Episode 855 | The Eight SaaS Moats that Still Work in the Age of AI

Summary of Episode 855 | The Eight SaaS Moats that Still Work in the Age of AI

by Rob Walling

1h 0m•October 20, 2026
Listen to Episode

Overview of Episode 855 | The Eight SaaS Moats that Still Work in the Age of AI

Rob Walling and Einar Volset (Discretion Capital) deliver a deep-dive on how AI is reshaping SaaS defensibility, M&A, and product strategy. The episode argues that AI is not just a feature layer—it’s creating major supply-side and demand-side shocks that can commoditize software, shrink seat counts, or even erase entire customer segments. In response, they outline 10 vectors of AI attack and identify 8 moats that still appear durable in the age of AI, plus a few offensive moves founders can use to adapt and grow.

Why This Episode Matters

  • The hosts believe many founders are underestimating AI’s threat to traditional SaaS moats.
  • They also think the industry is overreacting in some ways and missing the opportunity in others.
  • Their goal is practical: help founders understand:
    • what AI can break,
    • what still protects a SaaS business,
    • and how to position companies for survival, growth, and stronger exit valuations.

The Core Thesis: AI Is a Supply Shock and a Demand Shock

The conversation splits AI’s effect on SaaS into two buckets:

Supply-side shocks

AI increases the supply of “good enough” software and services:

  • customers can build more themselves,
  • competitors can reach parity faster,
  • open source is becoming more usable,
  • platforms may copy features,
  • agencies can deliver custom work faster and cheaper,
  • model providers can absorb wrapper products.

Demand-side shocks

AI can reduce or eliminate the need for the underlying customer or workflow:

  • the customer may disappear,
  • seat counts may drop sharply,
  • or the workflow/artifact you manage may no longer matter.

10 Vectors of AI Attack on SaaS

Supply-side attacks

  1. Customers vibe-code their own solution
    • They often only need a small subset of your product, not the full 10-year build.
  2. Competitors reach feature parity faster
    • AI compresses the time needed to copy products and integrations.
  3. Open source gets “good enough”
    • Non-technical buyers can now deploy more open source than before.
  4. Platforms ship your feature for free
    • Large platforms can quickly add your niche capability and bundle it into their core product.
  5. Custom software at SaaS-ish prices
    • Agencies can offer highly tailored solutions with recurring support and maintenance.
  6. The model itself absorbs your product
    • Foundational models can make some wrapper SaaS tools obsolete.

Demand-side attacks

  1. Your customer disappears
    • If AI eliminates the work your customer does, your market may vanish.
  2. Your seats shrink dramatically
    • Seat-based pricing is vulnerable when AI replaces human workers.
  3. Your customer’s role disappears entirely
    • If the job is fully agentic, the buyer no longer exists in the same form.
  4. The artifact or step you manage stops mattering
  • If AI handles the workflow, tools built around that step can become irrelevant.

The 8 SaaS Moats That Still Work

1) Buyers know and trust you

A moat built from:

  • brand,
  • deep buyer knowledge,
  • partner ecosystem,
  • internal champions.

Why it matters: a better product still has to overcome trust and distribution advantages.

2) You are the system of record

Your software contains the canonical source of truth for a business.

Examples: operations, schedules, balances, permissions, memberships, records.

Why it matters: switching means exporting data, retraining teams, rebuilding workflows, and risking downstream breakage.

3) You are allowed to act

Your product doesn’t just recommend actions—it actually moves money, sends payments, executes tasks, or triggers real-world outcomes.

Why it matters: software that can act autonomously is harder to replace than software that merely advises.

4) They don’t check your work

Customers trust your output enough that they don’t verify every action.

Examples: payroll, taxes, compliance-related workflows.

Why it matters: if humans must constantly review the work, replacement is much easier.

5) Someone official already said yes

Regulation, licensing, certifications, and approvals create friction for entrants.

Examples: HIPAA, legal or medical compliance, CEU/credential-linked training, e-sign disputes.

