🕶️ Town hit $1B in four months

Good morning, founders 🕶️

For two years, the AI pitch was capability: look what the agent can do. This week the money moved to a different question: what is the agent allowed to do? Town became a unicorn selling assistants that ask permission and write drafts instead of hitting send. Microsoft, which has invested in two frontier labs, now tells enterprises those models "cannot be trusted." DHS is pulling DeepSeek out of federal systems. And one consent sentence in a prompt decides whether a coding agent stays in its lane. Trust used to be a compliance checkbox. Now it's the product. So who's writing the permissions for yours?

In today's newsletter, we'll get into:

What does it look like when a consumer brand starts to break out?

Popsmith generated $5.8M in revenue in 2025 and is projecting $9M in 2026, with more than 100,000 customers and $20M+ in lifetime sales.

The brand has already earned national recognition through Oprah’s Favorite Things and partnerships with Costco, Sur La Table, and Williams Sonoma, where Popsmith is rolling out nationwide this holiday season.

Now Popsmith is raising capital to accelerate its next stage of growth, and individual investors have the opportunity to become shareholders.

Explore the Popsmith investment opportunity.

June 2026: Town launches publicly with a $55M Series A. A few weeks later: seven-figure ARR. Since July: paying customers up 4x, paying teams up 9x. October 19: a $108M Series B at a $1B valuation, co-led by Forerunner and Menlo Ventures, with a16z and First Round back for more. That's at least $160M raised by a company most of its customers met this summer.

Partly the product. Menlo says its team "roughly doubled" in five months, and the new hires "work for less than $100/month, don't need desks, and work 24/7." Those are Townies: the AI assistant Town gives every employee, wired into email, calendar, docs and 60+ tools. When your lead investor's due diligence was its own inbox, the term sheet writes itself.

Two choices, both boring on purpose. First, a narrow wedge: email and calendar, where knowledge workers burn 13+ hours a week. Second, permission-first defaults. Townies write drafts instead of sending, ask permission when needed, and share with each other only what their person would share. Nothing a customer connects trains a model. Founder Jean-Denis Greze, Plaid's former CTO, built something a security team could say yes to, then let it learn the job.

That's the founder takeaway: in agents, the default setting is the go-to-market. Capability wins the demo. "It won't hit send without you" wins every inbox in the company.

Now the stakes go up. Townies talk to each other and to customers outside the company. The first Townie-to-Townie mixup that reaches a client will show whether trust-by-default scales, or just demos well.

In this video, we talk about 3 pitch decks that have amassed over 20M views and are basically THE reference thousands of founders have been using for years.

And these are great decks, but you should NOT copy them unless you understand why they work, and the context when they were used.
We tear them down, study their structure, and show you what we extracted from them to build a framework to tell company stories.

Picture a landlord who owns stakes in two moving companies, then hands every tenant a pamphlet titled Never Leave Movers Alone in Your House. That's Microsoft this week.

Launching Guardian Agents, a tool that monitors a company's AI agents, security chief Hayete Gallot wrote that "frontier models, regardless of their origin, cannot be trusted." Not because they're malicious, she argued, but because they're unpredictable: trained to finish tasks, not to follow "organizational norms or judgment." