Dear SaaStr: Should I Run 20+ AI Agents the Way SaaStr Does?
The honest answer: probably not entirely like us. But you should start now — because the gap between teams who have been running agents for 12 months and those who haven’t is compounding every week.
Let us give you the real picture.
What We Actually Built, and Why
In early 2025, SaaStr had zero AI agents in production. Two of our top salespeople quit at our Annual event in May. Instead of replacing them, we went all-in.
By end of 2025, we had 20+ AI agents running across every function:
- 3 AI SDRs — Artisan for outbound sequencing, Qualified for inbound chat and conversion, Agentforce for CRM reactivation, now Monaco too
- Digital Jason — a Delphi-based AI clone trained on 20M+ words of SaaStr content, answering founder questions 24/7
- AI Amelia — operational support for events and sponsor questions
- AI Content Reviewer — vetting speakers and sessions at scale
- AI RevOps (Momentum) — tracking sponsor health and renewal signals
- AI Matchmaking — connecting CEOs and executives at events
- 20+ vibe-coded Replit apps — pitch deck analyzers, valuation calculators, startup scoring tools
Today it’s 3 full-time humans plus those agents doing what previously required 20+ employees. Same revenue scale. Actually more revenue — growth reversed from -19% to +47% year-over-year.
But here’s what that summary leaves out.
The Part Nobody Tells You: It’s More Work, Not Less
The three people at SaaStr right now work harder than our 20-person team did in 2020. That’s not a failure of AI — it’s the nature of high-leverage work.
What changed is what they do. Before: manually formatting newsletters, scheduling social posts, updating spreadsheets, endless coordination meetings. Now: strategic decisions, relationship management, and orchestrating 20 AI agents that each require daily oversight.
Our Chief AI Officer (Amelia) spends 30% of her time — every day — on agent management. Reviewing conversation quality. Catching edge cases. Retraining on bad outputs. Updating knowledge bases as our business changes.
The AI SDR that handles sponsor outreach? We went through 47 iterations just to get it to stop being too aggressive on pricing. It took 30 days of daily tuning. The first 1,000 emails needed manual review.
When one of our agents quietly stopped ingesting new data four months ago — a bug — it kept running, kept producing outputs, kept looking like it was working. But it had gone stale. No error. No alert. We only caught it when results felt subtly off and we pulled the thread. The lesson: you can’t train an agent and go away. Even one.
Budget 30–60 minutes per day for every two or three agents you’re running. That’s not optional. That’s the job.
The Results That Made It Worth It
Here’s what that investment produced, with real numbers:
Digital Jason (Delphi AI clone):
- 2.75 million total conversations over 12 months
- 45-minute average session duration — not a chatbot, an actual thought partner
- 65,000+ founder questions answered
- Users return repeatedly, building on previous conversations about specific hiring decisions, compensation structures, board dynamics
Artisan (outbound AI SDR):
- 3,221 emails per month from a single platform — vs. 75–285/month per human SDR
- 11–12% positive response rates on warm audiences (recent event attendees)
- 5.5% on colder audiences — still at or above industry average
- 11–13x more responses from the same lead pools
Agentforce (CRM reactivation):
- Deployed to ~3,000 leads the sales team had written off entirely
- 72% open rate
- Already closing deals from contacts with zero prior follow-up
- Generated 15% of SaaStr AI London ticket revenue from a segment that previously produced $0
Qualified (inbound AI SDR):
- $1M+ in closed-won revenue in 90 days
- $2.5M in pipeline
- 71% of one month’s closed-won deals came from AI-qualified inbound (historic average: 29–34%)
- Booked a six-figure deal on a Saturday at 6:02pm — something no human team catches
Total investment: $500K+ per year in tools, platforms, and management time. Total return: $2.4M+ in closed-won revenue directly attributed to agents, 8-figure overall revenue, and a business that runs on three people.
What the 20-Agent Stack Actually Requires
Before you try to replicate this, understand what you’re signing up for:
A dedicated AI operator. We have Amelia as Chief AI Officer. Someone whose primary job is managing the agents — not as a side project, not as 10% of a marketing manager’s role. If you don’t have someone whose job this is, start with one or two agents maximum and build from there.
Deep ICP and messaging work upfront. AI SDRs amplify your targeting precision in both directions. Sharp ICP → more of your best customers. Vague ICP → spam cannon that burns your domain. Do the work before you turn anything on.
Human-written frameworks, AI execution. The emails that book meetings are built on templates your best human SDR validated. The AI handles personalization, timing, and sequencing — but it’s working from messaging a real person crafted. Don’t let vendors write your templates.
A CRM as the center of gravity. Every agent we run reads from or writes to Salesforce. This creates real switching costs — agents trained on your data don’t transfer — but it also creates compound value. Every week of agent operation builds institutional memory that makes them harder to replicate.
Patience for a 60–90 day ramp. The first month is rough on every new agent deployment. Month two is when you see real signal. Month three is when you know if it works. Don’t judge any agent in week one.
Where to Start If You’re Not SaaStr
You don’t need 20 agents. You need one that works.
- If you have inbound leads you’re not following up on fast enough: Qualified or a similar inbound AI SDR. Speed-to-lead is still one of the highest-leverage variables in conversion, and no human team matches AI on response time at 2am.
- If your CRM has dead contacts nobody’s worked: Agentforce or a CRM-native reactivation tool. These leads already showed interest. The activation cost is low. The upside is direct.
- If you want to scale founder knowledge: A Delphi-style AI clone trained on your best content. If you’ve been writing, recording, or speaking for years, that corpus is an asset sitting idle. Digital Jason turning into a 2.75M conversation engine happened because the training data existed.
- If you want to go outbound at scale: Artisan, Apollo AI, or a comparable platform. But read the setup requirements above. This is the hardest agent to do well because bad targeting + AI volume = domain damage.
Start with the motion your team isn’t running at all — not the one they’re already doing. The leads your SDRs consider too small. The inbound you can’t respond to fast enough. The CRM contacts nobody has touched in a year.
That’s where the ROI is clearest and fastest, and it’s exactly where we found ours.
The Real Question
“Should I run 20+ AI agents like SaaStr does?”
Not on day one. But here’s what I’d ask instead: what’s the one motion you’re leaving completely on the table right now because you don’t have the people or the time?
Start there. Do it right. Give it 90 days.
If it works — and it usually does, when set up correctly — you’ll have your answer on whether to scale.
The companies that started this 12 months ago are already compounding the advantage. The ones that start this month will catch up. The ones that wait another six months will be behind in ways that are hard to close.
Have a question for SaaStr? We answer the real ones at SaaStr.ai.
