Summarize an Email Thread with AI: 15 Copy-Paste Prompts

How do you summarize this thread with AI?#

Paste the whole thread, oldest message first, under a prompt that names exactly what you want back: a one-sentence TL;DR, five bullets, or action items with owners and dates. Add a rule that the model must use only facts in the thread and write None rather than guess. Copy the prompt below and swap in your thread.

It is the four-header brief, the most useful default because it matches what you actually ask about a thread: what is it about, what was decided, who owes what, and what is still open. If you only need a quick triage, jump to the one-line TL;DR further down.

Can Gmail or Outlook summarize a thread for you?#

Yes, both have a built-in button, and it is worth trying first. Gmail shows a summary card at the top of a long thread: open the thread and click "Summarize this email". Google’s help page lists it for work, school and personal Google accounts, with the "Smart features" settings turned on and in supported languages only. It also warns that Gemini can be inaccurate.

Outlook has "Summary by Copilot", a box at the top of the conversation with a Summarize button (new Outlook, classic Outlook, Outlook on the web, the latest Outlook for Mac, and a Copilot icon on the mobile apps). Whether you see it depends on your Microsoft 365 setup. Microsoft announced a 2025 rollout of the thread Summarize button to users with Copilot Chat pinned even without a full Microsoft 365 Copilot licence, and admins can restrict it. Microsoft’s help page does not say whether Gmail or other non-Microsoft accounts added to Outlook are supported, so test one before you rely on it.

So when does a pasted prompt still win? When the built-in button is missing on your plan or account, when you read mail across several providers, or when you want control over the output: owners and dates in fixed fields, a decisions log, a version written for your manager. The built-ins give you one generic summary; a prompt gives you the shape you asked for. For a comparison of tools that summarize long threads without any pasting, see AI tools that summarize long threads automatically.

Which summary should you ask for?#

Match the output to what you are about to do with the thread. Each row links to a prompt below.

How do you paste an email thread into an AI safely?#

Get the input right first, because most bad summaries are bad inputs. Expand the quoted history ("show trimmed content") before you copy, or the model summarizes only the top reply and presents it as the whole story. Paste oldest-first so it reads the conversation as it happened; if you can only get newest-first, add one line saying so. Keep sender names and dates, which are the raw material for who owes what, and delete signatures, disclaimers and unsubscribe footers.

Then treat the paste as a disclosure decision. You are sending colleagues’ words, customer names and contract terms to a third-party service. Privacy settings differ by product and plan, so check the chatbot’s data controls for training and history before you paste anything sensitive, redact names and figures you do not need, and at work use your company-approved AI tool rather than a personal account. Never paste regulated or confidential material into a tool your employer has not approved.

• Expand the full thread before copying so the quoted history is included.
• Paste oldest-first, or tell the model the order in one line.
• Keep sender names and dates; strip signatures, disclaimers and footers.
• Redact names, numbers or clauses you would not want stored, and check the chatbot’s training and history settings first.
• At work, use your company-approved AI tool.

What prompts work in Copilot for Outlook and Gemini in Gmail?#

Inside Gmail or Outlook the thread is already in context, so you skip the paste and ask a short question. These two are the ones people search for most, and they work as written in a chat pane next to the message. Treat the answer the way you would any summary and check the dates against the thread.

For Copilot in Outlook, ask for action items by naming the fields you want, since a bare "summarize" returns a generic recap.

What is the best prompt for a one-line TL;DR?#

The TL;DR answers "do I need to care about this?" when you are triaging a stack of threads. Constrain it hard: one sentence, a word cap, and the bottom line first. Without the cap, models drift into three-sentence summaries that defeat the purpose. A strong output states the state and the ask in one breath: "Vendor agreed to the new pricing but needs your sign-off on the contract redlines by Friday." A weak one, "The thread discusses pricing," names the topic and tells you nothing.

How do you summarize several threads at once?#

When you have ten threads to clear, use one chat, not ten. Most chatbots accept several threads in a single message if you delimit them clearly, and you get a scannable list in under a minute. It is a triage tool, not a full summary: it drops detail on purpose. Once a thread passes triage and you need to act, move to the bullets or action-item prompts.

How do you get a clean bullet-point summary?#

When a thread does need you, the bullet recap is the workhorse: long enough to carry the substance, short enough to read in fifteen seconds. Two levers matter: cap the count so the model has to prioritize, and demand that every bullet be a complete, standalone fact so you do not get fragments like "discussed the budget". Bullets beat paragraphs for a recap because they strip out the connective prose that made the thread slow to read.

How do you extract action items with owners and due dates?#

This is the most valuable thing AI does to a thread, because commitments are what threads bury. "Maria will send the revised deck by Wednesday" can sit casually in the eleventh reply, and if you miss it the project slips. Two rules make extraction reliable. Force owner and due date into explicit fields and tell the model to write "Unassigned" or "No date" when the thread does not say, because a guessed owner looks authoritative and sends the wrong person chasing the wrong task. And separate firm commitments from vague intentions: "I’ll look into it" is not "I’ll send it Friday".

Use the flat list when you are moving items into a to-do app, and the table when you want to review the whole set and spot gaps, such as three tasks with no owner. For more on this, see our dedicated action-item prompts.

