Automatic follow-up after a meeting in 60 minutes, not next week
The meeting ends and within the hour the client has a summary in their inbox: what was agreed, who does what and by when. A workflow in n8n or Make starts from the calendar, from a CRM note or from a transcript, an AI agent writes the text from the actual decisions, tasks go to Asana, ClickUp or Jira, and the deal stage in the CRM moves on its own.
- 60 minutesthe usual upper bound between the end of a meeting and a summary in the client's inbox, with the decisions, the tasks and the deadlines
- 3 meetingsa day per person. Above that line the summary from the last call of the day gets written after 18:00 or not at all, and above that line an automation starts to earn its place
- 2 to 6 weeksfrom workshop to a working workflow, tested on your own meetings rather than on samples
The problem
Why does follow-up go missing when everyone knows it matters?
A meeting ends at 15:40, the next one starts at 16:00 and two calls still have to go out in the afternoon. The notes sit in a phone, in a paper pad or only in someone's head. The client waits for confirmation of what was agreed and nothing arrives, not that day and not the next. A few days later nobody remembers who was supposed to send the quote and who was supposed to book the next slot.
The same thing happens in customer service. A consultant promises during a call to come back with an answer, and that promise lives exactly as long as one person remembers it. Salespeople understand what follow-up is worth. The problem is that the calendar leaves no gap in which to write it, so the quality of a summary depends on how tired someone is at the end of the day.
The cost of the delay is countable. In an audit of 2,241 companies published in Harvard Business Review the average response time to an inbound web enquiry was 42 hours, and 23 percent of companies never replied at all. Companies that made contact within an hour qualified a lead nearly 7 times more often than those that waited an hour longer. The study is about enquiries rather than meetings, but the mechanism is the same: every hour without contact leaves less of the conversation standing.
- The client waits two or three days for confirmation and talks to a competitor in the meantime
- The task "send the quote by Friday" exists only in one rep's head
- Before the next call the rep rebuilds the history of the account from memory
- The deal sits at the "after the meeting" stage because nobody clicked the status across
Why now
What changed, so that a summary can write itself?
Meeting transcription became standard. Tactiq, Otter, Fireflies and Fathom write down a call in Google Meet or Teams, and the end of an event in Google Calendar or Outlook is a signal a workflow picks up over the API a second after the meeting. Nobody has to wait for someone to sit down with their notes in the evening.
A language model, Claude or GPT, reads the notes or the transcript and writes a summary from the actual decisions: what was said, what the next steps are, who is responsible and by when. The difficulty sits in the prompt. An untuned model treats a loose suggestion exactly like an agreed decision, and that destroys trust in the summaries inside the first week. This is why we test the workflow on your own meetings before it goes out to the whole team.
The third change is the AI agent. A workflow will send a summary, but it will not answer a manager asking which of this week's follow-ups never got a reply from the client. BOND, our company AI agent, answers that from the CRM and the mailbox in 5 to 10 seconds, in a plain sentence, with no report from anyone on the team.
What changes
What does the team get after every meeting?
A summary with the client in 60 minutes
The workflow starts from the end of a calendar event, from a note saved in the CRM or from a finished transcript. The client gets the email the same day, while the conversation is still fresh on both sides.
Three triggers to choose from: Google Calendar or Outlook, a CRM note, a transcript from Tactiq, Otter, Fireflies or Fathom
Decisions, not "thank you for your time"
The AI agent reads the notes or the transcript and writes out what was said, what the next steps are, who is responsible and by when. A loose suggestion the other side never accepted does not get listed as a decision.
A prompt that separates a decision from a suggestion, checked on your meetings before it runs for the whole team
Tasks with a deadline and an owner
Every action point from the summary lands in Asana, ClickUp or Jira as a task with a person and a date. A commitment stops depending on one person's memory.
The task is created in the same run as the email, with nothing retyped by hand
The CRM updates itself
The deal moves to the next stage and the summary is written into the account history in HubSpot, Pipedrive, Salesforce or LiveSpace. Before the next call the rep opens the record instead of replaying the meeting from memory.
