Half Your CRM Is Wrong: AI Prospecting in a World of 49% Exec Turnover

In this AI Day panel, moderator Amelia introduces Aura Sales' prospecting agent, which uniquely blends first‑party CRM data with massive third‑party signal libraries to generate highly personalized, dynamically updated outreach sequences. Steve recounts a concrete "fish‑tank" example where the agent mined a year‑old meeting transcript to craft a hook that secured a meeting, illustrating the power of internal data. Richa demos the product, showing how the agent builds contact lists from an 850 M‑record database, creates per‑contact multi‑step plans that adapt in real time to signals such as email opens, LinkedIn activity, leadership changes, or time‑zone considerations, and allows human reviewers to coach, edit, or approve each step. The discussion covers deliverability safeguards (email verification, SendGrid‑based mailbox warm‑up), handling stale or turnover‑affected contacts by refreshing profiles, and the importance of human SDRs for cold‑calling and pitch execution—tasks AI still can’t reliably perform. Participants compare channel effectiveness (cold‑call > email, LinkedIn ~60/40), explore broader use cases (recruiting, event marketing), and emphasize that AI augments but does not replace the salesperson, requiring a blend of data‑driven automation and human judgment. The session ends with a call‑to‑action to request a demo.
Chapters
One, I showed you guys a little bit about how we do outbound earlier today. We chatted, Jason and I chatted about some of the issues with prospecting tools, outbounding tools not having a lot of first-party data. And I mentioned that today you're gonna see how Aura Sales prospecting agent actually does use first-party data. And so that's actually super cool. They gave me a preview of it last week. I think it's a really great tool, especially if you, if you don't have a prospecting agent yet, or if maybe you're frustrated with the ones that you currently have. This one's a good one because it does have the first-party signals that you can use along with all those really great third-party data and signals as well. So I'll stop yapping about it before we talk about, um, YAPA. But Steve, I'll hand it to you. And if you guys have questions as we go, please drop them in the chat and we'll do Q&A at the end.
Cool. Thanks, Amelia. Uh, welcome everyone. Um, here I'm talking about another prospecting agent, so we'll jump in. This is my pet peeve. I'm sure many of you have received emails like this where you look at it and you can instantly spot like this actually wasn't personalized. It's hard to read. It's, it just, even when you see bold used and emojis, you kind of think, is this really a human that's writing this email? And so like we wanted to kind of like not do this as an example. In addition, I'm sure you've had many conversations or conversations like you see on the right where the AI SDR just talks to itself. Like I've seen my inbox just have a complete conversation with itself where I don't respond and it just keeps going and going and going. And this is really where we've got to in 2026 with a lot of prospecting agents. I get this every day. I had one yesterday. You can see on the left-hand side. Where I even asked it like, who's going to win, the Commanders or the 49ers, right? Nothing to do with like the pitch or what they were trying to sell me. It was just an AI bot like making me happy. So we don't want to do that. We had 2 options when we were coming to market. We can actually rush things and we can be like those other prospecting agents. Or we thought like, how about we solve the problem and we actually do some research and we look at like, why are we ending up with prospecting and outreach like you see in those slides, right? So we wanted to try and solve the problem. A few months later, obviously nothing happens even with vibe coding. You can't create something in a few hours to solve this problem. We actually came up with our first version of our prospecting agent and it booked a meeting, right? And we were like, awesome, like how did it book a meeting? And then suddenly like the SDR whose meeting it was was like, hey, I didn't mention a fish tank. Like, where the hell did a fish tank come from? And the agent actually made a reference to the prospect's fish tank. Now, like, I can assure you, many of you don't talk about fish tanks on any customer calls or outreach. And so, like, we started to, like, understand, like, how did it get that context or that hook, right? So this was the email that was sent. If you look at the email, it's actually pretty simple. It's relevant. It's got a nice hook. It's easy to read. But if you look at the top, you can see the subject is Chris, the fish tank still thriving, right? And so like we drilled into this and we were like, all right, like it's a good email. Like doesn't look like it's written by AI, but like you can see on the top, hope the fish tank crew is still thriving. So we started to think like, where actually did the fish tank come from? And this was this huge Slack conversation between product engineering, and the SDR team because no one could figure out where the fish tank came from. And sure enough, it actually came from a previous meeting that was in our CRM that was over a year ago. So Chris had already met with us. The timing wasn't right. It actually went through the actual meeting transcript and it pulled— you can see in the intro of the meeting, they talk about Chris's large home fish tank. Right. And so that was an example where the prospecting agent didn't just use like public signals or third-party data. It actually looked at our CRM for the past year and tried to find relevant things to try and hook Chris. No pun intended with a hook. We're talking about a fish tank. I thought you'd get that one. And so like, this is the difference we've engineered our prospecting agent to look at. Everything today pretty much relies on third-party data and signals. What if your prospecting agent could look at everything in your CRM over the years, and try and pull relevant meetings, Slack messages, conversations, and use that as a means to engage prospects and contacts. So that's kind of like the difference of, of, in a kind of like problem and what our solution looks like. I'm going to hand over to Risha and she's going to demo this.
