# How to hand off work to a colleague when you both use Claude or ChatGPT A practical guide to handing work to a colleague when your team uses Claude or ChatGPT: what to pass on, and how to keep the context and files intact. ## The short answer Your colleague does not need the whole chat to take over your work. They need what it adds up to: the task, the deadline, the facts, the decisions, the final files and the open questions. The simplest route is to ask your assistant for a structured handoff summary and send it, with the files, to one person with a deadline. Below you will find a seven-point checklist, a ready-made prompt and three methods that work without any extra tool. At the end we show how to do the same thing in one sentence with Colle. ## Why the context stays in the chat When you work with Claude or ChatGPT, a large part of the work happens in the chat. You upload the contract, talk through what the client wants and rewrite the proposal a few times until the price is right. An hour later the chat holds everything: what was decided and why. But you are the only one who can see it. When the task moves to a colleague, the usual round begins: an email saying "over to you", a hunt for the right file among several versions, a question about what the deadline was, and one more meeting to explain it all again. Sharing the chat itself helps less than you might think. Claude's Share feature, for example, creates a snapshot of the conversation: your colleague sees the chat up to the moment you shared it, and any later messages stay private. Attached files are not included, tool call data stays private, and on Team and Enterprise plans the link only works inside your organisation. Source: [Claude Help Center: sharing chats](https://support.claude.com/en/articles/10593882), last verified 2026-10-15. Even with the full chat, your colleague has to read it from the start and work out for themselves what still applies. Decisions shift over a long conversation: the price agreed in message twenty may have changed by message forty. So a handoff should not be a copy of the chat. It should be its conclusion. ## What you must pass on: 7 points Before you hand off the work, check that your colleague will get an answer to each of these seven questions. If one is left open, they will most likely ring you about it. 1. **What exactly needs doing.** One sentence with a verb: "prepare the proposal", "check section 4 of the contract", not "have a look at this". 2. **By when.** A specific date, such as 19 October 2026, not "next week". If the deadline depends on someone else, say so. 3. **Which facts are already known.** Amounts, quantities, contract numbers, contact people. Put each fact on its own line so nobody has to dig for it in the text. 4. **What has been decided, and why.** If you agreed a discount with the client, your colleague needs to know, or they may reopen the negotiation. A short reason stops the decision being changed for no good reason. 5. **Which files are final.** Only the versions to work from, with clear names. If files were left in the chat with your assistant, save them separately and attach them. 6. **What is still unclear.** What you do not know yet and what you are waiting for from the client or colleagues. This is the point people forget most often, yet it saves the most questions. 7. **Who to check with.** Who makes the final decision, who to call if something is unclear, and whether it is fine to call you. ## Three methods without a dedicated tool ### 1. A summary prompt The simplest method is to ask the same assistant you have been working with to write a handoff summary. It can see the whole chat, so it will gather the facts and decisions faster than you can. The key is to give it a clear structure and tell it not to guess. Copy this text into the end of your chat: Write a handoff summary of this chat for a colleague who has not seen it. Use only what was actually discussed in this chat. If something is missing, write "unknown". Do not guess. 1. The task in one sentence: what my colleague needs to do. 2. Deadline: the date (YYYY-MM-DD) and, if discussed, the time. 3. Facts: amounts, terms, names, contract numbers, each on its own line. 4. Decisions made, and briefly why. 5. What has already been done. 6. Open questions: what we do not know yet and what we are waiting for. 7. Files: the name and which version is final. 