Account history in the CRM, not in a paper pad and not in someone's head
A person approves where it matters
We agree which summaries go out immediately and which wait for one click from the rep: a formal client, a meeting with several threads running, a contract in the background. The team changes the template and the tone of the message itself.
The approval rule is written down in the workshop, template edits need no ticket raised with us
Questions about the funnel in a plain sentence
Which follow-ups from this week got no reply. How many deals have been sitting after the meeting for longer than 10 days. Whose task is past its deadline. BOND answers from the CRM and the mailbox in 5 to 10 seconds.
An answer from company data, with no CRM login and no report from anyone on the team
And in your company
What if our meetings work differently?
Four sentences we hear most often, and how we answer them.
We do not record meetings, our reps take notes on their phones.
Then the trigger is a note in the CRM or a short form after the meeting with three questions: what was agreed, by whom, by when. The AI agent writes the summary from that. A transcript is one of three routes, not a condition. In the workshop we check which route the team will actually keep up after a month, because the one that looks better on paper is often not the one people follow.
Our clients cannot receive an email written by AI.
The client gets an email from the rep, from the rep's own mailbox, after the rep clicks send. The AI agent writes a draft from the decisions, the rep adjusts the tone and approves it. For some meetings you can switch approval off, for others keep it on. That decision is yours and it sits in a written rule, not in a tool's default.
We have two salespeople, they will manage on their own.
If each of them has two meetings a week, they will, and we say so in the consultation. An automation starts to make sense at around 3 meetings a day per person, because that is when the summary from the last call gets written after 18:00 or never. We come back to the subject when the calendar fills up, or when a client asks yet again whether the quote is coming at all.
The model will confuse a decision with a loose suggestion and the client will get nonsense.
That is a real risk and it is the main work in the prompt. This is why the workflow starts on a handful of real meetings with a person approving every send. We correct the prompt wherever the model loses the thread, most often on meetings where two topics run in parallel, and only then do we switch it on for the whole team.
- Calendars and meetings
- Google Calendar, Outlook, Google Meet, Microsoft Teams
- Transcription
- Tactiq, Otter, Fireflies, Fathom
- CRM
- HubSpot, Pipedrive, Salesforce, LiveSpace
- Tasks
- Asana, ClickUp, Jira, Notion
- Email and notifications
- Gmail in Google Workspace, Outlook in Microsoft 365, Slack
- Automation layer and models
- n8n, Make, Claude, GPT, custom API integrations
What a follow-up workflow is made of
The pieces that most often go into a first rollout. Which trigger to pick and how much of this to switch on at once is agreed in the workshop, because a company working from Google Meet transcripts needs a different workflow than one whose reps take notes after phone calls.
| Element | What starts it or what it does | When it makes sense |
|---|---|---|
| Meeting transcript | a finished transcript from Tactiq, Otter, Fireflies or Fathom | online meetings, conversations with several threads |
| End of a calendar event | an event in Google Calendar or Outlook finishes, the rep adds three sentences | meetings in person and phone calls |
| Note in the CRM | a new note on a deal in HubSpot, Pipedrive, Salesforce or LiveSpace | teams already writing in the CRM that do not want another tool |
| Summary from the AI agent | decisions, next steps, owners and deadlines, sent after approval or straight away | always, this is the core of the workflow |
| Tasks | action points go to Asana, ClickUp or Jira with a person and a date | when someone other than the rep owns the delivery |
| Deal stage in the CRM | the status moves on with a follow-up date attached | a funnel where you want to see how many deals are waiting for a reply |
| Silence reminder | no reply from the client after 5 working days creates a task for the rep | a dozen or more open threads per person at once |
For most clients a first rollout covers one trigger, the summary and the tasks. The CRM stage and the reminder go in once the team trusts the summaries.
How we work
Five steps from a note in a pad to a summary in the client's inbox
Every step ends with something you can judge before you decide on the next one.
