Mar 18, 2026 · 1h 5m · saastr
10 Things to Know Before You Deploy Your First AI SDR
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this episode of the SaaStr Podcast, Amelia Ibarra and Jason Lemkin outline ten essential rules for deploying AI Sales Development Representatives (SDRs), focusing on replicating proven human sales playbooks, rigorous segmentation, and dedicated operational oversight.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Jason holds 18.8% of the talking time here. How this is scored →
Jason as informed peer 2.9 Guest teaching 0.6 Guest disagreement 0.1 Jason pushing back 0.4
speaking balance: gold is Jason, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 52:47 Amelia defending traffic volume thresholds
Amelia defends her high 10k-20k website traffic benchmark against Jason's challenge, explaining that inbound funnel math collapses without sufficient top-of-funnel visitor volume.
Hardest push from Jason ▶ 52:13 Jason challenging inbound visitor requirementJason pushes back against Amelia's 10,000 to 20,000 visitor threshold, arguing that early-stage startups without human coverage still benefit from having an agent answer inbound questions 24/7.
Biggest teaching moment ▶ 19:40 Amelia clarifying autonomous segmentation limitationsAmelia clarifies to Jason that no current AI SDR tool can autonomously generate high-converting hyper-segments without an operator manually configuring the audience parameters.
Jason holds their own ▶ 1:04:05 Jason on the SDR turnover and quality barJason delivers an incisive analysis of the SDR market, showing that AI SDRs do not need to outperform top reps to be viable, but merely exceed the consistency and retention of mediocre human SDRs.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Jason as informed peer | Guest teaching | Guest disagreement | Jason pushing back | Why |
|---|---|---|---|---|---|---|
| SaaStr Annual and AI Summit 2026 Announcement | 6 | 0 | 0 | 0 | Jason and Amelia introduce the session and outline Rule 1, emphasizing that AI SDRs only work when scaling an already proven human sales playbook. Jason contributes domain expertise on startup vs. enterprise failure modes, while Amelia outlines SaaStr's multi-agent setup. | |
| Rule 2: Automate Unglamorous and Monotonous Workflows | 5 | 0 | 0 | 0 | Amelia explains using AI SDRs for monotonous tasks like re-engaging neglected leads and sponsor portal follow-ups. Jason weighs in with operational color on build-versus-buy trade-offs for custom internal agents. | |
| Rule 3: Implement Ruthless Audience Segmentation | 5 | 2 | 0 | 1 | Amelia emphasizes ruthless audience segmentation rather than relying on a single large AI context. Jason asks whether any modern AI tools can perform this segmentation autonomously, which Amelia clarifies still requires human marketing management. | |
| Rule 4: Consistency Beats Brilliance in Execution | 0 | 0 | 0 | 0 | Amelia gives a monologue detailing Rule 4, explaining that consistent, 'pretty good' automated outbound messaging outperforms sporadic human attempts at perfection. | |
| Rule 5: Allocate Dedicated Human Oversight to Prevent Idling | 0 | 0 | 0 | 0 | Amelia delivers a monologue explaining why at least one to two dedicated humans are necessary to reload segments and keep AI agents from sitting idle once campaigns complete. | |
| Rule 6: Thoroughly Review Agent Outputs and Calibrate Prompts | 0 | 0 | 0 | 0 | Amelia walks through the necessity of reviewing almost every agent output during the first 30 days to catch branding quirks, hallucinated dates, and nuanced customer queries. | |
| Rule 7: Budget a Realistic Two-Week Ramp Period | 0 | 0 | 0 | 0 | Amelia explains why teams must budget a realistic two-week ramp period for domain warm-up, calibration, and context engineering rather than expecting instant turn-key deployments. | |
| Rule 8: Multi-Modal Formats and Communication Guardrails | 4 | 1 | 0 | 0 | Amelia details user preferences for chat over video/voice avatars and the strict guardrails needed for multimodal bots. Jason notes Delphi's similar metrics and suggests startups avoid video avatars initially. | |
| Rule 9: The Risks of Person-Dependent Deployments and Massive Scale | 0 | 0 | 0 | 0 | Amelia outlines the operational and legal risks of modeling AI agents on specific employees and highlights the massive session scale these agents can rapidly accumulate. | |
| Rule 10: Data Fundamentals, Traffic Benchmarks, and Lookalike Lists | 6 | 2 | 1 | 3 | Amelia argues inbound agents require 10k-20k monthly visitors to justify deployment, which Jason pushes back on as unrealistic for early startups. Amelia concedes the nuance and details using lookalike lists to scale outbound instead. | |
| Audience Q&A: Handling Drop-Offs, Prompt Injections, and SDR Quality | 6 | 1 | 0 | 0 | In the Q&A segment, Amelia discusses drop-offs, prompt injection attempts, and user sentiment. Jason closes with a strategic thesis comparing AI agent quality against the high turnover of mediocre human SDRs. |
Statements from this episode (21)
Ibarra: SaaStr operates over 20 AI agents in production
“Now we have, you know, more than 20 agents In our full stack, we've deployed more than one AISDR.”
