AI Research

Q: How many matches can one AI Research find?

You pick a stop target of 25, 50, 100 or no limit under Stop after; the run also stops if its Credit limit runs out first. Click Resume — even on a finished run — to open another pass the same size as what you originally chose, up to 200 matches in total without a target list or 1,000 with one.

Q: How do I get the matches into a list?

Tick the matches you want and click Add to list to file them into an existing list (free) or a new one (which costs credits); open a single match to view its full profile and save it to a folder instead. You can also turn on automatic hand-off so every qualifying hit goes to a list by itself — see "Getting AI Research hits into a list automatically."

Updated October 15, 2026Markdown

AI Research (KI-Recherche) finds audiences no filter can find: people or companies that did something visible on the open web. Open it with Find in the left navigation, then AI Research in the switch at the top right. The form has three steps: Describe the leads, Set up the research and Allowed AI tools. Two kinds of agent can run it: ListPlus — our own agent, working on the server, even with your window closed — or Your AI — an assistant you already use, such as Claude or ChatGPT, through a private link (its own topic: "Letting your own AI research for you"). Every match carries a fit score and the evidence behind it, and a run never returns the same match twice.

Describe the leads#

Type your audience in your own words, best in English — the field's own placeholder shows a full example ("CEOs and founders of B2B SaaS companies in the US that raised seed or Series A in the last few months"). Click Insert example to drop that exact brief into the field as a starting point, or use the microphone button next to it to dictate instead of typing. For the formula behind a strong brief and more example signals, see "How do I write a good AI Research brief?"

Allowed AI tools#

Step 3, Allowed AI tools, lists every source as its own switch, in folding groups that each show how many are allowed: Find (Database, LinkedIn people, Web search, Read web pages), Signals (Job postings, News), Community & intent (LinkedIn posts, Reddit, X, Hacker News) and Verify & enrich (Find LinkedIn). A switch is a permission, not an order: the agent may use a source you tick, and it queries one only when your brief calls for it — a database lookup when the brief names a role, place or industry, job postings when it asks for companies that are hiring, LinkedIn posts when it asks who talks about a topic. Ticking a source never triggers a query by itself, and a ticked source that the brief doesn't need costs nothing. Each tile shows what it's for and what it costs. Web search, reading pages, job postings, news, Hacker News and finding a real LinkedIn profile are free; Database, LinkedIn people and the community sources other than Hacker News are paid per call — see "How is AI Research metered, and what do paid sources cost?"; all prices are in "Price list: what every action costs".

Stop conditions and budget#

In step 2, Set up the research, Search goal → Stop after sets how many matches (or, with a target list, new records in it) the run looks for: 25, 50, 100 or no limit. Under Budget & model, Credit limit sets the credit budget for the run — you pick it from a list before you start (the options are in "Price list: what every action costs") — and the run ends at whichever of the two it hits first, or once a round finds nothing new. Keep searching until … are found (on by default, shown when you set a target) makes ListPlus start another round, up to 5 times and never past the credit limit, when the agent would otherwise stop early with fewer matches than you asked for.

Model picks the AI that does the work: the default is fast and cheap, the alternative has the strongest judgment but costs many times more per step. With the stronger model, a hint shows about how many steps your credit limit buys.

Sending hits to a list#

By default matches stay in the research; tick the ones you want and click Add to list to file them into an existing list (free) or a new one (which costs credits). You can also answer Automatically add new matches to a list? in step 2 with Yes, add them as they are found: pick a list (or a new one) and every qualifying hit is added on its own, with duplicates kept out — full detail, including the fit-score filter, in "Getting AI Research hits into a list automatically."

Watching a run#

Every run works on the server, so it keeps going when you close the tab, switch away or your laptop sleeps, and you can review or start a research on your phone too. In a workspace, researches are shared with the whole team: everyone sees them in the Researches column, and a run a colleague started shows their name. Click Stop at any time to end a run and keep everything found so far, or Resume on a paused, stopped or even finished run to open another pass the size of what you originally chose — see "AI Research always runs on the server" for how long an unattended run keeps working before it needs a Resume.

