Search for AI prompts for conference networking and you get event-planner content: agenda builders, session-track generators, post-event survey prompts. Underneath that sit a few generic networking lines about making small talk. None of it is written for the rep who has been handed a badge, three days, and an attendee list nobody has read.
These prompts assume the only real advantage an event gives you is that list, and that the work is choosing who to meet and preparing for them. The follow-up sequence afterwards belongs to trade show lead follow-up; this page owns the before, the during, and the triage decision you make on the flight home.
The attendee list is the only real advantage an event gives you
Everything else at a conference is available from your desk. The list is not, and it is usually treated as a directory to scan rather than a target set to work.
Here is the attendee or exhibitor list for this event, and here is my ICP and current pipeline. Split the list into three groups: people worth planning a meeting with, people worth a conversation if I happen to meet them, and people who are not a fit. For the first group, say which named account each one connects to and whether they are new or already in my pipeline.
Two hundred names become fifteen, and fifteen is a week you can actually plan. Research workflow is covered in the 15-minute account research brief, and event-side work sits in Events, Field Marketing & Community.
Prompts that pick the fifteen people worth your three days
Ranking by company size is the lazy cut. The better cut weighs whether you can plausibly get in front of them at all.
From the shortlist, rank the fifteen by expected value, scoring three things separately: account fit, seniority against what I can realistically reach, and how likely I am to encounter them given the sessions and sponsor commitments listed. Say which of the three is weakest for each name, and flag anyone I should ask for a meeting in advance rather than hope to meet.
The last flag is the one that changes behaviour. The people worth a booked slot should be emailed before you fly, not hunted in a hallway.
Preparing for a named person: prompts that find one true thing
Fifteen people is fifteen short preps, and the standard is low: one specific, verifiable thing that proves you did not read a title and wing it.
Here is what I know about this person and their company. Find the one thing I could reference in a first conversation that is specific, recent and clearly true, and say where it came from. Then list what I am assuming rather than verifying. If there is nothing specific enough to use, say so instead of inventing something.
The permission to return nothing is what keeps that prompt usable. Why specificity does the work is set out in the types of specificity that earn trust; prep craft generally sits in Meeting Prep & Discovery.
The thirty seconds you get, which is not a pitch
A hallway conversation is not a discovery call with less time. It has one job: earn a reason to talk again.
Write what I say in the first thirty seconds with this person: who I am, the one specific thing I know about their situation, and a question they would find interesting enough to answer standing up. No value proposition, no company description beyond one clause, and nothing that sounds rehearsed.
Take one prompt free, no account needed. Pro is $12 per month or $89 per year, and lifetime access is $149 — See pricing.
Models default to pitching because that is what sales copy looks like in training data. Strip it back every time.
Working the room you are in, not the one on the agenda
Half your shortlist will not show. The useful question mid-event is what to do with the afternoon you now have.
Three of my fifteen have not appeared and two sessions I planned were cancelled. Here is who I have actually met and what the remaining agenda looks like. Tell me where to spend the next four hours for the best chance of a conversation that matters, and what to drop.
Trigger-based attention is the same discipline off-site, covered in spotting buying signals with AI and in Prospecting & Pipeline Creation.
Triaging the badge-scan pile on the flight home
The pile is where events go to die. Treating every scan as a lead is why most of them get one generic email and nothing else.
Here are my scans and notes from the event. Sort them into three groups: real opportunities where a specific next step exists, people worth a single low-effort touch, and names to discard. For the first group, write the one-line next step for each. Do not draft sequences, and do not put anyone in the first group without evidence from my notes.
Discarding is the output most reps refuse to accept and the one that makes the rest workable. Once the groups exist, the sequence work is trade show lead follow-up, and the message craft is Outreach & Messaging. Day-shape for prospecting is in AI prompts for SDRs and follow-up timing in prompts for follow-up emails.
When an event is not worth attending, and how to argue that
Some events do not survive this process, and the list is what tells you before the flights are booked.
Working the attendee list against my ICP and open pipeline produced this shortlist. Write the honest case for not attending: what the list does and does not contain, what the three days cost in selling time, and what I would do instead. Then write the case for going. Do not favour either.
A shortlist that cannot reach fifteen names worth three days is the argument, and it is a better one than a feeling.
Why a model given an attendee list will tell you to meet everybody
Hand a model two hundred names and ask who matters and it will find a reason for most of them, because a plausible reason is always available from a title and a company. Force the third bucket and force the weakest-score column, or the shortlist is just the list again.
It also cannot know what decides these three days: that the person you want avoids the sponsor floor, that a competitor sponsors the party everyone attends, that your champion at a target account is here but has told nobody, or that the one real conversation will happen in a queue at 8am with somebody who was not on any list.
Use it for the parts that are genuinely work: cutting two hundred names to fifteen, finding one true thing about each, and sorting the pile honestly when you land. Keep the judgement about which room to stand in where it belongs.
The rest of the catalogue, by role and by stage, is mapped in AI prompts for sales. Browse the library → or read the full taxonomy.
Frequently asked questions
What are good AI prompts for conference networking?
Four carry the event: a triage prompt that cuts the attendee list into worth-planning, worth-meeting and not-a-fit, a ranking prompt that scores fit, reachable seniority and likelihood of encounter separately, a prep prompt that returns one verifiable thing per person, and a scan-triage prompt you run on the way home.
How do I decide who to meet at a conference?
Work the attendee list against your ICP and pipeline before you fly, then rank the shortlist on account fit, the seniority you can realistically reach, and how likely you are to encounter them given the agenda. Anyone scoring well on fit but badly on encounter should get a booked meeting, not a hallway hope.
How do I prepare for a conference with AI?
One short prep per shortlisted person, each aimed at a single specific, recent, verifiable thing you could reference, with your assumptions listed separately. Allow the answer to come back empty rather than invented, and write the first thirty seconds as a question rather than a pitch.
What do I do with the leads I scanned at an event?
Sort them before you write anything. Real opportunities with a specific next step, people worth one low-effort touch, and names to discard, with nothing in the first group that your notes cannot evidence. The discard pile is what makes the rest workable.
Is it worth attending this conference?
Judge it on the attendee list rather than the agenda, because the list is the only thing an event gives you that your desk cannot. If working it against your ICP and pipeline does not produce fifteen names worth three days, the honest answer is probably no.