The prompt you choose determines the quality of the output. A vague "summarize this" produces a vague summary. A specific goal produces actionable results.
Here is a library of proven prompts for different WhatsApp recap scenarios.
How prompts work in ThreadRecap
ThreadRecap uses goal-based analysis. Instead of writing free-form prompts, you select a goal (Summary, Action Items, Meeting Recap, etc.) and the system builds the right prompt for your chat.
But understanding what each goal does — and when to use it — helps you get better results.
When you write a free-form prompt, the AI has no prior knowledge of WhatsApp's export format: the timestamp layout, the `<Media omitted>` placeholders, the system messages that appear when someone joins or leaves a group, or the way voice notes are referenced before transcription. ThreadRecap's goal system encodes all of that context before your chat content is even analysed. The result is that a one-click goal selection consistently outperforms a carefully written manual prompt because the underlying instruction set is purpose-built for WhatsApp data.
ThreadRecap supports exports containing 60,000 or more messages, which matters for long-running project groups or community chats that accumulate months of history. At that scale, a free-form "summarize this" instruction tends to produce a shallow, high-level output that glosses over important detail buried deep in the thread. The structured goals are calibrated to surface relevant signal even across very large exports.
General summary prompts
Goal: Summary
Best for: Getting a quick overview of any conversation.
What it produces:
Key topics discussed
Important decisions
Notable moments
Overall tone and trajectory
Use this when you just need to catch up on a chat you have been ignoring.
When Summary is the right starting point
The Summary goal works well as a first pass on any export, regardless of size or context. If you are unsure which of the more specific goals applies, running Summary first gives you enough orientation to decide whether you then need Action Items, Decisions, or Topics. Think of it as triage rather than a final output.
For personal chats, Summary is usually the only goal you need. The other goals are designed around multi-participant, task-oriented conversations. A one-on-one conversation about travel plans or a family group catching up on weekend news does not benefit from decision logging or action item extraction in the same way a project group does.
Action item extraction
Goal: Action Items
Best for: Work chats where people make commitments.
What it produces:
Who committed to what
Deadlines mentioned
Dependencies between tasks
Items that are still unresolved
Use this after a planning conversation or project discussion.
What the Action Items goal actually looks for
The Action Items goal is designed to distinguish between a firm commitment and a vague suggestion. In WhatsApp chats, people often say things like "we should probably get that done by Friday" without any specific person accepting ownership. The goal flags those cases as unresolved rather than assigning them to the last person who spoke. That distinction matters when you are using the output to build a task list or update a project tracker.
The goal also surfaces dependencies between tasks. If one message says "once Ahmed sends the contract, Lisa will review it," the output will note that Lisa's task depends on Ahmed's prior action. This is the kind of structural detail that gets lost in a plain summary but is critical for project management purposes.
When a chat spans a long period, pairing the Action Items goal with a narrowed date range keeps the output relevant. A six-month project group will contain completed tasks from earlier phases alongside current commitments. Filtering to the last two or three weeks focuses the analysis on items that are still live.
Meeting recap
Goal: Meeting Recap
Best for: Work group chats that function as informal meetings.
What it produces:
Attendees (based on who spoke)
Agenda items (inferred from topics)
Decisions made
Action items with owners
Open questions
Use this to turn a messy group chat into professional minutes.
How Meeting Recap handles group chat exports
Many teams use WhatsApp groups as a lightweight alternative to video calls or email threads. The Meeting Recap goal is built specifically for this pattern. It infers attendees from participation, not from a formal attendee list, which means it accounts for the reality that some participants read and react without sending text messages. It also reconstructs agenda items from the sequence of topics raised, even when the conversation jumped between subjects.
Open questions are flagged separately from decisions. If someone asked a question that was never answered, or raised a concern that was acknowledged but not resolved, the Meeting Recap output surfaces it under open questions rather than folding it into the decisions list. That distinction is useful when you are preparing for a follow-up call and need to know exactly what still needs to be addressed.
Decisions and agreements
Goal: Decisions
Best for: Long threads where agreements get buried.
