Learn
Where does an AI agent platform store your data?
What gets persisted when agents run a process, why it has to be persisted, and how access to it should be controlled.
The short answer
Agents produce artifacts as they work: drafts, dossiers, comparisons, decisions. A platform either keeps those artifacts or throws them away. Keeping them is what makes replay, audit, and resumption possible, which means the meaningful question is not where the data sits but who is allowed to see it.
The two architectural choices
The first choice is to process in memory and discard. Nothing accumulates, but nothing can be replayed either. There is no audit trail of what the agent saw or produced, and a run that stops partway through cannot resume, because the working state is gone.
The second choice is to persist and govern. The artifacts of each stage are stored, so a run can be inspected after the fact, resumed where it stopped, and evidenced to an auditor. That choice is what most enterprise processes require.
Why access is the real question
Once artifacts persist, the risk moves from loss to exposure. A vendor dossier assembled for a risk review, a redlined contract, a price comparison: each is appropriate for some people at some point in the process and inappropriate elsewhere. Document-level permissions alone are a blunt instrument, because the same document is sensitive at one stage and routine at another. Visibility needs to follow the stage, not just the file.
Where VeroTX sits
In VeroTX, every artifact a WorkStream produces is stored and replayable. Access is governed by stage-specific audience rules, so visibility is a property of the stage rather than of the document. A document surfaced at legal review is not visible from an earlier stage. Every action is recorded in the Execution Ledger.
Where to next
Read deeper
Product and platform pages on the same topic, drawn from the VeroTX knowledge base.
Search everything in one place in the Research Center, or ask Veroli, which answers from the same knowledge base.