Oops! Something went wrong while submitting the form.
Most AI prospecting tools make a static list faster. Sai looks at who actually engaged with you this week — viewed, liked, commented, asked a question — and drafts outreach that references the specific thing they did.
The recording is a real session. The sheet on the right is what it produced.
Sai opens each profile, pulls the signal, and writes the row, live, in a real browser.
Eight columns, sorted by score, with a source link behind every claim.
A LinkedIn account you're already signed into, and a rough sense of who your ideal customer is — Sai asks before it starts.
A shortlist of people who engaged with you recently, what each one actually did, whether they fit your ICP, and a drafted message for each that references the specific interaction.
Under 10 minutes.
Run it weekly. Signals decay fast — a comment from three weeks ago isn't a signal anymore.
Less than most tools claim, and something different from what they're selling.
The standard pitch is that AI makes prospecting faster: it enriches a list, generates a first line for each row, and sends. That's real automation and it does save time. But look at what it optimizes — it makes contacting a static list of strangers more efficient. The list itself is unchanged, and the list was the problem.
The more useful application is upstream. AI can watch behavior that's already happening around you and tell you who's worth contacting right now. That's a different question from "how do I personalize 200 emails," and it's the one this task answers.
Something a person did that costs them attention.
The strongest ones are the ones almost nobody works:
None of these are intent data you buy. They happen in your own account, they're visible to you and to nobody else, and most people never look at them because there's no dashboard pointing at them.
There are also weaker signals that get treated as strong ones: a funding round, a job change, a company hiring for a related role. Those are events, not attention. They tell you a company might have budget. They don't tell you anyone is thinking about you.
Because a well-written message to someone who isn't thinking about the problem is still an interruption.
The industry spent five years optimizing message quality and the returns flattened, which makes sense — everyone got better at the same time, so the relative advantage disappeared. Meanwhile the variable with the most leverage barely gets touched: whether the person is currently paying attention to you.
Someone who commented on your post yesterday is in a different state than the same person next month. They remember what you wrote. They've already decided you're worth a moment. A message referencing their comment isn't cold outreach, it's a continuation of something they started.
This also means signals expire. A comment from three weeks ago has decayed into an ordinary cold contact, which is why this is a weekly task rather than a quarterly one.
It asks you two things first, then goes and looks.
Before it runs, it asks where your signals actually show up and who your ICP is. That second question is the one that matters, because the failure mode here is treating every interaction as a lead. Plenty of people who like your posts are peers, competitors, or job seekers. Without an ICP filter you get a list of your own audience rather than a list of prospects.
Then it opens your recent posts and reads who engaged — not just counts, but the actual comments, so it can tell the difference between "great post!" and a question about how something works. It opens the profiles of people worth checking. And it builds a shortlist where each row has the person, the specific thing they did, and why it qualifies.
It works in a visible browser inside your own logged-in session, so you can watch it read each comment and stop it at any point.
A draft per person, referencing what they actually did.
The prompt asks for messages that don't feel automated, and the thing that makes a message feel automated isn't tone — it's the absence of anything that couldn't have been sent to a thousand other people. A message that opens with the specific question someone asked in a comment thread is unfakeable, because it required someone to read the thread.
For deciding which of those drafts are worth sending and how to source the claims in them, our AI sales outreach page goes deeper on the quality bar. And when a signal is strong enough to justify a real conversation, the account brief workflow on our AI for sales prospecting page is what you run before the call.
Sai drafts. It doesn't send. Sending is a decision you make per message.
No, and it doesn't scale like it either.
This produces a shortlist — the people who engaged with you in a given week, which for most people is somewhere between five and thirty names. That's not a pipeline on its own. If you need volume, you still need a sourced list, and tools built for that do it well.
What this changes is where you start. Working the signal list first means the first hours of your week go to the highest-probability conversations available to you, and the cold list gets what's left rather than what's first. Most people have it backwards, not because they disagree, but because the cold list is the one sitting in a CRM and the signal list is invisible until someone goes and looks.
