Rob Walling's 8 SaaS Moats That Still Work Against AI: His MicroConf Talk Answered, With Timestamps
In The Only SaaS Moats Left in 2026, a MicroConf talk published in October 2026, Rob Walling asks how a software company survives once everyone has AI. His answer sits between "SaaS is going to zero" and "nothing has changed": AI can attack a SaaS business from ten directions, most of the classic reasons customers stay are wearing away, and eight moats still hold. He says a company with even one or two of them is probably doing well.
We transcribed the talk with transcribe.so and wrote each answer below in our own words, pointing to who said it and the exact second, so you can check it in the video. The percentages are his own estimates and examples, and we have not checked them. We are not affiliated with Rob Walling, MicroConf or TinySeed, and we sell a transcription service.
| Who is speaking | Role |
|---|---|
| Rob Walling, co-founder of MicroConf and TinySeed | Gives the talk |
Explore the whole talk below: its chapters, a summary of every section, search, and five questions we answered from the transcript, each opening the video at that moment. It also has its own page at /demos/rob-walling-saas-moats-2026.
Command Palette
Search for a command to run...
Chapters, section summaries and timestamps above are unedited transcribe.so output, so a few names are misheard ("IntervalSet" is Einar Vollset, his co-author). The questions and answers were written by us from the transcript, with each claim linked to the part of the talk where it is said.
What is the talk about?
How a SaaS company survives in the age of AI. Walling says the social media story that all SaaS is going to zero is wrong, but so is ignoring the change (Walling, 01:08). His own estimate is that somewhere between 10 and 20 percent of SaaS companies will decline or go to zero (Walling, 06:07). He co-wrote the talk with Einar Vollset, who runs a sell-side M&A advisory for SaaS companies and looked at the question from the buyer's side (Walling, 16:47).
What are the ten ways AI attacks a SaaS business?
Six hit supply, meaning more products like yours appear:
- Your customer builds it. They do not need your 50 features, only the three they use (Walling, 04:17).
- Competitors reach parity fast. He counted 46 AI note takers, almost all at feature parity, and says he could build a near-equal one in a few months (Walling, 05:01).
- Open source moves faster, and non-technical buyers can now run it with an AI assistant's help (Walling, 05:23).
- A platform you plug into ships your feature, because it can now build it in weeks and already has the customers (Walling, 06:27).
- Agencies sell custom software at SaaS-like monthly prices, since building it now takes a fraction of the time (Walling, 07:04).
- The model absorbs your product, as happened to apps that generated ad copy and images (Walling, 07:50).
Four hit demand, meaning fewer people pay for what you sell:
- Your customers go out of business. His example is translation and transcription agencies whose human work AI now does (Walling, 08:19).
- Seats disappear. A support team of eight people can become one or two people supervising AI agents (Walling, 09:39, 10:08).
- The thing you manage stops mattering, like software for take-home coding tests now that candidates use AI (Walling, 10:45).
- The human user is replaced by an agent. Software for receptionists or call centers has to add an agent itself, or someone else will sell the software plus the agent (Walling, 11:46).
"If you don't agentify your software, build some type of agent on top of it, I think you're going to zero in the next couple years." (Walling, 11:46)
Which classic moats is AI wearing down?
He lists nine reasons customers used to stay: unique features, a strong brand, owned traffic such as SEO, clever code, a UX the team knows, many integrations, high switching costs, many people coordinating in the app, and being the system of record (Walling, 12:29). AI shrinks most of them. Features are faster to copy, products converge, SEO gets muddier, clever code and integrations are easier to rebuild, and an AI agent can pull data out of an app even when there is no export (Walling, 13:57). Three survive: being the system of record, coordination between people, and brand plus the things around it (Walling, 15:30).
Why should a bootstrapped founder care about valuations?
