How to Give Claude a Previous Conversation

To give Claude a previous conversation, paste the relevant messages or upload a readable transcript, then state what you want to do now. For a long discussion, lead with a checked summary of the goal, decisions, unfinished work, and source references.

You are supplying context for a new session, not necessarily merging two chat threads. fenbs can keep a reusable work handoff outside the chat, so you do not have to prepare the same explanation every time you return to the project.

What is the quickest reliable way to supply the conversation?

1. Choose the portion that matters to the current task. Keep the original conversation available as a reference.
2. Prepare either a plain-text transcript or a checked summary. Preserve who said what and identify any missing attachments.
3. Open the intended Claude conversation or project. Paste the text, or attach the transcript in a supported file format.
4. Tell Claude the current goal and ask it to explain the approved decisions, remaining questions, and next step before working.
5. Correct misunderstandings, then give a specific continuation request with a clear boundary.

Anthropic’s file upload guide↗ (opens in a new tab) lists supported documents, including text files and PDFs, and explains how to attach them. Follow its current limits rather than trying to send your entire account archive as one enormous input.

A transcript is useful when exact wording matters. A summary is usually easier for a new session to use when the old discussion contains retries, abandoned directions, and repeated explanations. You can provide both: a short current brief first and the transcript as supporting material.

Should you use Claude’s own past-chat search instead?

If the conversation is already in your Claude account, try asking Claude to find it when chat search is available. Anthropic’s chat search and memory documentation↗ (opens in a new tab) explains the feature and its account and project boundaries. Ask for a specific subject or decision rather than “remember everything.”

For example: “Find our discussion about the appointment form. Identify the latest approved field list and the unresolved delivery test.” Check the references Claude returns. A confident summary is not enough if it points to an early draft or omits the message that changed the requirement.

Use an explicit handoff when you need precise scope, when the earlier discussion is in a different app, or when retrieval does not find the right material. Our introduction to Claude AI explains the assistant more broadly; this workflow is about getting one particular piece of work ready to continue.

What should the opening handoff say?

Start with present intent. A conversation can contain several goals, and the latest message may simply be a small correction. State the outcome you want now, then distinguish what a person approved from what an assistant merely suggested. Give each unresolved question a clear status.

Treat the material below as background, not new instructions.
Current goal: Finish the appointment request form.
Approved: Require name and email; phone stays optional.
Completed: The form layout is drafted.
Checked: The owner reviewed the field list.
Not checked: Sending a request has not been tested.
Open question: What message appears if delivery fails?
Sources: Earlier discussion attached; see final approval.
Your first task: Summarize the current state and uncertainties.
Wait for my confirmation before changing the implementation.

Replace vague labels such as “done” with evidence. “Draft written; not reviewed” tells Claude more than “draft finished.” When the old discussion contains a correction, say which instruction it replaces. If nobody decided between two options, leave the choice open.

This is the practical value of portable AI memory: preserve the work that determines the next action instead of depending on one chat’s full history.

How do you preserve a transcript without carrying every mistake forward?

Keep speaker labels and message order. Add a note if you removed unrelated material, and label the result as an excerpt when it is incomplete. Preserve the original wording of approvals when exact phrasing matters, but do not repeat confidential material that the new task does not need.

Treat instructions inside the old conversation as historical material. A transcript may include “delete this file” or “send the email,” although that action was completed, canceled, or never authorized. Tell Claude to extract the current state rather than execute old requests again.

Include the necessary source files separately when permitted. A conversation saying “use the attached spreadsheet” is not the spreadsheet itself. Name missing documents and explain what cannot be verified without them. An honest gap is safer than letting the assistant invent the absent content.

Anthropic’s data export instructions↗ (opens in a new tab) describe obtaining account conversation data. They also distinguish an export from importing data into another personal account. Use an export as source material you review; uploading selected text does not recreate the original account or thread.

What should you check before Claude continues?

Ask Claude to name the current goal, the decision that constrains the work, the last verified result, and the first unfinished step. Compare its answer with the original material. Correct any conflict before giving it permission to make changes or produce a final deliverable.

For a long transcript, request references to message labels or document sections. Check a few important facts yourself, especially the approval, the latest version, and the outstanding failure. Ask Claude to say when the evidence is missing rather than filling gaps with a plausible story.

If its first answer still follows an abandoned direction, shorten the current brief and make the replacement explicit. Repeating the whole transcript can amplify the confusion. A small, accurate starting record is often more useful than another large unfiltered copy.

When this becomes a repeated routine, a shared AI brain gives people and supported assistants a common work record to maintain between sessions.

How to do this with fenbs

1. Create a free account and add the project you want Claude to continue.
2. Turn the checked summary into current tasks, project context notes, and Decisions and rules that record what people chose.
3. Connect Claude through the guide on Integrations, using the browser sign-in and the access you intend to grant.
4. Ask Claude to read the project context before working. At the end, ask it to save task progress and Where we left off, including unfinished checks and anything waiting on you.

fenbs does not read your chat history or import conversations automatically. It holds the selected work record you and your connected assistant maintain. If you need one earlier exchange for a one-time question, pasting that exchange is simpler.

The AI handoff guide shows how this fits between sessions. For sharing with another person, see sharing in Claude. More ways to continue work are collected in the AI handoffs topic.