Workshop
We take apart what happens after a meeting today: who writes the notes, where they go, how long it takes before a summary is sent and at which point something goes missing. In one company it is a note with no structure, in another it is unclear who owns what between the rep and technical support, in a third it is simply no time left after the last meeting of the day.
Workflow design
We choose the trigger: the end of a calendar event, a new note in the CRM or a finished transcript. We agree where the tasks go: Asana, ClickUp, Jira or straight into the CRM. We build on n8n or Make, depending on what you already work with. We write down the approval rule and what is kept from a transcript, what is deleted and after how long.
Pilot on real meetings
The workflow runs on a handful of meetings with a person approving every send, before it goes out to the whole team. We check the quality of the summaries and correct the prompt wherever the model loses the thread, most often on meetings with several topics at once. Error monitoring is configured as standard.
Production and training
We switch the workflow on for the whole team. Reps see what a summary looks like before it reaches the client and learn to edit the template: one tone for a formal client, another for a long-standing partner. We do not leave you with a black box.
Measurement at 30 and 90 days
We count how many summaries went out within the hour, how many needed an edit before sending, how many tasks were closed on time and how many deals moved off the "after the meeting" stage. We retune the prompt and the approval rule against what actually arrives, not against what we predicted.
Before and after
How a salesperson's day changes
Before
- At 19:00 they write up three meetings from notes on their phone
- Before the next call they rebuild from memory what was agreed a month ago
- The client asks a week later whether the quote is coming at all
- The manager asks how many deals are sitting after a meeting and nobody knows without opening the CRM
After
- The summary goes out within 60 minutes of the meeting and the rep only approves it
- The account history is in the CRM, they open the record and read it
- The task "send the quote by Friday" sits in Asana with their name and a date
- Deals stuck after a meeting for more than 10 days show up from one question to the AI agent
The first summaries are visible in the pilot week. The quality of the prompt and the approval rule we judge after 30 days, the effect on the funnel after 90.
For the person signing this off
What a head of sales gets out of it
Follow-up stops depending on how tired a rep is at 18:00. The summary after the fifth meeting of the day has the same quality as the one after the first, because the same rule writes it from the same data. The sentence "nobody ever came back to me" stops landing on your desk.
The second thing matters more in front of the board. You see the funnel as it is: how many meetings happened this week, how many summaries went out, how many deals moved on and how many are standing still. Today that is an impression from the weekly review. After the rollout it is a number in the CRM.
- The same standard of summary after every meeting, whoever ran it
- Account history in the CRM, so a rep leaving does not take the client knowledge with them
- A pilot on real meetings with approval, before any client sees anything
- You change the template and the tone yourself, with no ticket raised with us
What it costs
The workshop has a published price, the rollout is quoted after it
The process workshop costs EUR 3,500 net per week, and runs from one to seven weeks. Follow-up is usually one week, because the process is narrow: from the end of a meeting to an email and a set of tasks. We quote the rollout after the workshop, once it is clear how many tools are in scope and how much work the prompt needs on your meetings.
You pay for the model, the transcription and the workflow tool directly to the provider, with no margin from us. A single model query costs the provider roughly EUR 0.005 to 0.07, and when n8n sits on your own infrastructure the workflow layer costs you nothing at all.
Process workshop
EUR 3,500 net
per week, and follow-up is usually one week
A self-contained piece of work. It ends with a map of what happens after a meeting, a written structure for the summary and a plan you could execute without us.
- A map: who writes notes, where they go, how long until the summary, where it goes missing
- The structure of the summary: what was agreed, by whom, by when, what is only a suggestion
- A trigger and tool recommendation: transcript, calendar or CRM note, n8n or Make
- The approval rule and the metrics to measure at 30 days
- Most popular
Workshop and follow-up rollout
Quoted after the workshop
scope follows from the number of tools and the work on the prompt
The option most companies choose. A working workflow 2 to 6 weeks after the workshop, tested on your own meetings.