Private equity makes about 70% of the acquisitions of companies like these, and buyers are now unsure whether a company will still be around in three to five years. The moats are what gives them that confidence, and even if you never sell, they are your defense (Walling, 16:47). Public markets already show it: he says HubSpot beat its earnings expectations and its stock still fell 21% (Walling, 18:17). He also stresses that none of the moats is on or off; score yourself from zero to ten and work on moving up (Walling, 19:50).
What are the eight moats that still work?
For each one he tries to give a big example, a smaller bootstrapped one, and a question to ask about your own product.
- Buyers know and trust you: brand, partners who recommend you, a champion inside the customer. He dislikes using QuickBooks but stays because his accountant uses it, and a better product does not get you into that conversation (Walling, 20:17). Ask: who has built their business or career around recommending or defending you? (Walling, 21:03)
- You are the system of record: the customer's daily operations run on you, with balances, approvals, statuses and other systems wired in. His examples are Jira and a TinySeed gym management app that grew to run nearly every part of a gym (Walling, 21:16). Ask: if your customer wants to leave, what can't they export? (Walling, 23:17)
- You're allowed to act: a tool that moves the money, not one that recommends a payment. Stripe Connect is his big example. Another is a churn tool that makes the save offer itself inside the cancel flow, instead of only flagging who might leave (Walling, 23:27). Ask: can your software act, or can it only advise? (Walling, 25:17)
- They don't check your work: Gusto runs his payroll, and he has never once looked at anything beyond the W-2 it issues him (Walling, 25:25, 25:46). Avalara does the same for sales tax. This trust is earned by being right for years, so a vibe-coded clone does not win the switch.
- Someone official already said yes: licenses, accreditations, certifications such as SOC 2 or HIPAA, or a board that has to approve you as a vendor (Walling, 26:56, 27:52). His small example sells courses that count toward a professional credential, which a copycat would need board approval to match. He adds US text messaging, where carrier registration and consent law are the moat rather than the code (Walling, 29:14).
- Everyone's already in it: the more people coordinate in your software, the more of them must agree to move. Slack is the big example (Walling, 29:53). The small one is a swim team app with every family signed up, so switching means the whole league has to agree (Walling, 30:52).
- Every customer makes it smarter: Stripe Radar learns fraud patterns across all merchants (Walling, 31:52). It only counts if three gates hold: each customer improves it for every other customer, customers can take their data but not what you learned from it, and the system keeps learning (Walling, 32:44).
- It's hard to keep running: a well-known indie maker who builds everything himself tried for weeks to replace a scraping proxy service and ended up paying for it, because the hard part is keeping it from being blocked (Walling, 33:56). At Drip, Walling had full-time people whose only job was fighting spam (Walling, 35:32). Ask: is it hard to build, or hard to keep running? (Walling, 35:58)
"No single merchant could ever build it, right?" (Walling, on Stripe Radar, 31:52)
How do you score your own business?
He turns the moats into eight questions, one each (Walling, 36:26). The one he likes best:
"And what do you maintain that you resent maintaining? That's my favorite." (Walling, 37:06)
Then he suggests one move toward each moat: a vertical tool builds an ecosystem of people who recommend it, a point solution becomes a system of record, a reporting tool learns to act, a calculator becomes the output nobody double-checks, a company in a regulated industry becomes an approved provider, a single-seat tool builds coordination between users, a company that collects data builds a learning loop, and infrastructure you babysit becomes a burden nobody else wants to carry (Walling, 37:28).
"Probably doing pretty well if you have one or two of these, I think." (Walling, 38:32)
What to copy
- Score yourself from zero to ten on each of the eight moats, then pick the one you are closest to and make it stronger. A moat counts even when you are a small company.
- If your users are people an agent could replace, ship the agent yourself before a competitor sells your software plus one.
- Treat the work you resent as a possible moat. Anti-spam, uptime and compliance are exactly what a vibe-coded competitor does not want to take on.
More from this series: Jason Cohen's ideal bootstrapped business, answered, every post in SaaS founder talks, answered, and our free YouTube transcript tool if you want to check a talk like this yourself.