[Un]Churned · 20 Feb 2026 · From the week of 16 February
75% of SaaS Companies Will Disappear ft. Brett Queener (Bonfire Ventures)
These are notes on the conversation, checked against its transcript. The episode itself has the full discussion.
In brief
Host Josh Schachter interviews Brett Queener, Managing Director of Bonfire Ventures and an early Salesforce executive, about why traditional SaaS moats have become table stakes. Queener revisits his prediction from 18 months earlier that 75% of customer-facing software companies would disappear and says he is still writing about it. He argues that the defensible advantage left is an embedded judgment layer in agentic products, and that buying software increasingly resembles hiring an employee. The episode also covers how the role of operators shifts from doing the work to reviewing and managing agents.
For founders
- Brett Queener says systems of record and accumulated context used to be rights to win but are now just the price of entry, which he calls a right to play rather than a right to win.
- Queener says the defensible layer in agentic products is the judgment layer: the organization's way of working captured in the product and fed back to the user over time.
- Queener says some of his founders now have to find and refine product-market fit every three to six months, which he contrasts with five years earlier.
- Queener's biggest prediction from 18 months ago is the return of the buy-versus-build dilemma, which he thinks will eliminate many software companies that do not deserve to survive.
- Queener says agentic products improve quickly with user feedback and retain that feedback in memory, so users keep giving it.
For revenue leaders
- Queener says buyers of agentic software should evaluate it like hiring an employee, asking whether they want to hire it rather than whether they want to buy software.
- Queener says that in products requiring configuration, user frustration may reach a CSM as a logged request and show up in the product months later.
- Queener says leaders should get hands-on with AI tools, starting with vibe coding an app, so they understand what is now possible.
What was said 13, most useful first
Traditional SaaS moats such as the system of record and accumulated context are now table stakes, a right to play rather than a right to win. Listen
He says moats he once thought would let a company win a game have become the price of entry. He gives a basketball analogy: being fit gives you the right to play but does not win the game. He presents this as a shift he and others have experienced over the last 18 months.
“They're just table stakes. So you don't have them, you can't play, but they're not a right to win”
Listen to the episode Strategy & market Link to this Report a problem
Buying makes sense when a product brings in best practices that a buyer cannot easily replicate and lets the buyer inject their own way of working. Listen
He gives a hypothetical: a personal-injury (PI) lawyer gets a product from a team of 250 people who only work with PI lawyers, so 80% of the hard work is already done, with best practices built in that improve as it works with more PI lawyers. He says the buyer can then inject the way they want to work. The example is hypothetical rather than a customer he describes.
“And then it allows me to inject the way I want to work, the way I write, the way I look at cases. Yeah, I'll buy that software.”
Queener predicts that the return of the buy-versus-build dilemma will eliminate many software companies that do not deserve to survive. Listen
He calls the return of buy versus build the biggest prediction he made 18 months earlier. He says he thinks it is good news because it removes frauds and companies that do not deserve to exist. He states this as his view.
“this is the biggest prediction that I'd made 18 months ago, which is the return of the buy versus build dilemma. And it's very real.”
Listen to the episode Strategy & market Link to this Report a problem
Queener defines a judgment layer as the question of whether an agentic product ships with enough embedded judgment that a customer would not build it themselves. Listen
He frames it as a baseline test: does the product know enough about the skills the customer wants done that it is better to use it than build it. The bigger point he describes is capturing how an organization or person wants to work and feeding that judgment back to the user. He says this was never the case in SaaS or enterprise software.
“Does it ship with judgement? Does it know enough about the skills that I want to get done if we're moving in an agentic world”
Users give agentic products more feedback because the product improves quickly from it and keeps that feedback in memory. Listen
He describes writing his article with agents, which improved quickly when he gave them feedback and retained it. He says this makes users more willing to keep giving feedback repeatedly. He cites his own experience with the agents rather than a broader measurement.
“When you give them feedback, they improved so quickly and they were, and that memory stays there that you're much more willing to dig in and give it feedback”
The buy-versus-build question should be reframed as whether you want to hire an employee or an engineer. Listen
He says people should stop talking about buying software and instead think of it as hiring an employee. He presents this as the way he thinks about it, and he describes the return of the buy-versus-build dilemma as the biggest prediction he made 18 months earlier.
“stop talking about buying software. Do you want to, you're hiring an employee. And so the question is. Do you want to hire an engineer?”
Listen to the episode Strategy & market Link to this Report a problem
The human role is moving from operator doing the work to manager and reviewer of agents that are assigned outcomes. Listen
He describes agents as semi-autonomous employees: rather than giving step-by-step instructions, you assign an outcome, and the agent builds a team to do the work. He says the human's role shifts to review and management. He presents this as the new era he believes is underway.
“you assigned an outcome. It can be a big outcome. And then you tell it to build the team to work with you, to go do the work.”
Queener advises leaders to get hands-on with AI products, starting with vibe coding an app, so they can see how the old workflow of PRDs and handoffs breaks down. Listen
He says that vibe coding an app shows the old world of writing PRDs, handing them to dev, and waiting through information loops is slower than iterating directly. He says it is almost irresponsible for leaders not to do this. He suggests starting with vibe coding before moving to Claude Code.
“start with vibe coding, vibe code an app like, oh, this is pretty cool.”
Listen to the episode Leadership & culture Link to this Report a problem
A product that requires users to configure settings or workflow rules can turn friction into requests that reach a CSM and may reach the product months later. Listen
He describes a user who has to change personal settings, configuration or workflow rules, and says the frustration gets logged as a request or heard by a CSM. He says it may show up in the product nine months later. He offers this as an example of how feedback travels in the old model.
“what that gets logged as a request or CSM hears it. Maybe it shows up in a product nine months later”
Listen to the episode Retention & customer success Link to this Report a problem
Some of his founders have to find and refine product-market fit every three to six months, which he says was not the case five years ago. Listen
He presents this as part of the journey that everyone has been feeling, describing founders chasing a thesis that keeps shifting. He gives the three-to-six-month cadence as what some of his founders face now. The figure is his description of his portfolio.
“some of my founders have to find and refine product market fit every three to six months. That wasn't the case five years ago.”
Listen to the episode Strategy & market Link to this Report a problem
His forecast from 18 months ago that 75% of customer-facing software companies would disappear, with 50% of categories disappearing at the same time, has held up well enough to keep writing about. Listen
Queener says he had AI agents analyze whether he was right enough to deserve writing a new piece, and the first half of that piece reviews how the moats he expected have changed over the last 18 months. He frames the prediction as a bold call he made 18 months before the episode. The claim is his own assessment of his prediction.
“I thought I was prescient like I wrote 18 months ago, that 75% of all customer facing software companies would disappear and 50% of all categories would disappear at the same time.”
Listen to the episode Strategy & market Link to this Report a problem
Queener describes the last two and a half years as a copilot era and the last 60 days as the start of an agent era led by tools such as Claude Code and Claude Cowork. Listen
He says the copilot era had a human instructing an assistant at each step. He says the new models are using their own code-building agents to build later versions of themselves. He presents this as a shift he has observed in the last 60 days.
“We're in a brand new era. What I'll call the Claude code Claude Co work.”
Good judgment, including from people trained in the humanities, will matter more in the AI era. Listen
He describes this as the revenge of the dilettantes after colleges shut down humanities and classics programs to produce STEM graduates. He says that if you have good judgment, the new tools reward you. He presents this as a view from the judgment-layer argument rather than a measured result.
“It's revenge of the dilettantes. Like if you have good judgement, whoo wee.”
Listen to the episode Hiring & team building Link to this Report a problem