When people ask how AVA knows their organization's data, they usually expect a technical answer about machine learning and model training. The real answer is more direct than that — and more reassuring.

AVA does not guess. She reads.

What happens the moment you connect a platform

The instant you connect a platform to ThatApp, a read process begins. ThatApp retrieves every record in every app or workspace on that platform — field values, comments, file metadata, revision history, relationships between records — and copies it into your private data lake in MongoDB.

This is not a sample. It is not a representative subset. It is everything, as it exists right now, plus every version of it that existed before.

From that moment forward, every change on the source platform — a new record, an updated field, a deleted item — is reflected in your lake within seconds. AVA's knowledge of your organization is continuously current.

What "knowing your data" actually means for AVA

AVA does not hold a memorized summary of your data in her context. She holds a live connection to your lake.

When you ask AVA a question — "How many open projects do we have?" — she does not recall a number she memorized from a previous session. She runs a query against your actual data lake, right now, and returns the real current count. Her answer is as fresh as the data that backs it.

This distinction matters. It means:

  • AVA's answers reflect what is actually in your systems today, not a snapshot from when you last asked
  • Records you added yesterday are already in her answers
  • Records someone deleted this morning are reflected as deleted
  • There is no "re-training" step where you push data to AVA — the lake is always the source and it is always live

What AVA learns about your organization's structure

Beyond raw records, AVA reads and retains your organizational context:

  • Which apps and workspaces you have, what they are named, and what fields they contain
  • The relationships between apps — which records reference which other records
  • How data flows through your workflows based on what field values change over time
  • Your team's naming conventions and terminology as they appear in your actual records

This structural knowledge is what allows AVA to answer questions like "find all projects that are behind schedule" without you having to explain what "behind schedule" means in your data. She has read enough of your records to understand how your organization uses its own fields.

Your data stays in your lake

AVA reads from your lake. She does not move your data anywhere else, aggregate it with other organizations' data, or use it to train any external AI model.

Your data lake is a MongoDB collection private to your organization. It is encrypted at rest and in transit. ThatApp employees cannot access your data without your explicit permission, and that access is logged.

The organizational context AVA builds from your data does not leave your environment. Other ThatApp customers — regardless of their platform, their industry, or how long they have been a customer — have no access to your data and no influence on AVA's understanding of your organization.

What AVA does not know

AVA knows what is in your connected platforms. She does not know what is in platforms you have not connected. She does not have access to the internet, to industry databases, or to general AI training data during your conversations. Her knowledge is deliberately narrow: it is entirely about your organization, your data, and your history.

If you ask AVA about a platform you have not connected, she will tell you that platform is not in your lake and offer to help you connect it.

Related: How Your Data Lake Works · Understanding Backup Behavior · How Your Data Is Isolated