- Everything in the workshop
- A workflow in n8n or Make wired into the calendar, the transcript tool, the CRM and the task tool
- A prompt that separates decisions from suggestions, tuned on your meetings
- A pilot with approval, then production with error monitoring
- Training for the team on editing the template and the tone of the message
Follow-up and an AI agent for sales
Quoted after the workshop
for teams that want to ask about the funnel rather than click through the CRM
Everything in the rollout plus BOND, which answers questions about follow-ups and deals from company data in 5 to 10 seconds.
- Everything in the follow-up rollout
- An AI agent wired into the CRM and the mailbox: which follow-ups got no reply, how many deals are stuck after a meeting
- A reminder for the rep when a client has not replied after 5 working days
- Access for the reps and for the manager, agent rollout around 2 months
If the workshop shows that a template in the CRM and a calendar reminder are enough, we say so and show you how to set it up. The workshop ends with a document that stands up without us, and a rollout you do not need never starts.
Frequently asked questions
What is automatic follow-up after a meeting?
It is a workflow that reads the notes or the transcript once a meeting ends, writes a summary of what was agreed and sends it to the participants, usually within the hour. In the same run it creates tasks in Asana, ClickUp or Jira and moves the deal in the CRM to the next stage. A person approves wherever the rule says they should.
How long does a follow-up rollout take?
One week of workshop and 2 to 6 weeks of rollout. The spread comes from the number of tools being integrated and from how much work the prompt needs on your meetings. A team working from transcripts with one CRM sits near the lower end, a team with notes in three places nearer the top.
Do we have to record meetings for this to work?
No. A transcript from Tactiq, Otter, Fireflies or Fathom is one of three routes. The second is the end of an event in Google Calendar or Outlook, after which the rep adds three sentences. The third is a note in the CRM. In the workshop we pick the one the team will actually keep up with.
How does the model know what was agreed and what was only a suggestion?
From a prompt that separates the two categories, and from tests on your own meetings. An untuned model treats both the same way, which is why the pilot runs with a person approving every send, and why we correct the prompt wherever it loses the thread, most often on meetings with two topics at once. Only then does the workflow run for the whole team.
Can the client tell that an AI agent wrote the email?
The email goes out from the rep's mailbox, under the rep's name, and the content is what the two of them agreed in the conversation rather than a generic note. The rep can adjust the tone before sending or trust the rule and send straight away. You edit the template and the tone yourselves, with no ticket raised with us.
What happens to the meeting transcript and the client's data?
We pass only what is needed to write out the decisions, and in the workshop we agree how long a transcript is kept before it is deleted. Recording a call requires telling the participants, and that stays on your side. Personal data is processed under <a href="https://eur-lex.europa.eu/eli/reg/2016/679/oj" rel="noopener">GDPR 2016/679</a> in the EU, and under the law of the client's own market when they sit outside it.
Which tools do we need to have in place?
Google Calendar or Outlook, plus a CRM or a task tool. The rest, meaning the work of joining those systems into one workflow, is ours. You do not need an IT department or any knowledge of language models. You need one person who can describe what happens after a meeting today.
Does this work in customer service, outside sales?
Yes. In a customer success team a consultant promises during a call to come back with an answer, and that promise is usually the thing that goes missing. The workflow turns it into a task with an owner and a deadline, and the client gets a summary of what was promised. The same mechanism works in consultancies, agencies and financial brokers.
Let us find out where your follow-up goes missing today
The free consultation takes 30 minutes. You tell us how many meetings a day your team runs, who writes the notes and how long it takes before a summary goes out. We show you which trigger makes sense for you, what can be switched on first and what needs the CRM sorted out beforehand. You do not have to share any systems at this stage.
We run at most two rollouts at a time, so that someone is genuinely sitting on each of them.
Book a free consultationPS. If anyone on your team wrote up a meeting after 18:00 this week, or did not write it up at all, that is exactly the problem. The client gets the decisions in 60 minutes and a commitment stops living in one person's memory. That is where the workshop starts.