Hello everyone, I'm Richa. I lead product at RSL and I wanted to give you a quick demo of how the prospecting agent works. This is an orchestrated sequence to give you a full demo within like 5 minutes, but the product is live and you can test it out for, for your instances. So We are different from a sequence. We do not create a sequence for your prospecting. We in fact change it every single time the prospector is hit by a particular sequence. So while typical prospecting creates a sequence, we actually change it live for each contact that we are prospecting to. And humans have full control over this agent. They can review the plan, they can coach the agent, and they can therefore trust how the prospecting agent acts on their behalf. So let's get started. Suppose we are trying to prospect VPs or CROs at major transportation and logistics companies in Washington State. You will just prompt the agent to find you the contacts. It will look into our 850 million contacts database to find the relevant contacts that are meeting the criteria. It will then find the most qualified VPs and CROs and make a list for you and start planning for those particular contacts. How should I reach out? It will then plan different steps for each contact. Somebody will get 9 steps, somebody will get 7 steps, and depending on how they have previously been contacted through the system or have recently had any activity on LinkedIn, the right steps get prioritized. The plan comes live. It is relevant in terms of what you're reaching, who you're reaching out to, why you're reaching out to them, And what's relevant to their business. All those signals are basically baked in. Once you confirm the plan, it will start reaching out to those people and it will basically figure out, have they clicked on an email? Have they responded to your LinkedIn message? Has a buyer signal been received from their account? Maybe a new leadership came into their account and then it will start to change the plan as the new signals arrive into the platform. And so, for example, here Marcus opened the first follow-up email within 48 seconds, a medium signal confirming continued interest. And so it will replan the next steps based on what Marcus did, right? And you can tweak this prospecting sequence any single time. So say, for example, you're done with 20 contacts and you want to add more prospects to the campaign. You tell the prospecting agent to do that and it will add and find more people. For your prospecting use case. Suppose you want to run a similar campaign for Director of Operations at Freight Forwarded in Texas. You could do that, and it would then create a very similar prospecting session for your prospects in Texas. You can also give it feedback like change— make sure to reach out in a reasonable time zone, and it will reach out. So far, We have seen immense success using our own prospecting agent, and because it thinks with you on your behalf with every new signal that it receives. I'll stop here, and Steve and I are happy to take any questions that you guys have.
Yeah, you got one question, Marisha. How do you verify and ensure deliverability?
Yes. From what I'm seeing. So we leverage a lot of tools, including the right kind of industry best practices around email deliverability, as well as LinkedIn deliverability, to ensure that we are ensuring the right emails are being reached out to the right people. We in fact work with multiple vendors to get the right emails across the board and then verify and validate those emails before sending it out. So your open rates and click rates are high as well as you are basically ensuring that a person is actually reached out through multiple channels.
And how are you guys doing that through like specific mailboxes that you're warming up for the users?
Yes. So you can connect your primary mailbox like Gmail or Outlook, Or you can also like connect your secondary mailboxes and we have a full SendGrid integrated infrastructure so we can warm up the new mailbox that you have connected and then like distribute load between those.
And there's a question on the livestream. How have you guys tackled some of the turnover in, let's say like SaaS companies, go-to-market companies? Like does your prospector kind of account for contacts changing over quickly as well?
100%. So that's like a big signal for us. We ensure that we are tracking each contact's live profile. So let's say, for example, if you have engaged 20 contacts in a prospecting session, we will basically find the latest information about that particular contact. If they've moved position, we're going to use that as a signal and maybe reach out to the same person but in the new company and find the new company email. To go to.
And then you find the, the relevant people at the, the current company that they're trying to reach.
Exactly.
Cool. Um, another question we had is when it— if it does hit a stale contact, how does the agent handle that? Does it pass it over to a human or does it do something automatically?