What it produces:
Every decision identified with context
Who agreed to what
Conditions or caveats mentioned
Unresolved disagreements
Use this before a follow-up meeting to review what was already decided.
The difference between a decision and a suggestion
One of the most common sources of confusion in WhatsApp work groups is that suggestions, tentative agreements, and firm decisions are all expressed in the same conversational register. "Let's go with option B" might be a firm decision or an opening proposal, depending on what follows. The Decisions goal analyses the surrounding context to determine whether a statement was confirmed, modified, or left open.
Conditions and caveats are captured alongside each decision. If the group agreed to proceed with a vendor "assuming the pricing comes back within budget," that condition is recorded rather than being collapsed into a simple "agreed to use vendor X." This makes the Decisions output more reliable as a reference document before a follow-up meeting, because it preserves the nuance that the original conversation contained.
Unresolved disagreements are also surfaced. If two people expressed opposing views and the thread moved on without resolution, the goal flags it so the follow-up meeting can address it directly rather than discovering the disagreement mid-discussion.
Topic extraction
Goal: Topics
Best for: Long chats that cover many subjects.
What it produces:
Distinct topics identified
Key points per topic
Participants involved in each topic
Resolution status
Use this when a 500-message chat covered everything from logistics to strategy.
How Topics handles overlapping and interleaved conversations
Group chats rarely stay on one subject for long. A discussion about a product launch might pivot to a budget question, return to launch logistics, then branch into a side conversation about a team member's availability. The Topics goal segments the chat into distinct subjects even when those subjects are interleaved across the conversation rather than appearing in clean sequential blocks.
Each topic entry includes the participants who contributed most to that thread, which is useful when you need to know who to follow up with on a specific subject. Resolution status tells you whether the topic reached a conclusion or was left open. A topic marked as unresolved is a prompt to either address it in the next meeting or open a separate thread where it can be handled properly.
Choosing the right goal
Situation
Best Goal
Catching up on a chat
Summary
After a work discussion
Action Items
Group planning session
Meeting Recap
Before a follow-up meeting
Decisions
Long multi-topic thread
Topics
Personal conversation review
Summary
Tips for better results
Use date ranges
If the chat spans months, narrow the date range to the period you care about. A focused input produces a focused output.
Filter participants in group chats
In large group chats, select only the key participants. This removes noise and focuses the analysis on the people who matter for your purpose.
Voice notes often contain the most important information — the decisions, the detailed explanations, the emotional context. Always include media in your export if voice notes are part of the conversation.
Run multiple goals
You can analyze the same chat with different goals. Run a Summary first to get the overview, then Action Items to extract tasks, then Decisions to log agreements.
Understanding what gets lost when voice notes are excluded
When you export a WhatsApp chat without media, voice notes appear in the text file as `<Media omitted>`. The AI sees only that placeholder, not the content of the recording. In many project groups, voice notes are where the substantive discussion happens: someone records a two-minute explanation that would take ten minutes to type, or records a decision verbally that never gets confirmed in writing. Excluding those recordings from the analysis produces an output that can be significantly incomplete. ThreadRecap uses Whisper for transcription, which means including media in the export brings voice note content into the analysis on the same footing as text messages.
When free-form prompts make sense
ThreadRecap's goal system handles most use cases. But if you need something very specific — like "extract all mentions of the Q3 budget" or "find every time someone mentioned the client deadline" — a general-purpose AI tool with a custom prompt might complement your ThreadRecap analysis.
The difference is that the chat analyzer handles the hard parts (parsing, voice notes, long chats) while you focus on what you want to learn.
Combining structured goals with targeted follow-up
A practical workflow for complex projects is to run the relevant structured goals first, then use the output as a reference when writing a targeted follow-up query. For example, run the Decisions goal to get a structured log of agreements, then search that output for a specific vendor name or budget figure rather than asking the AI to search the raw 60,000-message export for it. The structured output is smaller, cleaner, and easier to query precisely.
This approach also works across time. If a project group has been active for several months, you might run Topics on the full export to understand the shape of the conversation, then narrow the date range and run Action Items on the most recent four weeks to get a current task list. The two outputs together give you both historical context and current priorities without requiring a single prompt to do everything at once.