If code was never your moat, this is your moment
Gil Dibner

The tech and venture market is reeling from the "SaaSapocalypse." While I’ll leave the public market forecasting to the analysts, one thing is clear: the AI moment has ruthlessly exposed the inefficiencies of the SaaS era—specifically, the massive compression of multiples for companies whose only moat was a recurring subscription model. It has destroyed the illusion that the ability to code up a glorified CRUD application is a sustainable barrier to entry.

Of course, that was never the full story. While most SaaS success stories were—ultimately—CRUD applications, the reasons for their growth were many and rarely had to do with their ability to generate code. It was always about PLG or sales-driven GTM success, subtle points of product perfection, network effects, brand, and reputation.

Still, there was enough confusion around the sources of success that engineering teams ballooned to irrational sizes. Funding rounds ballooned. Headcount counted as a success metric. Many SaaS applications received multiple rounds of venture funding and unicorn valuations pursuing strategies that never could have led to sustained value over a long time horizon. The moats simply weren’t there. We may or may not be in an AI bubble, but we were definitely in a SaaS bubble and it’s definitely over.

Hard Tech, Hard Markets

At Angular Ventures, we’ve got the idea of “hard tech, hard markets” hard-wired into our consciousness. As a result, we can confidently say that we didn't just avoid entering new investments at peak valuations—we also avoided investing in anything that had a whiff of a “glorified CRUD application.” Before, during, and after the AI moment, we have invested only in companies with a real moat (be it technical or otherwise). What makes that perennially difficult is that as technology advances, the definition of a "technical moat" keeps rising. We’ve got to raise our bar even faster to keep up.

One consequence of this, however, is that if you look back at our portfolio over the past few years, there are a number of companies that share a common set of features: in essence, these are companies chasing hard markets. They sell to discerning enterprise customers with high standards and very demanding requirements for any vendor. These are customers for whom flashy demos are worthless and procurement decisions are a labyrinth of stakeholders and procurement. We are talking oil & gas, logistics, procurement, shipping, etc. Often, this means long sales cycles, complicated relationships, internal politics, advanced integrations, and very high product specs. This stuff can’t be vibe-coded, and—even if it could be—the code itself is barely 10% of the value the company creates for shareholders.

The litmus test is simple: if we gave 20 brilliant Stanford students $100M, could they build the business? Not the code itself, but the business? If the answer is yes, the moats are likely too shallow. If the answer is no, the market, the tech, or both might be hard enough for sustainable value to accrue.

Engineering has moved from obstacle to accelerant

For companies in this category—those serving hard markets and winning in them slowly but surely—the SaaSapocalypse moment is both frightening and filled with promise. It is frightening because of the real fear that software revenue—however difficult to obtain—is being treated as essentially worthless. To my mind, that is a panic-induced generalization that only captures a portion of reality; many companies will thrive in this climate as markets adjust.

For founders able to put their irrational fears aside, today’s climate is actually a huge opportunity. For companies in this category, the generation of code has historically been a cost center and a source of drag. AI-accelerated coding is a huge unlock for these companies. In the go-go times, the slow apparent progress of these companies was a source of consternation for some. Today - against the backdrop of a world questioning the value of code itself - their ability to navigate complex sales cycles is a source of comfort not concern. 

If you are lucky enough to be the CEO or founder of a company serving a hard market, the message is simple: You have never faced a more favorable set of conditions in which to grow your business. AI-accelerated coding effectively eliminates the challenges of data integration, feature development, and customization. There is virtually no capability that your customer might demand that you cannot realistically demo to them—and deliver—within a very short time, certainly shorter than their sales cycle.

There is no longer an excuse for a feature not being ready or an integration being too difficult. Those days are over. You can be the harbinger of rapid value delivery in your market. These are the conditions where the industry insight, reputation, and trust you have been quietly building for years emerge as deeply strategic assets. You should not fear the SaaSapocalypse. Your only fear should be not taking full advantage of the massive opportunity being handed to you to move faster than ever.

If your ability to win in a hard market is your real moat, this is your moment. LFG!

Could the future of software be fluid
How do we get the best of AI without losing the soul of software?

The future belongs to young missionary teams
Why it makes more sense betting on youth in the current moment

The AI-native enterprise playbook
Ten real-time observations on a rapidly evolving playing field

No more painting by numbers
It’s the end of the “SaaS playbook.

