Aurasell's Prospecting Agent Mines Your CRM, Not Vendor Lists

An AI prospecting agent booked a meeting by opening an email with a question about a prospect's home fish tank — a detail that existed nowhere except a meeting transcript sitting in Aurasell's own CRM for over a year. That demo, presented by Aurasell CMO Steve and head of product Richa at SaaStr AI Day, is the vendor's argument for why first-party data beats the third-party lists most AI SDR tools run on.

The setup: earlier that day, SaaStr's Jason and Amelia had flagged the structural weakness — every team buys from the same data vendors, so every agent writes the same emails to the same people. Aurasell's answer is an agent that also reads CRM history, past meeting transcripts, and internal conversations. Amelia previewed it a week before and plans to benchmark it against the agents SaaStr already runs. That benchmark does not yet exist; treat the fish-tank story as a single anecdote, not a win-rate study.

How did the fish tank get into the email?

The agent emailed a prospect named Chris, asking whether the fish tank was still thriving. No SDR had mentioned one. A Slack thread across product, engineering and sales traced the source: Chris had met Aurasell more than a year earlier, and the transcript's small talk covered his large home fish tank. No third-party vendor has that record. The caveat came from Steve himself: a personal hook alone gets a shrug — "Hey Steve, I saw you like golf" books nothing. The hook must lead to a real problem, and recency matters; Aurasell researches how old a signal can get before a six-month-old detail underperforms one from last week.

Where does first-party data carry the most weight?

• Startups out of stealth: founders take dozens of feedback meetings early; a year later, with the product changed, those contacts are the best list to revisit.
• Enterprises launching new products: years of CRM conversations hold account context that can open doors for the new line.
• Most outbound, in Steve's view, needs both first-party and third-party signals, chosen per campaign goal.

What did the live demo show?

Richa's walkthrough had no fixed sequence — the core difference from a standard sequence tool. The agent built a per-contact plan: one person got nine steps, another seven, based on prior contact history and LinkedIn activity. Reps review and coach each plan before anything sends. In the demo, a prospect named Marcus opened a follow-up 48 seconds after arrival; the agent logged medium interest and rewrote his next steps. Email clicks, LinkedIn replies, and account changes like new leadership all trigger replans. When a contact opens three emails and clicks without replying, the agent switches the next step to a phone call — a switch reps should make anyway. The demo searched an 850 million contact database; Aurasell did not disclose its vendor or verification rates.

What about turnover and deliverability?

The agent tracks each contact's current profile. If someone changes companies, it reaches out at the new address and identifies the right replacement contact at the old one. When LinkedIn is stale, it searches press coverage. On deliverability, Richa said Aurasell sources emails from multiple vendors, verifies before sending, and uses a built-in SendGrid setup to warm secondary mailboxes and spread volume.

Does it work beyond sales?

One data point: Richa launched a recruiting agent for PMs in India, left on vacation, and returned three days later to 20 booked meetings — the agent replied to candidates within two minutes, in their own time zones. That fixes a real agent weakness; Amelia noted agents defaulted to UTC a year ago.

The harder number sits inside Aurasell's own team: 14–16 meetings a week, about 60% from human cold calls on a power dialer, most of the rest from LinkedIn — almost none from email. Steve tests SDR candidates by having them write a prospecting email without AI, arguing that creativity and call control are what beat competitors running identical tools. Amelia said Databricks' CRO screens SDRs the same way; agents won't pick up the phone at step four. The benchmark against SaaStr's existing agents will show whether the first-party approach scales beyond one fish tank.