow? A corpus looks backward at records you already hold. micro1's workflows track produces new records, because employees grade an AI agent on live tasks. In the undated talk, Revland also described a possible top-up about six months after a corpus deal, covering data created since. Our [corpus vs workflows page](/answers/micro1-corpus-sale-vs-workflows-track) compares the two. ## Do other companies use the word? Not in the pages I read. Troveo calls it company data exports. [micro1's data partnerships page](https://www.micro1.ai/data-partnerships) says operational data and datasets (all). If a buyer says corpus, ask what it includes and which tools are in scope. ## What should you do before a corpus deal? Count it by tool, since micro1's data manifest form asks for a rough size per tool. Then sort it by who holds the rights, because a corpus includes material that isn't yours to license. The [rights check](/rights-check) asks whose records these are and what your contracts allow, and our [managed readiness service](/managed-readiness) works through the count with you without touching your data. micro1 may pay us a referral fee if a company we introduce goes ahead with it. [How we are paid](https://vaultfish.com/disclosure) Figure: comparison of Two ways to take part in micro1's program: Corpus deal (emphasized: Covers the whole archive, Whole archive, once), Workflows track (New records by task). In micro1's program a corpus is the whole archive, moved once, while the workflows track repeats by task. [Apply to micro1's data partnership (paid link)](https://req.micro1.ai/post/74520e8b-4673-4da5-b855-53609aaf6710?referralCode=f82e519e-5d64-4e0e-9574-760c84371cf2&utm_source=referral&utm_medium=share&utm_campaign=data_referral) ยท [micro1's own data-partnerships page](https://www.micro1.ai/data-partnerships) _Paid link. [How we are paid](https://vaultfish.com/disclosure)_ _Vaultfish helps you run the process. We are not lawyers and this is not legal advice. If you need counsel who knows this area, we can point you to one._ ## Sources - [Wikipedia](https://en.wikipedia.org/wiki/Text_corpus) - [micro1's data partnerships page](https://www.micro1.ai/data-partnerships)