How do you see which of a bookkeeping firm's customers have not heard from you in a year, from email and software together?

Customers who go quiet show up on a list: your AI puts the latest messages next to the file details, with what Bombos remembers, and for each customer you see when there was last contact and what has been stalling since.

Works in your own Outlook, next to your own software.

Seeing who has gone quiet is the start; every conversation and promise stays in one shared memory.

Book your half hour

Together we map what your AI forgets. Free, and the plan is yours, with or without Bombos. Join the plugin waitlist →

This is how it goes now, and how it goes with Bombos.

Now

The promise "we call at least once a year" is not tracked anywhere, so quiet customers leave quietly.

You know which customers you speak to most often. Who you no longer speak to is exactly what you do not know, because that produces no email. To find out, you go back through Outlook for each customer and put the information from your software beside it. A document the customer sent is not the same as a reply from your team, and a phone call is missing if nobody recorded it.

Withbombos

The list of quiet cases sits ready, with what has been stalling for each customer since.

You start with a list of customers whose contact has been quiet for a year according to email and your software. Your AI puts each customer's last email beside their details. The list shows the last recorded contact, what has stalled since and the reason. You add a missing conversation or correct an entry, then decide whose contact to pick back up.

After three months

After three months, no customer leaves without someone having spoken to them. The list is ready, with a reason beside every name. Your team calls with something to discuss, and the promise to check in once a year is simply kept.

What's ready for you then

  • See which customers have heard nothing for a year
  • The reason for each conversation is already there
  • Keep your promise to check in every year
  • Call with a reason, not out of nowhere

Where do you see that?

The list of quiet customers sits with the other prepared work. For each customer you read the reason and the date of the last recorded contact. The status is "ready for you" or "waiting on your input". You look at the correspondence found and the file details next to the entry. Your software does not need to be open for that review. So you do not have to work out, name by name, what each entry is based on.

If a question from the office is still open, you read it next to the customer. If you know of a more recent phone call, you factor that into your review of the list. You decide who you want to reach out to; the entry on the list does not send any message. With every proposal you see where it came from.

How does it start?

In the first week we set up visibility into quiet contact together, during a setup conversation of about an hour. You show us the mailbox with customer correspondence and the software with the customer details. We agree on when you consider contact quiet, which messages count, and who reviews the list. You give access to the email and details needed for that task. The correspondence stays in your existing systems; nothing has to move and no course is needed.

After that, your AI starts preparing the list. After three months, no customer leaves without anyone having spoken to them. Every task starts with proposals. For the first names you review whether the reason holds up. If a message counts differently than intended, your correction becomes the rule for the next list. Bombos keeps that explanation as a rule; without it, the memory does not hold how you judge contact.

Once a task type has gone well for a while, you decide how much autonomy you give your AI, with Bombos as the memory underneath. You approve messages to the customer; you set that in the AI or agent itself. The rules for the list go in as settings, without custom work, even if you do not work with computers much. This list is a start, not the whole offer: after that, every correction becomes a rule Bombos keeps, so the next task starts from what you already agreed.

Every customer in one shared memory.

A fast-growing firm first loses touch with its quiet customers. Attention naturally goes to the people who call.

Bombos is the memory layer under the AI your team already uses: Claude, Claude Code, ChatGPT, Copilot, Grok, Codex, Cursor or your own agents. One plugin gives them all one shared memory: your team's knowledge, the rules it learns from your corrections, and what is already in your email, documents and business software.

You see where every answer came from. Bombos runs in the cloud or standalone on your own server, and one person can try the plugin with their own agents first.

The pilot runs three months for €2,000 instead of €3,000, then €1,000 per month, excl. VAT. Book your half hour to talk through your situation.

What Bombos does, in three steps

  1. It reads what comes in

    An email, a message, a photo or a PDF: your AI reads it, finds the client and the details through Bombos in the systems you already have, and sees what is missing.

  2. It drafts the work, you approve it

    The proposal sits on one list with the reasoning next to it. You read it, adjust it and approve. Only then is anything sent, booked or filed.

  3. It remembers what you decide

    Every correction becomes the rule for next time. Once it has gone well for a while, you give your AI more autonomy step by step, with Bombos as the memory underneath.

What does Bombos not do here?

Bombos does not act here itself: it gives your AI the memory. Without a connection between your AI and Outlook and your software, it stays a proposal.

Your AI does not work out from an old message which customer is going to leave. The list shows what has been recorded about the contact. A customer might meanwhile have discussed something with your colleague by phone. That is why your AI asks you once what contact happened outside email and what was agreed there, and Bombos keeps the answer. That way you can factor that knowledge in when you review the list, instead of reading dissatisfaction into silence.

When in doubt, your AI puts the question to you. An entry can be wrong or incomplete; you check the reason before reaching out to the customer. You approve before an outgoing message goes to a third party, through any channel, even for picking contact back up. You set that in the AI or agent itself. The list does not predict who is leaving; it helps you see who you want to talk to again.

The first half hour · free

What does your AI forget?

  1. 01You tell us which AI you use and where it keeps starting over.
  2. 02We show what a shared memory changes there, on your own examples.
  3. 03You keep the plan, with or without Bombos.
Piet Baudoin
Piet BaudoinYou talk to Piet, from the Bombos teamDutch team from Utrecht and Nijmegen · data in Europe
We reply the same working day.