How an AI Assistant Categorises and Prioritises Enquiries: Rules to Set Up for Your Morning Briefing

BOTrento Team22 October 20268 min read

What "categorising" and "prioritising" actually mean

These are two separate jobs, and mixing them up is the most common reason a first setup disappoints.

Categorising: what is this message?

The assistant reads the email text and decides what kind of message it is. For a small B2B company the useful categories are usually few: new enquiry, reply to a quote you sent, request to change an existing quote, a promise someone made ("I'll get back to you by Friday"), and everything else.

Prioritising: what needs a person first?

Priority is a ranking inside those categories. A new enquiry with a date, a venue and a clear scope outranks a vague one-liner. A reply to a quote you sent last week outranks a general newsletter question. Priority depends on your rules, not on how the email is written.

One limit to keep in mind: the assistant works from the text of the email. It does not read attachments, so a brief that lives only in a PDF will not be reflected in the ranking. Ask people to paste key details into the message body, or note this in your process.

The rules and data the assistant needs from you

A ranking is only as good as the inputs. Before you switch anything on, gather these five things.

1. Enquiry types. Name the handful of types you actually handle, for example equipment hire, full event production, one-off service or support request.
2. What "urgent" means for you. A date inside a set window, a named deadline in the email, or a customer waiting on your answer. Write it as a plain sentence.
3. Customer signals. Existing customers, repeat buyers, partners or key accounts you always want near the top. A simple list of domains or company names is enough. Export it from Pipedrive, HubSpot or a spreadsheet as CSV or XLSX.
4. Response-time targets. How quickly you want to reply to each type: same day for new enquiries, a couple of working days for general questions, and so on. These are your own targets, not something to copy from a template.
5. Deal-size or effort hints. Words or situations that suggest a larger job, such as multi-day, multiple locations or a tender. Be honest about which ones really change how you treat an enquiry.

If you want a broader check of whether your inbox is ready, the AI assistant readiness checklist for sales covers mailbox access, written prices, approval rules and privacy.

Set up your priority rules: a step-by-step method

You do not need a complicated model. Start with a short rule set you could explain to a new colleague in a couple of minutes.

1. Pull a batch of real enquiries from your inbox or CRM export. Use real ones, including the messy ones.
2. Label each by hand with its type and a simple priority: reply today, reply this week, low. Do not overthink it.
3. Write down why for the ones that surprised you. "Existing customer", "event is soon", "no date given" are the reasons that become rules.
4. Turn the reasons into rules in the form "If X, then higher priority". Keep the list short enough to remember.
5. Add exceptions you never want missed: a key account, a complaint, a message that mentions a deadline.
6. Test on the labelled set. Compare the assistant's ranking with your hand labels and adjust the rules where they disagree.

Start with one process, such as new enquiries only, and widen later. The split of tasks to hand over and tasks to keep human helps you decide where to start.

What a good morning briefing looks like

The point of ranking is a short list you can act on before your first coffee has cooled. A useful briefing shows, per item, the category, the reason it ranks where it does, the customer, any date mentioned and a suggested next step.

Compare this with a sorted inbox. A sorted inbox still makes you open every message. A briefing tells you which few to open first and why. For a worked walk-through of the routine, see what a sales AI assistant does in your inbox each morning.

It also needs to include promises, not just new mail. If you wrote "I'll send the quote by Friday", that commitment belongs on the list on Friday morning. The same applies to quotes waiting for a reply; if this is your weak spot, read how to stop forgetting quote follow-ups.

What to review before you act on a recommendation

Treat the ranking as a well-prepared suggestion, not a verdict. Check these points each morning, especially in the first weeks.

• Category. Was a change request wrongly treated as a new enquiry?
• The reason. Does the stated reason match what the email actually says?
• Dates. Confirm any date or deadline against the original message and your calendar.
• Customer match. Did it link the sender to the right company, or miss that they are an existing customer?
• Missing information. If the key detail was in an attachment, the assistant could not see it.
• Draft wording. Prices, availability and commitments in any draft are yours to confirm.

Keep a short tally of wrong rankings in the first weeks. Each miss is either a missing rule or a rule that is too broad, and both are quick to fix. If speed of reply is the goal, these steps to reply to sales enquiries faster are a good companion.

How BOTrento handles this

BOTrento Chief of Staff is a managed process rather than a tool you configure alone. It watches enquiries and quotes in your mailbox and in the data you upload. On every working day it prepares the Today 1+3 overview in the app and by email, and it drafts replies, quotes and follow-ups in four languages (CS, EN, DE, FR). It also catches promises such as "I'll be in touch by Friday" and reminds you of them.

Nothing is sent without a person's approval. BOTrento does not send anything on its own, and it does not confirm dates or bookings. You can see the result in the process demo.

Practical details: Seznam and other mailboxes connect via IMAP. Gmail, Google Workspace and Microsoft 365 work for now by forwarding or import, with direct connection in preparation. CRM or spreadsheet data comes in as CSV or XLSX. You can feed your own calendar in through an iCal link so the overview can flag meeting clashes. AI use is included in the plan with a monthly usage limit; you get a warning at 80% and new work pauses at 100% until the next month, with no automatic extra charge. Plans are on the pricing page.

On privacy: data is stored on servers in Germany (EU). The text needed for a task is processed by a commercial language model provider through its API, so a server in the EU is not a closed computation. For GDPR questions in your specific case, speak to a lawyer. If you run an event or production business, see the page for events and production.

Try the rules on your own enquiries

The fastest way to learn whether your priority rules hold up is to run them on real mail. The 14-day free trial needs no card and no API key of your own. The 14 days start with your first successful data upload, and you have 7 days to prepare. Registration is assisted: someone from our team contacts you within 24 hours on working days and helps connect your mailbox.

At the start of the trial you also get a "Lost money" report, which looks back over your mailbox when it is connected via IMAP. There is no guarantee of results, which is exactly why you test it on your own data first.