No, if it's a stale contact, we're going to refresh that information as live as possible. Like, and if, if the Human gives feedback that hey, I need you to find out further new information because sometimes people also don't update their LinkedIn profiles. We'll try and look up any news that has happened. Most execs would have certain news come out for them or PR come out for them. Use that and then leverage that from our prospecting in our prospecting session.
Yep. Another question from the live stream is: Do you guys have any data you can share on the? types of outbound you've seen work on your prospecting that maybe works a little bit better with first-party signals versus just third-party?
Yeah, I think Steve can speak a little bit to it, but like the fishtank example we gave was like purely first-party, right? Like we had a previous conversation, like a rep reached out many months back and then we used that and we got a meeting out of it.
Yeah, I think it varies depending on where you are as a company. So for example, a lot of startups when they come out of stealth, they'll do lots of meetings just to get feedback. Then like a year later, the product will totally change and then they'll want to go back to those people to try and get that, to get their feedback. Similarly, if you're an enterprise and you've had lots of conversations in the past and you've got a brand new product, like there's lots of signals that you already have about the customer, the contacts in your CRM that you could probably use to get interest on your new product. So I think like a combination of both. I don't think it's one or the other. I think it's like whatever makes sense. I think also as well, it's connecting the hook to the problem. I think you see a lot of outreach where it's just a hook that's personal and it doesn't really identify a problem for the contact. It's kind of like, hey Steve, I saw you like golf. It's like, great, thanks. But it doesn't get into like, how is it gonna help me? Like, what problem are you making me aware of that I might need to solve? And I think like having as many signals available to you as you can, so you can pick and choose. Like one thing Rishi didn't talk about was just the recency of the signal. Like there's a lot of research we do where we look at like there might have been a signal 6 months ago, but like, do we really want to use that versus a signal that was a week ago that is probably going to strike a chord? So I think it depends on what you're trying to achieve with the outreach. I mean, Rishi even used this for recruiting. So like to recruit PMs, we used it to fill dinner seats. So when we do field marketing, we use that to fill seats at some of the events we do. So there's a wide variety of use cases you can use it for.
Yeah, I was gonna ask you, like, as a CMO, like, how do you think about the Prospector tool for— and I actually, I don't know, do you run the sales team at Oresell? I actually don't know the answer to that question.
No, I don't. Jason can be here. Jason's our CEO.
He runs sales.
Yeah, he runs sales.
Yeah. So, okay. So then as just, you know, this, as the CMO who's running marketing, how do you, what if other than like events you just mentioned, that's a good use case. I talked about that one this morning too. And using a tool like prospecting for that. Do you have any other ones that come to mind for maybe marketing specific versus sales specific?
I mean, I look after SDRs. So recruiting of SDRs is important. I think for me though, like email and LinkedIn is a valuable channel and you don't want to abuse it. Like I feel like we've abused it the last 20 years in marketing because it was so easy to send an email. I feel like I'm pretty holy about like reaching out or contact for the fear of like, are they going to hit unsubscribe?
Yeah. Because then they're gone forever.
Yeah, exactly right. And so I think where you add value and you know you can help someone, like I think it makes sense. So, um, like for us, like a lot of our go-to-market hangs out on LinkedIn, right? So LinkedIn outreach and honestly voice calls is how we get our personas. Yeah. Yeah.
There's a question in the chat kind of related, um, which is like between email, LinkedIn, cold calling, which have you seen be maybe the most productive, um, in getting through?
Yeah. So, um, I would say cold calling. And LinkedIn is probably a 60/40 split.
And is your cold calling— just to clarify, is your cold calling through an agent or a human cold calling?
Um, we have a power dialer, so we, we do it. It's humans on— unfortunately, it's humans on the end of the phone that book meetings. Imagine that. Uh, yeah, so, um, our SDR team, um, is booking 14 to 16 meetings a week. It says 60% is voice, the rest is LinkedIn. You get email every now and then. But like, it's hard to get sales leaders to respond to an email because they get— I mean, they're just inundated every day.
Yeah.
And
oh, go
ahead. Tying it back to the prospecting agent as well, like depending on what response you're getting and where you're getting, it will prioritize steps accordingly. So for example, let's say you sent 3 emails and you did not get a response, you know, it will basically tell you, why don't you— like, you've done 3 touchpoints, the person is opening the email. but, or is clicking on the links per se, but has not really responded back, why don't you give her a call, right?
Yeah.
