How do I stop pasting customer details into ChatGPT every time?
Summary
Ending repeat pasting takes four steps: check your data policy, clear customer details from ChatGPT memory, keep one project per customer, and check the context of each answer. The steps rely on memory and project controls that vary by plan, region and workspace settings (OpenAI Memory help, 2026).
Saved memories are stored separately from chat history, so deleting the original chat does not automatically delete a saved memory (OpenAI Memory help, 2026).
Temporary chats do not create or update memories, even with personalization on (OpenAI Memory help, 2026).
With project-only memory, chats cannot reference conversations outside the project, and chats outside the project cannot reference it (OpenAI Projects help, 2026).
A ChatGPT project holds up to 40 files on Business and Enterprise, 25 on Go and Plus and 5 on Free (OpenAI uploads help, 2026).
These steps cover ChatGPT memory and projects as OpenAI's help pages describe them in October 2026. Enterprise, Edu and regulated workspaces can restrict memory and project sharing, and an admin setting overrides any step here. Whether your contracts allow customer data in ChatGPT at all is a question for your company policy.
How do you stop pasting customer details into ChatGPT, step by step?
You stop pasting customer details into ChatGPT by moving each customer's context out of the chat box and out of memory, into one project per customer that you keep current.
Check the data policy first. Confirm that company policy and the customer contract allow customer data in ChatGPT, and on which plan. By default, OpenAI does not use content from Business, Enterprise, Edu or ChatGPT for Healthcare workspaces to train its models; for personal accounts it may, when "Improve the model for everyone" is on (OpenAI Memory help, 2026). For contact names and other personal data, the GDPR requires data to be "limited to what is necessary" for the purpose (GDPR Article 5(1)(c), 2016).
Clear customer details from memory. Open Settings, then Personalization, then Memory, delete every saved memory that names a customer, and delete the chats where you shared those details (OpenAI Memory help, 2026).
Create one project per customer. Under Project settings, set Memory to Project-only memory; the change may take a few hours to take effect (OpenAI Projects help, 2026). Project-only memory appears only when memory is on (Reference saved memories and, outside Enterprise, Reference chat history, where your account has separate controls), and on Business or Enterprise when the workspace allows memory (OpenAI Projects help, 2026).
Check every answer against the four checks in the table below before anything from it reaches the customer.
What does ChatGPT memory remember between chats?
ChatGPT memory remembers relevant preferences and details from past chats, saved memories, custom instructions, files in Library and content from connected apps, depending on plan and region (OpenAI Memory help, 2026). OpenAI states that "Memory does not retain every detail from every conversation" and that ChatGPT decides which available information is relevant to a response (OpenAI Memory help, 2026).
Memory works in two layers. Saved memories are details you asked ChatGPT to remember, or that it saved as useful context; they are stored separately from chat history and stay until deleted. Reference chat history lets ChatGPT draw on past conversations, and what it derives from them can change as ChatGPT updates what is most useful to remember. If you turn Reference chat history off, information remembered from past chats is scheduled for deletion within 30 days, but saved memories are not deleted (OpenAI Memory help, 2026).
The memory summary in settings is a high-level view and does not necessarily include every detail or source that memory can reference (OpenAI Memory help, 2026). For customer work, that makes memory a summary, not a record. It suits a standing preference such as "report values in SI units"; it does not suit the pH measured at the last visit or the revision of the datasheet the customer uses (illustrative examples).
Why is ChatGPT memory the wrong place for customer details?
ChatGPT memory is the wrong place for customer details because it is one summary shared across your chats, it keeps no exact record, and removing a detail takes several separate deletions. On plans other than Enterprise and Edu, when memory is enabled on an account it stays active across all chats, including chats in projects with default memory (OpenAI Projects help, 2026). This suggests that a detail saved while working on one customer can shape an answer about another, with nothing in the answer to show it.
Removal is spread over many places. To remove remembered information, OpenAI lists the memory summary, saved memories, regular and archived chats, files in Library and connected apps, and notes that deleting a chat alone does not necessarily delete a saved memory created from it (OpenAI Memory help, 2026). OpenAI may also keep logs of deleted saved memories for up to 30 days for safety and debugging (OpenAI Memory help, 2026).