Ibarra: Most companies only need one AI SDR vendor
“Most of what you will need to do slash want to do with your agents, you'll probably be fine using just one vendor. Versus using the four or five that we have for outbound.”
Lemkin: AI SDRs fail unless an outbound sales playbook is proven first
“If you haven't gotten any outbound to work, Buying an AI to do it isn't, it's not going to help you. You have to prove it. Any of the agents we use, any of the dozens of agents we use, we are taking a playbook that at least works a little bit.”
Ibarra: SaaStr Annual agency hours dropped over 90% via AI automation
“I was looking at the amount of hours our actual, our events production agency had done for January and February for this year for SAS Reannual that's coming up in May versus the amount that our agency had done last year in January and February for the SAS Rean…”
Lemkin: SaaStr uses Agentforce, Qualified, Artisan, and Monaco for AI SDRs
“For the most part, we are using the current four third-party agents that we use, right? We are using Agent Force from Salesforce, Qualified, which Salesforce is buying. We're using Artisan, which we use all the time as a core outbound agent. And recently we've…”
Lemkin: Salesforce is acquiring pipeline generation platform Qualified
“Qualified, which Salesforce is buying.”
Lemkin: Current AI SDR tools cannot autonomously auto-segment effectively
“None of the AI tools we use today, which are all great, but none of them can really use AI to auto segment in a way that can deliver these results.”
Ibarra: AI SDRs perform far better with hyper-specific audience segmentation
“We've run these for 10 months now, it just works a heck of a lot better to have accepted segmented and have the context that you're giving the agent hyper specific to that segment.”
Ibarra: SaaStr sent roughly 340,000 messages across Artisan, Qualified, and Salesforce
“I will tell you, we've sent this is just artisan, but it's sent 40,000 messages. Okay, I think our qualified is later on this deck that's sent like a 100,000 messages because that's an inbound. And then with Salesforce, we're probably close to like 200,000 mes…”
Ibarra: AI SDR messaging only needs to be pretty good, not brilliant
“Pretty good. I would say is the bar. It does not need to be brilliant. It does not need to be the best email on planet earth or even an email that you might write yourself. It just needs to be pretty good to get this person down the path you want them to go.”
Ibarra: Deploying AI SDRs requires at least one to two dedicated humans
“You need at least one, but ideally two essential humans to run all this and to deploy your AI SDR.
you can't have zero people.”
Ibarra: AI SDRs quickly finish sequences and idle without human-fed segments
“Your agents will sit idle if you don't give them new segments, new context, new people to go out to. It will finish its sequence in a couple days, and then it will sit there until you give it the next one.”
Ibarra: Teams must read all AI SDR outputs for 30 days
“Read everything. I would say in the first 30 days and then read almost everything. Every day after that. So if you're rolling out your first AISDR agent, you are going to want to read everything. And maybe you're going to want to read everything before you hit…”
Ibarra: Warming up domains and IPs for AI SDRs takes two to three weeks
“Some specifically to with an outbound agent, they can take a little bit longer, especially if it's warming up anything like specific IPs or domains or dedicated evil addresses. Sometimes that process just takes two to three weeks in and of itself.”
Ibarra: Buyers should distrust vendors marketing instant AI SDR deployments
“Nothing we have found is instant. So I would be wary of anything that you see is like an instant AISDR. Oh, roll it out today. I'd be wary of that for a lot of reasons.”
Ibarra: Most prospects prefer chat or email over voice/video AI SDRs
“And what we've found for the last year is that most people still prefer to chat with an AISDR. Over chat or over email versus either phone, voice, or video.”
Ibarra: SaaStr logged over 1.5M sessions on Qualified AI alone
“We've had more than 1.5 million sessions just overall, just in qualified, not counting our other agents with, you know, text-based communication or video communication.”
Ibarra: Inbound AI SDRs need 10,000 to 20,000 monthly website visitors
“I, if you have less than, let's say, 10 to 20,000 visitors a month, it's probably not worth it.”
Ibarra: SaaStr doubled CMO Summit attendees in one week via Clay lookalikes
“I happen to use Clay, and we had actually doubled the amount of CMO Summit attendees in a week just by using a lookalike.”
Ibarra: 99% of Inbound Prospects Are Satisfied Interacting with AI SDRs
“But for the most part, 99% of people are happy to talk to Amelia because we're busy. Like, I can't come answer your questions about tickets or Where are the hotel room block is? Or when did, when does the ticket code expire or when do prices go up and what's t…”
Lemkin: Well-Trained AI SDRs Deliver Better Interactions Than Mediocre Humans
“So what we found is if, and it might be hard to quantify that, maybe it isn't, but if you get above a certain line, if you invest the time to train your agent, if you read the email exceptions every day, if you upload all the documents, you're going to find th…”