Which is better for action items, a list or a table?#

A table adds two columns a list usually omits, priority and dependency, which are the next questions once you can see the whole set: what is urgent, and what is blocked. Tell the model to infer priority only from explicit signals in the thread ("urgent", "by end of day", "this is blocking the launch") and leave it blank otherwise.

How do you build a decisions log and surface open questions?#

A thread is mostly discussion, and discussion is not decision. The key instruction is to separate a decision from a proposal: "We should probably go with option B" is a leaning, "Confirmed, we’re going with option B" is a decision. Capture who made the call and why, because a decision without its reasoning is the one that gets relitigated six months later. Ask the model to flag anything that looks agreed but was never explicitly confirmed, so the ambiguity is surfaced rather than smoothed over.

How do you find the open questions and who owes what?#

Open questions are the inverse of the decisions log: everything raised and never resolved. They quietly stall projects, like the "can you confirm the budget?" that scrolled off the screen. The second half, who owes what, tells you whether the ball is in your court. If three people are waiting on a number only you have, you want to know before you close the tab. When you are about to reply, the sharper version is to pull the exact questions you must answer.

What about catching up, hard facts, conflicts and stakeholders?#

Four more prompts are worth keeping on hand. After time away, ask for the story, not the state: a flat summary says where things stand but not how they got there, and the closing rule surfaces the message sent to you directly. When the details are the point, extract dates and numbers with a rule to show both values when something changed, so a stale date does not slip through. For a thread that has gone sideways, lay out the disagreement neutrally before you reply; for the negotiation case, see summarizing a negotiation thread. And when you are forwarding the gist to one person, name them and their concern so the model drops everything else.

How do you summarize a very long thread that won't fit?#

Most threads fit comfortably in a modern chatbot, so for everyday conversations paste the whole thing and ask once. The problem is the months-long mega-thread or the reply-all that grew to two hundred messages. When it is too big to paste, or so big that the important sentences get diluted, summarize in chunks, carry the running state forward, and merge. Trim first: the repeated quoted blocks under every reply are the biggest offender, and deleting them can cut the length dramatically without losing a fact.

1. 1

   Trim the dead weight

   Delete repeated quoted history, signatures, disclaimers and footers before you split anything.

2. 2

   Split into chronological chunks

   Break the thread into ordered chunks that each fit comfortably, oldest first. Keep each message whole and label the chunks 1, 2, 3.

3. 3

   Summarize each chunk, carrying context forward

   Run the per-chunk prompt in order, with the running decisions and open questions from earlier chunks pasted above, so each pass keeps the state of the conversation.

4. 4

   Merge the chunk summaries

   Paste all the chunk summaries together and run the merge prompt, which reconciles items across chunks into one summary.

5. 5

   Spot-check against the source

   Open the original thread and verify the dates, numbers and owners. Multi-pass summaries are where small errors compound.

What does the final merge prompt look like?#

Per-chunk summaries are fragments until you reconcile them. The merge pass is what turns a question raised in chunk 1 and answered in chunk 3 into one clean decision. Skip it and you have summarized the pieces, not the thread.

How do you turn a summary into your next action?#

A summary is the means, not the end. Once the thread is in the conversation, keep going: the summary tells you what the thread needs, and the next prompt produces it. The most common chain is summarize, then reply. Run the prompt below right after a summary prompt and the draft is grounded in the actual exchange rather than a template. Two other chains are worth knowing: turn each action item into a calendar entry, or, for a thread full of options, ask for the trade-offs and a recommendation, with the caveat that you make the call.

What mistakes quietly ruin AI thread summaries?#

The failures are quiet, plausible-looking errors. Most trace back to a few causes, and the fixes are the rules you have already seen in the prompts above. A smooth, well-formatted summary feels trustworthy, which is exactly why a wrong detail is dangerous: it looks like the real facts.

• Hallucinated facts: an invented owner, date or decision. Force "facts only", write None or Unassigned instead of guessing, and spot-check any load-bearing number.
• Lost nuance: a tentative "maybe" flattened into a firm "yes", or an "if the budget allows" dropped. Separate settled from proposed, and read the source when stakes are high, as with a contract or a personnel matter.
• Incomplete input: only the top message copied, or the thread pasted newest-first without saying so. Most "the AI got it wrong" is really scrambled input.
• Trusting one pass on a huge thread: smooth is not the same as accurate when the content is diluted. Chunk it.
• Forgetting the paste is a disclosure: check the tool’s data controls before pasting anything confidential.

How does AI Emaily summarize threads inside your inbox?#

The prompts above make a paste-based workflow as good as it can be, but the loop itself has friction: open the thread, expand it, copy, switch tabs, paste, and then carry the answer back to your inbox. AI Emaily is an AI-native email client that does the summarizing where the thread already lives, so there is nothing to copy out.

In Manual mode a thread is summarized when you click Summarize with AI; in Copilot and Autopilot, summaries are generated for you, and you can ask the AI about a single email or a group. Copilot drafts replies for you to approve before anything is sent, in the voice you set in your Personal Context plus per-client profiles. Autopilot handles routine mail within your rules, with undo and an audit trail. It works across Gmail, Outlook and Microsoft 365, iCloud and IMAP providers, and AI Emaily does not train models on your mail.

Try it with a 7-day free trial on Pro or Autopilot; Pro is $12.42 per month billed annually or $14.99 monthly, and the pricing page has the current details.