8. Who to check with and who signs off. If a decision changed during the chat, give only the final one. Keep it short, no introduction, 250 words at most. Read the summary before you send it: your assistant does not know what was agreed on the phone or in a meeting. Then copy it into an email or message and attach the files. **Pros:** works with any assistant and any plan, and takes a few minutes. **Cons:** you still have to attach the files by hand, and the handoff lives in email or a chat app, so you cannot see whether your colleague has picked up the work or finished it. ### 2. A shared project Both Claude and ChatGPT have projects: places that keep chats, files and instructions together. In a shared ChatGPT project, a member with chat access can see and use the project's chats, files and instructions, and on Business, Enterprise and Edu plans a project can hold up to 40 files. Source: [OpenAI Help Center: What is a project in ChatGPT](https://help.openai.com/en/articles/10169521-what-is-a-project-in-chatgpt), last verified 2026-10-15. This works well when several people work on the same client or topic all the time. **Pros:** files and instructions in one place, and your colleague can carry on in the same context. **Cons:** you both have to use the same assistant and the same workspace, and a project has no "who" and no "by when". It is a shared shelf, not a task for a specific person. Before you set up a shared project, think about whether every member should see every chat in it. ### 3. A handover document The third method is a shared document wherever your team already works: Google Drive, SharePoint or similar. You paste in the summary your assistant wrote, add links to the files and share it with your colleague. **Pros:** it does not depend on the assistant, you can add to it, and it stays with the team after the handoff. **Cons:** access has to be managed by hand, the document goes out of date quickly, and your colleague has to load it into their own assistant separately before they can carry on. Whichever method you choose, stick to one rule: one handoff, one person, one deadline. A long email to three colleagues covering ten jobs often ends with each of them assuming someone else is doing it. ## How it works with Colle Colle turns a handoff into one sentence in the assistant you already use. You add the Colle connector to Claude or ChatGPT, and handing off becomes a simple request to your assistant: > Pass the draft contract to Rachel: price €12,400, sign by 20 October 2026. Your assistant finds Rachel, writes a summary, records the facts (the price and the deadline) and attaches the file. Rachel gets a notification in her browser or on her phone: who sent it, what it is and when it is due. She accepts the work with one click, asks her own assistant "what did I get?" and finds the summary, the facts and the files. Rachel does not have to use the same assistant: a handoff can travel from Claude to ChatGPT and back. A few more examples of how it sounds: - "Give Tom the client brief and our meeting notes." - "Send Mark this invoice, to be paid by Friday." - "Hand this complaint to James with the order number." - "Send to Rachel and Tom, contract review by 19 October." Colle finds people by name the way people actually say them. That matters in languages where names change their ending in a sentence. In Lithuanian, "send it to Jonas" becomes "nusiųsk Jonui", and the surname Petraitis becomes "Petraičiui"; Colle still finds the right colleague. If there are two Toms in the company, your assistant asks which one you mean before it sends anything. One handoff can go to up to five people, and original files up to 25 MB are uploaded through a secure link. A handoff you have sent has a status. You can see when your colleague accepted it. They can send the work back with a reason, such as "the signed page is missing", or mark it as done. If the deadline passes, you both see that the work is overdue. Sent it to the wrong person? Cancel the handoff. All of this stays in the case, together with the history of the facts. Work sent to a person is visible only to the sender and the recipients. If you send work to a team space, every member except you gets a notification, and one of them takes it. Find out more in [how Colle works](/how-it-works/) and on our [security](/security/) page. For a fuller list of what to cover when work changes hands, see our [project handover checklist](/blog/project-handover-checklist/). [Join the Colle pilot](https://www.trycolle.com/#pilot) ## Frequently asked questions ### Does my colleague need to use the same assistant? Without a dedicated tool, usually yes, if you want to share a project or a chat. With Colle, no: handoffs travel between Claude and ChatGPT. Sending from ChatGPT needs a ChatGPT Business or Enterprise plan. ### Is it enough to send a link to the chat? Rarely. A Claude share link shows the chat only up to the moment it was shared, and attached files are not included. Your colleague has to work out for themselves what still applies. A link is good for showing how you reached a conclusion, but not for handing over the work. ### What if some of the agreements were made on the phone? Add them to the summary yourself. Your assistant only knows what was in the chat, so always read the summary and fill in the gaps before you send it. ### How much does Colle cost? €15 per active person per month, or €12.50 per month when you pay yearly, excluding VAT. People who only receive notifications do not pay. The first 7 days are free, and you do not need a card. More on the [pricing page](/pricing/). --- Canonical page: https://www.trycolle.com/blog/hand-off-work-to-a-colleague-with-claude-and-chatgpt/ · Last updated: 2026-10-15