Watch what a rep actually does in the two minutes before a discovery call at most companies. They open the CRM record, which has a name, a company, an email and a form submission from nine days ago. They open LinkedIn to find out what the person does. They open the company site to find out what the company does. They search for recent news. They scroll the email thread to see what was already promised. They check whether anyone else at the company is already in a sequence, and usually give up on that one, because the answer lives in a different tool.
Then they join the call, ninety seconds late, and open with a question the prospect already answered on the form.
The rep is not underprepared. The preparation was never assembled, it was left as an exercise for the person with the least time to do it.
What a brief actually needs
A useful pre-call dossier is short and answers five questions. Anything longer does not get read in the ninety seconds it has to be read in.
- Name, title, company, email
- The form they filled, if it was a form
- Lead source, probably wrong
- A notes field, probably empty
- Who they are, role, seniority, what they actually own
- Why now, the signal that brought them in, and when
- What they asked for, in their words
- What we have already said to them, across every channel
- What to avoid, an open complaint, a competitor already in play, a pricing conversation someone started
Why most stacks cannot produce it
Not because the data does not exist. Because it exists in six places, and no single tool holds all of it.
The enrichment provider knows who they are but not what they asked. The form tool knows what they asked but not what was said next. The sequencer knows what was sent but not what the website visit signalled. The scheduler knows the meeting exists but nothing about the person attending. The CRM has a subset of all of it, synced on a delay, with the free text stripped.
Assembling a brief from those six requires a human to visit all six. That is precisely the work that gets skipped when a rep has back-to-back calls, and it gets skipped most reliably on the busiest days, which are also the days with the most meetings.
This is the strongest argument against the idea that AI in sales exists to replace reps. Nothing about that pre-call scramble is a rep performing badly. It is a rep doing unpaid integration work because the systems will not talk to each other. Remove the scramble and you have not replaced anyone, you have given them back the first ten minutes of every call.
The version that reads well
A good brief is written for the ninety seconds it gets. Two lines on the person. Two on the company and why it is plausibly a fit. The specific trigger, dated. A short list of what has already been said to them, in order, including anything sent automatically on your behalf. Then the caution: the objection raised last time, the fact that they went quiet for three weeks, the colleague who is also in a sequence.
What makes that possible is not a better summarisation model. It is that one system saw the capture, the enrichment, the qualification, the routing and every message, so there is a single record to summarise in the first place.
A test worth running
Pick a meeting on a rep's calendar tomorrow. Ask them, without warning, what they know about the person. If the answer takes them to a browser tab, the brief is not being assembled anywhere, it is being reconstructed, from scratch, by your most expensive people, several times a day.
