The whole is greater than the sum of its the parts. Code automation is the #1 use case for Agentic AI 🏄‍♂️ 🌊 Well done @AnthropicAI marketing team 👏

10 years into Enterprise Software and for the first time I’m hearing: building is easy, distribution is hard. Software in the AI era feels a lot like Consumer Internet — code, intelligence, workflows all commoditizing fast. The best AI companies will obsess over distribution and GTM creativity the way the best Consumer companies obsess over growth.

If you’re a Founder or a CEO looking to get your entire team AI-pilled, @AnthropicAI is giving $1,000 in usage credits to every new user (up to $10M per organization) who activates Claude Code or Claude Cowork for the first time. support.claude.com/en/articl…

B2B Marketing in 2026 — doing things that don’t scale and doing it over and over again! 🫡 @MonacoGTM

In the last 18 months too much capital chased a tiny pool of companies — the enabling layer for AI: models, chips, neo clouds. The building blocks are in place and compute capacity is actively getting built. In the coming 12-18 months, we'll see the attention and capital rotate to Apps re-imagining workflows end-to-end with AI. The tell: faster horses, or the car? Automating the human workflow is a faster horse. Re-imagining the problem entirely with AI is the car. Never been a better time to be an AI Apps Investor! 🤩

SaaS stocks are singing Fly Me to the Moon 🚀 🎶 💃 $TWLO $HUBS $DDOG $SNOW

Some companies keep trying.. like they should!

For two decades, developer infrastructure companies optimized for the first five minutes of developer adoption. Then AI coding agents showed up and the bottleneck shifted from getting developers to install your SDK to getting it implemented correctly across the entire organization. That's where Agent Skills come in. These are small, installable context packages that teach AI coding agents what your company’s best solutions engineer knows about how your product works. One command, and every interaction the agent has with a codebase carries deep, opinionated knowledge of your SDK. Great SDK documentation has always been a form of distribution. Agent Skills are documentation that executes. The companies building them now are pre-loading their adoption curve into every AI-assisted developer workflow. Read more on Agent Skills from Battery's @webaficionado and @BarakSchoster: okt.to/CD5Ftb

The clearest signal a SaaS company has become an AI company? Revenue is accelerating on net new adds. Only 5⃣ companies pass that test: Gong*, Hightouch, Intercom, Notion, and Ramp. Kudos @ivanhzhao and @akothari 👏 *denotes a Battery portfolio company

AI washing at its best!

ClickHouse might deliver on the “App Store” vision Snowflake / Cloud Data Warehouse never quite pulled off. Who’s building next-gen AI Apps on top of @ClickHouseDB? 🤔

Two years ago, companies were paying $200K for "Prompt Engineers." Now every knowledge worker prompts AI daily — ChatGPT, Claude, whatever. Great technology democratizes itself. It's only a matter of time before anyone can build end-to-end agentic apps.

So much of the world’s AI innovation is happening within a 4 sq mile radius of San Francisco… and yet, you ask someone living in SF vs. Palo Alto vs. South Bay about AGI, whether SaaS is dead, or if Claude eats AI Apps, you’ll get completely different answers. What a time to be alive! 😇

Who’s next on the PE SaaS wipeout list? UserTesting, Pipedrive, Salesloft, iCIMS would be my top picks.

🎯💯

"Cursor for X" may just be Claude. Its core primitives are universal enough that, with the right skills, plugins, and MCP connectors, you can replicate practically any junior knowledge worker inside Claude.

Agent skills fix the gap. Wrote about it with @webaficionado @BatteryVentures : battery.com/blog/agent-skill…

Most dev tools fail to close the gap between “SDK installed” and “SDK instrumented well.” As the puck shifts from Humans to AI Coding tools, the question becomes: can you teach the agent what your best SE knows? Say hello to Agent Skills! @BarakSchoster battery.com/blog/agent-skill…

Your margin is my opportunity. Ola and Uber were a duopoly in India for nearly a decade. Collective they deployed over $5B of venture capital into India to build trust, educate the market, and create demand-supply liquidity. You can hail an Uber in minutes in NYC and just as quickly in Delhi, Bangalore, or Mumbai. It was a knife fight between these two players and India was the last market Uber refused to cede to a local player. Then comes Rapido and says zero commissions and flat subscription fees for Drivers. They started in one city (Bangalore) and just Bike taxis and gradually expanded to every major city across all 3 modes - Bikes, Cars and Auto Rickshaws. With just $500M in total funding (1/5th of Ola's), they now have 74 million MAUs, more than Uber and Ola combined. If any company in this category goes public in India, it's likely Rapido. Now they're going after food delivery market with the same zero-commission model, aiming to disrupt the Swiggy-Zomato duopoly. If they can pull this off in a low-trust market like India, you have to wonder: is there an opening for consumer AI companies in the US to run a similar play against Uber and DoorDash? Fascinating case study🙂 @peegeekay @TheKenWeb the-ken.com/newsletters/two-…

At this point, it almost feels like OpenAI’s product roadmap is Anthropic’s changelog 😏