Getting the Most Out Of Semblian AI Chat: A Complete Prompting Guide

Chat examples:

Weak ❌

Stronger ✅

Find what was said across your meetings

One of Semblian's most powerful capabilities is recall across all of your meetings, not just one. You don't need to remember the meeting title or date — Semblian searches your history for you.

Find what someone said about a topic:

Across any meeting in the last 90 days, what did [Person] say about [Topic]? Quote the line and tell me the meeting date.

Find a specific moment or commitment:

In a meeting with [Person] from [Customer], they said if we delivered the dashboards we discussed, it would prove the value to them. Which meeting was that, and on what date?

Pull every mention of a topic:

Tell me everything that has been discussed about [Topic / Customer / Project] in my meetings. List each mention chronologically with the meeting date and a 1-line summary.

Find a meeting by topic, not by title:

We had a strategy meeting (or a couple of them) where we discussed [Topic]. Can you find those meetings for me and list the dates?

Stay focused on a single meeting:

Answer based only on the meeting from [Date]. [Your question.]

Always include a time window ("in the last 30 days") and ask for citations ("quote the line and link the meeting"). It makes the answer faster and easier to verify.

Generate reports, summaries, and documents

Most documents you write after a meeting follow a template. Semblian fills the template for you in seconds — as long as you tell it which template to use.

Status report:

Generate a [weekly / monthly] status report based on my status meetings in [time range]. Sections: progress this period, decisions made, blockers, next milestones with dates. Tone: professional, shareable with stakeholders.

Executive brief:

Compile my meetings from the last [N] days into an executive brief. Include: top decisions made, open commitments (who → what → when), blockers, customer signal worth my attention. Bullets only. Keep it under one page.

Eisenhower matrix as a status doc:

Based only on the added meetings, generate an Eisenhower Matrix. Add context and a status update at the top. At the bottom, include important dates and the timeline. This is going to stakeholders, so keep the tone polished.

A specialized format (SOAP, OKR, SOW, hiring debrief):

Turn this meeting into [SOAP notes / OKR review / 1:1 record / hiring debrief]. Use the strict structure for that format. Remove duplicate details. Where the meeting didn't cover a section, write "Not discussed."

Statement of work:

Answer based only on the given meeting and create a comprehensive draft of the statement of work to be presented to [Client]. Include: scope, out-of-scope, deliverables, timeline, milestones, assumptions, dependencies. Flag anything missing as [TBD] rather than inventing it.

One-pager:

Create a one-pager on [product / feature / topic] tailored for a [persona] audience. Lead with the 3 outcomes they care about most, then 3 supporting bullets, then a short closer. Tone: confident, no jargon.

Tell AI chat the audience and length up front. Add "don't invent anything that wasn't in the meeting — flag missing info as [TBD]" when you need a strict, evidence-based document.

Draft emails and follow-ups

Post-call emails are usually the most time-consuming part of the day. Semblian multi-meeting chat drafts them from what was actually said.

Standard post-call follow-up:

Based on my call with [Account] on [Date], draft a follow-up email to [Contact]. Include: a 1-line recap of what we agreed on, 2–3 next steps with owners and dates, and a clear ask for the next meeting.

Tone: warm, professional, short.

Personalized customer follow-up:

Draft a personalized follow-up email summarizing decisions and next steps from my last meeting with [Customer]. Mention the points they emphasized so it sounds like I was listening.

Sensitive or high-stakes reply:

Draft an email to [Person] in reply to their message that [summary of their position]. Make clear that [your position], using context from our meetings. Tone: respectful, firm, factual. No jargon.

Internal handoff (e.g. AE → CSM after a sale):

Based on my discovery and contracting calls with [Customer], draft a handoff email for my CSM team. Include: customer goals, decision makers, technical environment, agreed-upon use cases, risks, and next milestones.

Always name the tone ("warm and professional," "direct," "apologetic but firm"). Specify length ("under 100 words," "3 short paragraphs"). Semblian handles non-English drafts — end the prompt with "Output in [language]" if needed.

Track action items, decisions, and commitments

Semblian AI chat is great at keeping everyone honest — yourself included. Three patterns cover most of what you'll need.

Action items as a table (the most popular use case in our data):

From [meeting / set of meetings], pull all action items into a table. 

Columns: Action, Owner, Due date, Source meeting, Status (open / done). Sort by due date.

Decision log (different from action items — Semblian will surface decisions separately when you ask):

List every decision made in [project meetings / time range].

Columns: Decision, Date, Decider, Context (1 line), Reversible?

Pull only decisions, not discussions.

What someone owes you:

List every commitment [Person] has made to me across all our meetings. Group by status: done, in progress, slipped, no update.

