Stored with Yield on AWS
Your original CSVs, prepared analytical files, and previews live in private, encrypted storage.
orders.csv Data ready to query
Original CSV and analytical files are not uploaded to OpenAI.CONNECT → ASK → ACT
Your business data.
A clearer way to use it.
You already have the information. Yield helps your team ask better questions and find useful answers—without building a report from scratch.
How it works What the AI sees Privacy & access
- 01 · Connect
Start with data you already have.
Upload a CSV export into a named dataset, such as Orders or Products. Check the preview, then add business context to explain your terms and definitions.
Your files, organized for analysis - 02 · Ask
Ask a question, not a query.
Ask in everyday language. Yield can work across your ready datasets, or you can choose which ones to focus on. It queries your data to support the answer.
One question can connect several datasets - 03 · Act
Understand the answer. Take the next step.
Get a direct answer, with a table when it helps. Dataset names and snapshot dates let you check how current the underlying information is before making a decision.
Useful context, with a source you can check
A QUESTION YOU MIGHT ASK
Less digging.
More understanding.
Compare products, explore a sales trend, or ask what changed. Yield brings the result back into a conversation you can keep building on.
Illustration with synthetic data, not a live customer answer.You
Which products brought in the most revenue in September?
Yield
Your field jacket led September sales with $18,420 in revenue. Here are your top three products.
| Product | Revenue |
|---|---|
| Field jacket | $18,420 |
| Weekend bag | $12,600 |
| Everyday tote | $7,800 |
Orders · Oct 1, 2026 Products · Oct 1, 2026
FOLLOW THE INFORMATION
What reaches the AI?
There’s a difference between storing your files and sending context to generate an answer. Here’s how that works.
Relevant context
Processed by OpenAI
Your question, relevant conversation and business context, private preferences, dataset structure, and limited query results help generate the answer.
Question & contextDataset structureBounded results
Query results can contain business information. AI processing is not entirely inside your workspace.Yield requests that OpenAI not store response application state. That does not mean zero provider retention: separate provider monitoring and retention rules still apply.
ACCESS WITH INTENTION
Shared knowledge.
Clear boundaries.
Your business is the boundary.
Server-side membership checks restrict access to your business’s data. Choosing a dataset never grants access to another business.
The right access for each teammate.
Owners manage people and settings. Data Managers can update data and business context. Chat Users can ask questions without changing datasets.
Your chat stays yours.
Teammates share business datasets, but each member has a private conversation. Personal memory is separate from shared business context and can be reset.
Queries have limits.
Data queries run with read-only rules and execution limits in an isolated process. Storage is encrypted and application traffic uses HTTPS.
A few things worth knowing.
How current are the answers?
Answers use your ready datasets. For now, you update them by uploading a new CSV; Yield does not automatically sync your source systems. Snapshot dates show when the underlying data was uploaded. A failed replacement leaves the previous successful dataset available.
Can I change or remove my information?
You can replace datasets, restore retained snapshots, delete datasets, clear your chat, and separately reset private memory. The active snapshot and four previous successful snapshots are retained. File cleanup waits for active processing and queries to finish. Contact support for verified account or business deletion; provider retention is separate.
Will the AI always get it right?
No. Answers depend on your data, business definitions, and the question. AI can make mistakes. Check the source dates and important results against your records, especially before a consequential decision. Yield supports your judgment; it does not replace your systems of record.
What should I upload during the pilot?
Use ordinary business records you are authorized to share. Do not upload health records, payment-card details, government IDs, passwords, or other highly sensitive information. This is a usage restriction, not a claim that Yield automatically detects or removes those fields.
START WITH A REAL QUESTION