How do I write a good AI Research brief?

Updated Markdown

The brief you type is the only input AI Research works from, so a specific one finds far more real matches than a vague one. See "AI Research" for the basics of how a run works; this article goes deeper on writing the brief itself.

The three-part formula#

Aim for one sentence with three parts, ideally written in English:

  1. Role — who you're looking for (or the kind of company, for a company-level search).
  2. A visible signal — something the person or company actually did publicly, that shows up somewhere on the open web.
  3. Industry or region — narrows the pool so the agent doesn't have to guess what "relevant" means.

Example: "Heads of Sales at fintech startups in Germany that are hiring SDRs right now" — role (Heads of Sales), signal (hiring SDRs right now), industry and region (fintech, Germany).

Signal examples that work well#

Under Try an example, the AI Research start screen shows six example cards, each built around one kind of signal: Who's exhibiting (trade shows), Who's building sales (job postings), Who just started (new role), Who just raised (funding), Who's on stage (speakers at an event) and Who's near you (local businesses). Each card is tagged Companies or Contacts for what it returns; click one to load its full brief into the field and use it as a template, or click Insert example for another ready-made brief. The cards and Insert example show while the field is empty, and the card set follows your app language. A microphone button next to the field lets you dictate your brief instead of typing it.

Person or company — there's no switch to flip#

A ListPlus-run research has no on/off toggle for people vs. companies — the agent reads your brief and decides. Name a role (CTOs, Heads of Sales, Founders) and it searches for people; describe a company-level trait with no role attached — a funding round, a product launch, a technology migration — and it searches for companies instead. If you want one mode specifically, make that clear through wording rather than looking for a switch.

ai-research · research-costs-and-sources · categories-audience-database.

FAQ#

What counts as a "visible signal"?#

Something the person or company actually did publicly that shows up in a job post, a forum thread, a press release, a company page or a LinkedIn post — hiring for a role, complaining about a tool, switching technology, raising money, launching a product, or speaking at an event. A concrete signal is what lets the agent find and verify real matches instead of guessing.

Where do I find example briefs to start from?#

While the field is empty, the AI Research screen shows six example cards under Try an example — such as Who's exhibiting, Who's building sales or Who just raised — each tagged Companies or Contacts. Click a card to load its full brief into the field and adapt it (for example your own region or industry), or click Insert example for another ready-made brief.

Can I dictate my brief instead of typing it?#

Yes — use the microphone button next to the input field to dictate your brief instead of typing it (tooltip Dictate; click again to stop and transcribe).

How do I get AI Research to return companies instead of people?#

You don't switch modes; the agent infers it from your wording. Naming a role (Heads of Sales, Founders, CTOs) returns people, while describing a company-level trait with no role — a funding round, a product launch — returns companies instead.

Should I write my brief in English?#

Written in English if possible — the agent still understands other languages, but English briefs tend to find more sources, since most of the public web content it searches (job posts, forums, press) is in English.

Why did my brief return few or no matches?#

A vague brief gives the agent nothing concrete to search for and verify, so it returns few or no real people. Add a specific role, one visible signal, and an industry or region, following the three-part formula above.