Sensitive information can appear in memory if you share it with ChatGPT (OpenAI Memory help, 2026). Customer contacts' names, roles and phone numbers are personal data, which the GDPR defines as "any information relating to an identified or identifiable natural person" (GDPR Article 4(1), 2016), and a memory that serves every chat is hard to keep "limited to what is necessary" for one customer's purpose (GDPR Article 5(1)(c), 2016). Keep memory for how you like to work, and keep customers somewhere you can scope and delete.
How do you keep customer context in a ChatGPT project?
You keep customer context in a ChatGPT project by giving each customer its own project with project-only memory, project instructions and one customer brief file that you replace after every contact. With project-only memory, saved memories are not referenced in the project, and chats can reference other conversations in the same project only (OpenAI Projects help, 2026). Project instructions apply only within that project and override global custom instructions (OpenAI Projects help, 2026).
Write the customer brief as a short document with fixed headings:
Account and sites, with the equipment or lines you support
Contact roles, with no more personal data than the work needs
Products in use, each with the datasheet and SDS revision the customer holds
Last visit: date, what was measured and what was found
Open issues, each with its status
Commitments: what was agreed, by whom, and the due date
Add the current datasheets and SDSs next to the brief. A project holds up to 40 files on Business and Enterprise, 25 on Go and Plus and 5 on Free, and only 10 files can be uploaded at the same time (OpenAI Projects help, 2026). Uploading a file with the same name as an existing project file can add a second copy (OpenAI Projects help, 2026), so delete the old brief or revision before uploading the new one.
What context gets lost when you paste customer details by hand?
Pasting customer details by hand loses whatever you did not think to paste: earlier visit notes, open issues, the document revision the customer actually uses, and who agreed to what. Each gap changes the answer in a predictable way.
Without earlier visit notes, the answer treats a recurring problem as new and suggests checks that were already done. Without open issues, it can recommend a change that conflicts with a trial still running at the site. Without the customer's document revision, it quotes values from whichever revision you pasted, or from general knowledge. Without the record of commitments, a reply can promise something a colleague already ruled out, or forget a date the customer was given.
Pasting also leaves the details behind. A pasted chat stays in chat history, and when memory is on, past chats are one of the sources ChatGPT can draw on for later responses (OpenAI Memory help, 2026). The customer brief in a project replaces both problems with one document you control and update.
The method is the same with any tool: decide which records a customer question needs, keep them in one place per customer, and update them after each contact. Lubinpla reads the record on the screen you are working in and lets you name members, projects, work logs, tasks and documents with @, so the customer's context comes from the record instead of a paste.
How do you check that ChatGPT used the right customer context?
You check that ChatGPT used the right customer context by reading each answer against four points before it reaches the customer: the customer named, the last visit, the document revision and stray data from other customers. When ChatGPT uses personal context, Sources may appear below the response, but they may not show every factor that shaped it (OpenAI Memory help, 2026). The table below decides whether an answer can be used or must be corrected first:
Check | Pass | Fail and what to do |
|---|---|---|
Customer named | The answer uses the customer and site from the project brief | Wrong or missing customer: confirm you are in that customer's project, then ask again |
Last visit included | The answer reflects the latest visit date and findings in the brief | Older visit used: replace the brief, delete the old copy, ask again |
Document revision | Values match the current datasheet or SDS revision in the project files | Old revision quoted: delete the superseded file and check the value in the document itself |
No other customer's data | No product, site, contact or value from another account | Stray data: check that project-only memory is on and that the detail is not a saved memory, then delete it |
The fourth row matters most: one detail from another account in a reply is a disclosure, not only an error. Even when all four checks pass, the answer can still contain confidently stated false content, which NIST names confabulation (NIST AI 600-1, 2024), so values still get checked against the document.
Frequently asked questions
Does Temporary Chat stop ChatGPT from remembering customer details?
Temporary chats do not create or update memories, and choosing Unpersonalized before you start keeps existing memories and custom instructions out; the choice cannot be changed after the conversation starts (OpenAI Memory help, 2026). Saving a temporary chat converts it to a regular chat that follows your personalization settings. It suits a one-off question, not context you want to keep.
Can teammates see my memories through a shared customer project?
No. Shared projects automatically use project-only memory and cannot access an individual member's memories or context from outside the project (OpenAI Projects help, 2026). Members do see the project's chats, files and instructions, so put in a shared project only what every member may see.
Does deleting a customer project delete its files?
Deleting a project permanently removes its chats and instructions, along with files stored only in the project; files saved separately in Library are not deleted (OpenAI Projects help, 2026). When a customer relationship ends, delete the project and check Library for copies.