So the 4th step, it will automatically change from maybe like a LinkedIn step or another email step to a call. And therefore it's kind of like almost like your EA working for you to see what happened with this person. What do you need to do next to make sure you're booking a meeting? It's, it has one goal, say getting you to book a meeting. Goes after that goal with all the resources and channels and data available with it.
Yeah, it's fine. We were talking about this this morning too, of like, you know, the agents are just relentless on trying to hit their goals. And then, you know, if you look at something like the Anthropic data where they're like, okay, 80, you know, 80% of agents are just waiting on humans and they're just sitting idle. I'm like, yeah, it's because agents work 24/7, man. They're relentless. If you tell it to go book you a meeting, it's going to start doing everything it can to help you book that meeting. So.
Yeah, it's actually important the agent knows the time zone the contact is in so they're not reaching out at 3 o'clock in the morning and then everyone's like, why do you? So that's another thing we noticed.
Agents do have, sometimes they can struggle with time. I think they've gotten a little bit better for sure, but agents in the early, like even a year ago, they would get so confused on what, like even I would talk to my agent and be like, oh, I'll send it at 6 AM. And it'd be like, what do you mean? It's already past 6 AM because all the agents are in UTC. Right? Like they're not in Pacific time.
I have a very funny story about this one where I was actually on a vacation like last to last week watching fall colors. And I started a recruiting agent to basically go and prospect PMs in India and the time zone was completely opposite. And while I was vacationing, my agent was talking on LinkedIn with candidates, trying to give them information, trying to get them to book a meeting on my calendar. After 3 days I came back, I had 20 meetings booked, my calendar was full, and it was happening in a very different time zone because it was respective about their time zone and responding as quickly as possible, sometimes in like 2 minutes on LinkedIn, because it knows that if you don't respond fast enough on LinkedIn, you lose interest.
Yeah, it's an interesting one. One thing I wanted to ask you, Steve, because you brought up that you have the humans doing the cold calling still. How have you, you know, is this something that you now like filter for in the recruiting process for SDRs? Like, do you have them do cold calls since it is so important to your— Yeah. Yeah, you do?
Yeah, I mean, we don't just do cold call. I'll tell them, go to our website, don't use AI, and give me a prospect an email so that like it teaches them how quick they can learn what we do. And then like, can they actually do the manual craft? I feel like the human creativity is lost sometimes today with AI, and that's what you need to, To punch above the competition.
Yeah, no, it's, it's, it's so interesting. We were just talking, we recorded, um, I think it's coming out next week. We recorded with the CRO of Databricks and he was saying he, he filters SDRs coming into Databricks on very similar, like he makes them do cold calling, go to offices, like everything that obviously they have AI, it's Databricks, like they have AI too. And he's like, no, what I, what I search for in hiring is actually things that the agents can't do. Like, sure, agents can do account research, they can do automated emails, they can follow up in the right time zone, they can send, you know, hopefully the right message at the same time. But they're not, to your point, they're not going to pick up the phone when it comes time for step 4 in the sequence and it's time to call them and actually, you know, dial for it. And so it is something where it's so, it's so interesting that he looks for that maybe even more so now in hiring than previously. Yeah. Because It's, I feel like it's, and we were talking a little bit about how it's becoming like a lost art because so many of the new SDRs, like, grew up with AI. So like, how do you get them to do these skills that actually, these are the skills that the agents won't do or teach you? Yeah.
I mean, look, here's a sales leader that just replied to us this morning. Great pitching, call control. You're going places.
Yeah. Consensus is a good one.
Yeah. Like, and it's like, if you're in sales and you can't give a good pitch to a sales leader, like, It's kind of like you're probably not going to do well.
Yeah. Okay. I mean, the agents do help a lot, but they'll only take you, you know, part of, part of the way there. All right, cool. Well, thanks guys for doing this. For folks who have questions for you or want to try the prospecting tool and agent for themselves, um, what's the best way for them to do that?
Yeah, just go to our website and request a demo. We'll be in touch and you can take it for a test drive.
Awesome. Yeah. Thanks. I'm excited to get my hands on it as well. I want to, I want to benchmark it against our other agents. So I'm excited to do that next. So thanks guys for joining us and thank you everybody for watching our AI Day. This is our last session of the day. So thanks for joining. Hopefully this is super helpful and we will see you guys next time. Thanks guys.
Thanks.
Bye everyone.
More from Jason

How Clay, Figma, and Higgsfield Price AI, and How to (Finally) Get Your Pricing Right

Claudeforce vs. Replit vs. Lovable. Our Agents Pick Our Vendors Now. Who Wins?