ENTERPRISE/TECH NEWS

Doomsday Debate. Derek Thompson argues the “AI debate” is really four separate arguments - usefulness, whether models think, bubble dynamics, and moral valence - and we keep conflating them (often along a Bay Area vs. everyone-else cultural fault line). In a conversation with The Atlantic’s Josh Tyrangiel, they lay out three job scenarios (no displacement, slow change, fast change) and talk through what is most likely to happen. A great listen and read.

Layoffs incoming? Jared Sleeper isn’t the first to highlight this, and won’t be the last, but wow do some of these companies have massive headcounts. As he writes: “Yes, UiPath still has more employees than Anthropic. Infer from that what you will.”

HOW TO STARTUP

How OpenAI uses AI. Lenny’s latest episode with Sherwin Wu from OpenAI is a must-listen. The two most interesting sections were: (1) how software engineering has changed internally at OpenAI (almost everybody uses 10-20 Codex threads running parallel) and (2) how if you’re building AI-powered products, recognize that the “scaffolding” (vector stores, agent frameworks, etc.) are likely going to be made obsolete by model advancements.

End of exponential. More Anthropic for you. On Dwarkesh’s podcast this past week, Anthropic CEO Dario Amodei argues that we are rapidly approaching AGI, potentially within the next one to two years. He describes the imminent arrival of a "country of geniuses" contained within data centers, capable of solving complex problems in biology, coding, and engineering at superhuman speeds. Worth a listen, even if you question Dario’s intentions and what game he’s playing.

OpenClaw. Peter Steinberger, creator of OpenClaw, is joining OpenAI, just 2 months after launching OpenClaw to the world. OpenClaw will live on as an open source project (and OpenAI will continue to support it), but it sounds like OpenAI is also making a bet on multi-agent workflows inspired by what Steinberger created.

OpenAI Slack. I found this post strangely compelling: Swyx makes the case that OpenAI should “build Slack” as the natural next step in owning the multiplayer interface for work, and argues Slack has left enough product and pricing resentment (API friction, channel fatigue, weak discoverability of AI features, reliability issues) to create a real opening. By layering an org’s social/work graph onto ChatGPT, they can make agents truly collaborative, and turn chat into the orchestration surface for everything from coding to enterprise workflows. In a world where Anthropic is converging chat + cowork + code into one desktop, “OpenAI Slack” is framed as the bold, hard-to-pull-off move that could re-take the initiative and lock in distribution.

HOW TO VENTURE

Tool-shaped. A viral essay by Will Manidis reframes the current AI moment as a boom in tool-shaped objects: things that feel like work (tokens, agents, dashboards, “Something Big Is Happening” slop) but don’t reliably translate into real-world output. The punchline is that consumption has become the product: people track token budgets and orchestration complexity the way they track capex, even though the tokens → value relationship is often a “cloud,” not a curve. For founders and early-stage investors, it’s a sharp warning: separate measurable business impact from performative activity, and don’t let “the number going up” (usage, spend, agent chains) masquerade as traction.

10x, 10x, 10x. Anthropic raised $30B at a $380B post-money valuation, after 10x-ing revenue over the past 3 years from $100M to $1B to $10B. Generational growth, generational company.

The case for scale. Erik Torenberg, of a16z, makes the case for scaling venture. He reframes the “VC should be boutique” take as a kind of moral myth (like hubris in Greek tragedy) then argues the underlying economics have shifted: outcomes are bigger, companies stay private longer, and founders demand more than capital. The core claim is that modern venture competition is less “picking off the sushi boat” and more “building the best interface to help founders win,” which naturally rewards firms that can compound services, reach, and reserves into a moat. It lands on a barbell prediction: a few scaled platforms + a long tail of true specialists will thrive, while mid-sized “neither-here-nor-there” funds get squeezed.

Firebolt President Hemanth Vedagarbha talks about leveraging AI to cut CAC, reduce churn, and boost ARR at SaaStr AI annual summit.

Reco raises $30M B round for a total of $85M to meet rapidly growing demand for AI SaaS security among enterprises.

DualBird
Data Platform / Automation Engineer (Tel Aviv, Haifa)

FalkorDB
Core Developer (Tel Aviv, Hybrid)

Sourcix
Backend Engineer (Tel Aviv)