So about a year ago, we built “Digital Jason” — our first AI Agent. It was all we had until June/July 2025 — and we learned so, so much from it.
Digital Jason is an AI version of me, trained on:
- 13 years and 4,600+ posts of SaaStr content
- thousands of founder Q&As
- every podcast, every SaaStr AI session, every Workshop Wednesday
- and pretty much everything I and all our speakers know about scaling B2B companies from $0 to $100M+.
The idea was simple: I can only take so many meetings. I can only answer so many questions at Annual, on X, in the newsletter. But founders have the same questions at 2am that they have at 2pm. What if there was a version of me that could help anytime, anywhere?
Here’s what happened.
Digital Jason just crossed 2,756,758 total conversations over the past 12 months. Up from essentially zero the year before.
Let’s break that down, because I think there are real lessons on scaling your first AI Agents. From a horizontal one that does it all (or tries to), to 30+ AI Agents that now specialize.
The Numbers
Here’s the 12-month view:
- 2.76M total conversations with Digital Jason
- 2,048 highly active users engaging for hours regularly (our super power users)
- 371,156 total messages exchanged
- Average session duration: 45 minutes and 35 seconds
- 6,185 hours of Digital Jason time created
And here’s what February 2026 alone looked like:
- 460,913 conversations in a single month
- 39,104 messages
- Average session: ~14 minutes
- 651 hours of agent time
That February number tells the real story. We’re now running at roughly 460K conversations per month, which annualizes to north of 5.5 million conversations. That’s 2x the trailing 12-month total already.
Let”s be clear: if nothing else, there’s no way I could have had all those conversations … myself.
Try it here (it’s 100% free).
The Evolution: From Everything Agent to Expert Agent
In the early days of our AI Agentic Journey (early 2025), Digital Jason did everything. Because it was our only agent. Support tickets? Digital Jason. Helping sponsors figure out which package to buy? Digital Jason. Qualifying inbound leads? Digital Jason. General SaaStr questions? Also Digital Jason.
It was the only AI agent we had, so we threw everything at it.
And it worked — kind of. The problem is that one agent trying to do everything is like one employee trying to run sales, support, ops, and advisory simultaneously. The quality suffers. You can’t optimize a single agent for “answer a founder’s deep question about when to hire a VP of Sales” AND “help a sponsor figure out booth placement for Annual” at the same time. Different jobs, different data, different tone.
So we did what any good B2B company would do: we specialized.
Over the past year, we built out 20+ dedicated AI agents to handle all the jobs Digital Jason was never meant to do. Three Artisan AI SDRs running outbound campaigns (one for cold, one for warm, one for event-specific — with 5-7% response rates). A Qualified inbound chat agent that’s driven $1M+ in closed revenue. An Agentforce agent resurrecting dead leads at a 72% open rate. Monaco for creating true outbound. A Momentum RevOps agent syncing everything to Salesforce. An AI content reviewer vetting 1,000+ speaker submissions. An AI matchmaker connecting CEOs at events. And more.
Each one trained specifically for its job. Each one managed daily. Each one getting better every week.
And that freed Digital Jason to do what it actually does best: help founders think through how to scale their B2B + AI businesses.
That’s what it focuses on now. The deep questions. “I’m at $3M ARR and my lead investor wants me to pivot to AI-native. Is it too late?” “How do I structure comp for my first AI sales rep?” “What does good NRR look like when half my revenue comes from AI agents?” The stuff that requires 13 years of pattern-matching, not a product FAQ.
Here’s what happened.
Digital Jason just crossed 2,756,758 total conversations over the past 12 months. Up from essentially zero the year before.
Let me break that down, because I think there are real lessons for every B2B founder thinking about deploying AI — and especially about what it means to “clone” your best expert.
The Hockey Stick Is Real
Look at the trajectory. From March through about October, Digital Jason was averaging 5K-9K daily conversations. Respectable, but linear. Then something broke through around November-December, and by January-February we were hitting 18K-22K conversations per day.
That’s not incremental growth. That’s a step-function change.
What drove it? A few things, all compounding:
- Digital Jason got smarter. Every month we shipped improvements — better prompts, richer data, better conversational flows. The SaaStr.ai valuation tool alone processes hundreds of thousands of startup valuations monthly now. When an AI expert actually gets better every week instead of every decade, usage compounds fast.
- Distribution kicked in. 400K+ newsletter subscribers and a strong organic presence mean new tools spread fast when they deliver real value. We didn’t spend a dollar on paid acquisition. Digital Jason grew entirely on word of mouth and utility.
- Session depth tells the whole story. 45 minutes average session duration is not a chatbot metric. That’s a working session. That’s a founder running valuation scenarios, iterating on their pitch deck, pressure-testing their go-to-market strategy — with a version of me that has 13 years of context and infinite patience. Nobody spends 45 minutes with a gimmick.
- People told their work friends. That always works.
What “Cloning Yourself” Actually Looks Like
Here’s the thing nobody tells you about building a Digital You: it doesn’t replace you. It scales the parts of you that are already documented.
I’ve written 4,600+ posts. I’ve answered thousands of questions on stage, on video, in comments. That corpus is massive. And it turns out that maybe 80% of the questions founders ask me are questions I’ve already answered — sometimes dozens of times. “When should I hire a VP of Sales?” “How do I price my product?” “What does good NRR look like at $5M ARR?”
Digital Jason handles that 80% now. And it handles it well — with specific data, real frameworks, and the right nuance. Not generic AI slop. Real SaaStr-grade answers.
That frees up the real me to focus on the 20% that actually requires a human: the deeply specific strategic questions, the relationship-driven intros, the judgment calls that only come from pattern-matching across 100+ portfolio companies.
What This Means for Every B2B Founder
If you’re a founder or CEO who has spent years building expertise — through content, through customer conversations, through hard-won pattern recognition — you’re sitting on something incredibly valuable that you probably haven’t thought about this way.
Your knowledge is an asset that can be cloned.
Not perfectly. Not for everything. But for the 80% of repetitive, high-value-but-time-consuming interactions that eat your calendar alive? An AI version of your best expert can handle it. Today. Not in some theoretical future.
Here’s what we learned building Digital Jason:
- Depth of training data matters more than model sophistication. We didn’t build some custom model. We fed 13 years of real content into well-engineered prompts and workflows. The content was the moat, not the AI.
- Session duration is the only metric that matters. If your AI clone’s sessions average 2 minutes, you built a FAQ bot. If they average 45 minutes, you built something people rely on for real work. Optimize ruthlessly for depth.
- Compound effects are massive. We went from ~5K daily conversations to ~20K in about 4 months. That’s 4x growth with essentially zero marginal cost. Try scaling a human expert team that way.
- The economics are absurd. Digital Jason, along with our other 20+ AI agents, is generating $1M+ in revenue. The cost to run them is a fraction of a single employee. This is a big part of why our growth flipped from -19% to +47% year-over-year.
Clone Your Most Valuable Expert
Most B2B companies are still treating AI as a feature they bolt onto their product. They’re shipping “AI-powered” badges while running their internal operations the same way they did in 2023.
We went the other direction. We cloned the most valuable expert in the company — me — and deployed that clone to handle millions of conversations that I physically could never have. Then we built more agents around it. 20+ and counting.
2.7 million conversations. A 45-minute average session. A hockey stick that’s still accelerating.
That’s not a chatbot. That’s a digital expert that works 24/7 and gets better every single week.
And Digital Jason is be better at answering most of these questions than I am at this point. He’s read every post I’ve ever written. I definitely haven’t.
