Is SaaS dead and distribution the only moat?
And is distribution the only moat left?
We are back with Episode 14, and honestly, we’re hitting our stride. We were joking right before hitting record that we wish we had started this five or ten years ago. But as Austin pointed out, the game has changed. In a world where AI lets anyone build anything, the only moat left is distribution.
If everybody can create anything with AI apps then your advantage is your distribution... if you have no ability to distribute you can make whatever you want but how does it get found? — Austin
This week was packed. We covered the chaos of desktop agents, the “Mad Men” style advertising wars between Claude and ChatGPT, and the absolute monster earnings from Big Tech.
Here is the breakdown.
The Rise of “OpenClaw” (fka Moltbot)
If your feed was anything like ours this weekend, you saw Moltbot (or Claudebot, or now “OpenClaw” after a likely cease and desist) everywhere.
This isn’t just another chatbot in a browser tab. This is an open-source agent that installs locally on your computer and has “hands.” It can move your mouse, type for you, and read your files. It’s proactive, meaning it doesn’t just wait for a prompt—it can ping you on WhatsApp to tell you you’re late for a meeting.
The creator, Peter Steinberger, is embracing the chaos.
The guy’s just arrived... he’s like ‘Oh yeah I’ve made my money... this is pure open source go nuts.’ It’s just complete anarchy. — Pranav
While the security implications of giving an AI full access to your local machine are terrifying to some (Austin included), the potential for a centralized context that connects your files, apps, and preferences is the holy grail we’ve been waiting for.
The Ad Wars: Claude vs. Everyone
We watched a fascinating ad from Claude that positions privacy as their killer feature. The premise? A guy asking AI for advice on talking to his mom, only to get served a hyper-targeted ad for “mature dating” based on his private query.
It was hilarious, but it sparked a debate between us.
Austin loved the execution:
It captures... that GPT is pretty lazy... and the ad at the end was just the cherry on top... telling you that you’re sharing private information which could be utilized to advertise for something that’s completely unrelated.
But Pranav played devil’s advocate. Does the average person actually care about privacy? Google is a $3 trillion company built on our data.
Do people really care about their privacy... to the degree that Claude is like... saying? ... I mean the numbers say otherwise
Follow the Money: Google, Meta, and Accenture
Speaking of Google, their earnings were absurd. They hit $400 billion in annual revenue for the first time. YouTube ads alone are generating $60 billion.
The most interesting stat? 50% of YouTube viewership is now happening on TV screens. The lines between “digital video” and “television” are completely gone.
Meta is also making moves for 2026. They are going all-in on Reels. They are surfacing 25% more same-day Reels, and they are pushing the “TikTok-ification” of their algorithm hard.
Then there is Accenture. They reported that AI revenue jumped 100%. They are shifting from “projects” to “end-to-end solutions.” Essentially, they are becoming the “human layer” for managing AI agents for the Fortune 500.
The “SaaS Crash” & Vibe Coding
There was panic on X this week about SaaS stocks crashing (Figma, etc.) with people claiming AI “vibe coding” will replace software subscriptions.
Austin’s take? If it’s a simple wrapper or a utility, you should probably just build it yourself.
There’s almost no SaaS application that you can’t build with enough time... so the question is like is the value that you’re paying worth the value received from not having to go build it with an agent.
However, complex systems like CRMs or two-sided marketplaces (like Day AI raising $20M for autonomous CRM) aren’t going anywhere. You pay them for the network effect and the maintenance, not just the code.
Other News (The Pulse)
We didn’t get to everything live, but here is what else is happening in GTM and Tech:
Publicis Targeting LiveRamp: Rumors are swirling that Publicis Groupe is looking to acquire LiveRamp. This would be a massive move for a holding company to own the “Identity Layer” of the internet.
Amazon’s Agentic Advertising: Amazon launched a Model Context Protocol (MCP) server. This is “plumbing” that allows AI agents to buy and manage ads directly, moving us toward machine-to-machine marketing.
The Trade Desk Plummets: TTD stock dropped significantly, signaling a potential “AdTech Reset” where performance is judged by business outcomes rather than just programmatic scale.
LinkedIn Video is Booming: Video ad revenue on LinkedIn is up 30%. If you aren’t doing B2B video, you are falling behind.
Shoppable YouTube on CTV: YouTube officially rolled out shoppable formats for TV. You can now buy products directly from your television screen via QR codes, bridging the gap between awareness and performance.
Google Ads API v23: A nerdy but huge update—channel-level reporting is finally available for Performance Max. We can finally see where the money is actually going (Search vs. YouTube vs. Display).
See you next week. Go build your distribution.
I agree that distribution is a moat but then you need to rinse and repeat with retention. One example (and I have spoken about this to a recent vibe coder who came out of retirement at 67-years old after decades as a CEO because he was bored): a vibe coder creates an already existing solution but is able to distribute to a low-price, underserved market that is well-below the economic return point for the incumbents. The quality of the product can go upstream but the solo vibe coder can continue to penetrate the underserved market with a super-low cost solution. Then they have to focus on retention, but what's interesting at this level is the switching cost can become higher than the reduction in price provided by the next vibe coder with a lower price. So retention might be less of an issue than we might suspect.