Cite the meeting date for each.

What you owe someone else (great before a 1:1 or customer call):

List everything I've committed to [Person] across all our meetings.

For each: what I committed to, when I said it, and whether I've delivered it.

Multi-topic status sweep (a great Monday-morning prompt):

What's the latest on:

- [Topic 1]

- [Topic 2]

- [Topic 3]

For each, give me 3 bullets: where we are, who owns it, what's blocking. Pull from any meeting in the last [N] days.

Always ask for a table when you need owners + due dates + status. It exports cleanly to Excel, Notion, or any project tool.

Get ready for an upcoming meeting

Three prompts cover almost every prep scenario.

60-second daily plan:

Build my plan for today. From my upcoming meetings and any open commitments, give me: 
1) top 3 priorities
2) prep needed before each meeting (1 line each)
3) follow-ups I owe from yesterday.

Bullets only.

Pre-call prep brief:

I have a call with [Contact / Account] in [time]. From any prior meetings and context, give me: 
1) what we last discussed
2) the 3 questions I should ask
3) the 1 outcome I should drive in this call.

Full account or project catch-up (for picking up an account, joining a project late, prepping for a renewal):

I'm jumping into [Account / Project]. From the last 90 days of meetings, give me: 
1) timeline of what we've discussed
2) open issues / risks
3) recent commitments from us
4) recent commitments from them
5) anything that's changed in tone or stakeholders.

Renewal or QBR prep:

Build a renewal prep brief for [Account]. Across the last 180 days of meetings: what value moments did we create, what concerns has the customer raised, who are the champions vs detractors, and what's the cleanest 3-bullet renewal narrative I can lead with?

Discovery call prep (sales):

Using my [internal sync / prep] meeting on [Topic], build a discovery prep doc for my call with [Prospect]. Sections: what we know about them, top 3 hypotheses to test, suggested discovery questions, likely objections + responses, recommended next step.

Match the depth to the stakes. A 9 a.m. internal standup needs 60 seconds of prep. A renewal call needs the full brief.

Compare options, vendors, or ideas

When you're evaluating tools, vendors, candidates, or product directions, Semblian produces side-by-side comparisons grounded in your meetings, with web context where it helps.

Standard comparison table:

Create a side-by-side comparison table of [Option A], [Option B], and [Option C]. Use the information from my meetings and combine it with information from the web. Make sure it is well-structured and clear. Do not add meeting titles as references.

Comparison with your own criteria as rows (the most useful version):

Build a comparison table of [Option A], [Option B], [Option C].

Rows: [Your criteria — e.g. price, integrations, support, time to implement, security posture]. Columns: each option. Pull what's been said in my meetings; supplement with the web where my meetings are silent. Keep it neutral.

Pros / cons summary:

For each of [Option A], [Option B], [Option C], summarize the pros and cons that came up in my meetings. Quote the line that supports each pro / con and cite the meeting.

Always specify the criteria as rows. Without them, the table will be generic. With them, it becomes a decision tool.

When a prompt doesn't work

If the answer feels generic, off-topic, or just not what you wanted, walk through this checklist before re-prompting. Fixing one of these almost always rescues the chat.

The 6-question diagnostic

1. Did you tell AI chat which meetings to use? Add an anchor: "based only on this meeting," "across the last 30 days," or "in any meeting from Q1."
2. Did you use a specific verb? Draft, list, compare, build a table, generate a one-pager — these work much better than "give me" or "tell me about."
3. Did you say who the output is for? "For my board," "for an internal team," or "as an email to a customer."
4. Did you name the format? Bullets, table, email, one-pager, summary. Naming the format is the #1 fix.
5. Was the goal one thing — not three? Two clean prompts almost always beat one tangled prompt. Split it up.
6. Did the prompt require AI chat to invent facts? If you're asking about something that wasn't in any meeting, anchor to what was actually said.

Three quick re-frames to try

If your prompt didn't work, try one of these as a follow-up message in the same chat:

Try anchoring it: redo the answer based only on the meeting from [Date], or across the last 30 days.

Output that as a table with columns: [Column 1], [Column 2], [Column 3].

Split this: first, just give me the facts retrieved from my meetings. Then in a second message, I'll ask you to draft.

When to start a new chat vs. refine the current one

• Refine the current chat when you're tweaking format, length, or tone of the same answer.
• Start a new chat when you're switching topics, switching meetings, or the conversation has gone in a confusing direction. A clean prompt in a clean chat almost always beats a long thread.

Still stuck?

If a prompt still isn't working after applying the checklist, share the prompt and the response with our support team and we'll help you tune it.