Costs#

The usual AI costs apply: the agent's reasoning is billed by usage, with no flat fee per match, and the stronger model costs more per step. On top, each tool you allow under Allowed AI tools that isn't free is billed per call it makes while the run works. Which sources cost credits: "How is AI Research metered, and what do paid sources cost?"; running out of credits mid-run: "Not enough credits."

For advanced users#

A resumed run reuses the switches, target and budget it started with — change any of those by starting a new research instead. Without a target list a run holds at most 200 matches in total; with one, up to 1,000, since the list itself absorbs and dedupes the hits instead of the run's own memory. Finding a real LinkedIn profile (identity) is always free and runs to ground a match before it's recorded, which is also why the agent can't invent people.

Related topics: research-costs-and-sources · write-a-good-research-brief · research-results-to-a-list · background-runs · your-ai-researches · not-enough-credits · what-is-a-workspace.

FAQ#

What is AI Research and when should I use it?#

AI Research finds people or companies that did something visible on the open web — hiring for a role, complaining about a tool, raising money or speaking at an event — that no database filter can find. You describe the audience in a sentence, pick which of our own agent or "Your AI" (an assistant you already use) should search, and matches come back with evidence and a fit score. Use Segment Search instead when your audience is defined by role, industry and company size alone.

How do I describe a good target group?#

Type the role or company trait, a visible signal, and an industry or region, best in English — the field's placeholder shows a full working example, and clicking Insert example drops it straight in. You don't pick person vs. company mode yourself; the agent infers it from your wording.

Does ticking a source mean the agent searches it?#

No. The switches under Allowed AI tools are permissions: the agent may use a ticked source, and it queries one only when your brief calls for it — the database when you name a role, place or industry, job postings when you ask for companies that are hiring, LinkedIn posts when you ask who talks about a topic. A ticked source the brief doesn't need is never called and costs nothing; an unticked source doesn't exist for the agent, even when the brief would need it.

What does an AI Research cost?#

The usual AI costs apply — the agent's reasoning is billed by usage, with no flat fee per match — plus the paid sources the agent actually calls — only ones you allowed under Allowed AI tools, and only when the brief needs them — each billed per call. See "How is AI Research metered, and what do paid sources cost?" for which sources are paid, and "Price list: what every action costs" for the prices.

How long does a research take and can I close the window?#

Several minutes for a small target, longer for a bigger one. Yes — every research runs on the server and keeps going after you close the tab, switch away or your laptop sleeps, for a solid block of unattended work before it needs a manual Resume.

What does "Your AI" mean?#

Instead of ListPlus's own agent, you can hand the research to an AI assistant you already use — such as Claude or ChatGPT — through a private link that reports matches back into the same cockpit. It's a separate way to run a research, covered in "Letting your own AI research for you."

Why did the research find fewer people than I asked for?#

Only people or companies with real evidence on the web are returned — the agent never invents matches. With Keep searching until … are found on (the default), ListPlus starts another round, up to 5 times and never past the credit limit, until it reaches your chosen number or stops finding anything new; turning on more sources, or broadening the brief, usually finds more.

How many matches can one AI Research find?#

You pick a stop target of 25, 50, 100 or no limit under Stop after; the run also stops if its Credit limit runs out first. Click Resume — even on a finished run — to open another pass the same size as what you originally chose, up to 200 matches in total without a target list or 1,000 with one.

How do I get the matches into a list?#

Tick the matches you want and click Add to list to file them into an existing list (free) or a new one (which costs credits); open a single match to view its full profile and save it to a folder instead. You can also turn on automatic hand-off so every qualifying hit goes to a list by itself — see "Getting AI Research hits into a list automatically."

How long are my AI Research results kept?#

A research is kept for 7 days after you start it by default, whether running, finished or stopped. With automatic hand-off to a list, the research and its link are kept for 30 days instead; the records themselves live on in the list.

Can more than one research run at the same time?#

Yes — there's no fixed limit on how many researches run at once. Each one spends credits as it works, so your balance is the only real limit.

What is a fit score?#

Every match gets a fit score from 0 to 100, with a one-line reason, rating how well it matches your brief on role, company, place and signal. See "Getting AI Research hits into a list automatically" for using it to filter what reaches a list.