Why it matters: AI can clone code, but not instantly replace approvals or legal risk management.

6) Everyone’s already in it / everybody has to move

This is the network-effect / multi-stakeholder adoption moat.

Examples: Slack, school platforms, multi-user operational systems.

Why it matters: the software only needs to be “good enough” if moving would require many people to coordinate and retrain.

7) Every customer makes you smarter

Each customer improves the product for all others through shared learning.

Examples: fraud detection, recommendation engines, operational intelligence.

Why it matters: this creates a compounding learning advantage that new entrants can’t easily copy.

8) It’s hard to keep running

Some products are not hard to build, but are hard to operate reliably.

Examples: deliverability infrastructure, scraping services, Cloudflare-style infrastructure, anti-spam systems.

Why it matters: the moat is in the operational complexity, not the code.

Key Takeaways on “Old” Moats

The episode argues that several classic SaaS moats are weakening:

  • unique feature set,
  • complex codebase,
  • years of engineering effort,
  • integrations,
  • UX familiarity,
  • and high switching costs.

These still matter in some cases, but AI reduces their protective power because:

  • code is easier to recreate,
  • agents can learn interfaces,
  • and competitors can ship faster.

The Offensive Playbook: How SaaS Can Win in the AI Era

1) Use AI to build faster

This is the table stakes move:

  • ship more features,
  • improve product velocity,
  • strengthen existing advantages.

2) Make your product agent-friendly

Founders should think about two strategies:

  • lock agents out if necessary,
  • or welcome agents via APIs/MCP-style access and monetize that traffic.

The hosts frame this as a new kind of SEO:

  • human SEO for buyers,
  • agent SEO for AI systems choosing which tool to use.

3) Become the worker, not just the tool

This is the big strategic shift.

Instead of selling software that humans use to produce an outcome, sell the outcome itself:

  • an AI agent that does the work,
  • a service that replaces labor,
  • a workflow that delivers results directly.

Why this matters: labor budgets are much larger than software budgets, so the upside is much bigger if you can capture part of the labor spend.

4) Build a learning loop

The hosts emphasize this as one of the strongest long-term advantages:

  • the product learns from actual use,
  • improves over time,
  • and becomes more valuable across the customer base.

This is especially powerful if the learning is:

  • hard to export,
  • grounded in real-world behavior,
  • and not easy to simulate with synthetic data.

Strategic Implications for Founders

For bootstrapped and small SaaS companies

  • Don’t assume “we’re niche” means you’re safe.
  • Don’t assume “customers can’t vibe-code this” is enough.
  • Start thinking about which moat you already have—or can build.

For companies considering M&A

  • AI defensibility is now part of the valuation story.
  • Buyers care more about whether a business can survive AI disruption.
  • Positioning and narrative around moat quality matter more than before.

For product strategy

  • Don’t just protect the existing product.
  • Look for adjacent workflows where you can own the outcome.
  • Think beyond seat-based pricing if AI is reducing seats.

Notable Themes and Insights

  • Leadership matters again: AI can meaningfully disrupt even established SaaS businesses.
  • Distribution and trust are becoming more valuable: especially for incumbents with real customer relationships.
  • The best businesses may become AI-augmented service-like products rather than pure software tools.
  • A lot of SaaS may still survive—but not all of it:
    • the hosts repeatedly suggest a meaningful percentage of SaaS companies will plateau, decline, or go to zero.

Resource Mentioned

  • microconf.com/moats
    • The episode’s companion resource page with:
      • summary materials,
      • slides from the Iceland talks,
      • and additional Q&A content.

Bottom Line

The episode’s central message is that AI is compressing old software advantages, but not eliminating defensibility altogether. The SaaS winners in the AI era will likely be companies that:

  • own trusted distribution,
  • sit on canonical data,
  • can act in the real world,
  • operate in regulated or hard-to-move environments,
  • and build products that learn continuously.

At the same time, the biggest opportunity may not be in defending the old SaaS model—but in turning SaaS into outcome delivery and capturing more of the labor budget.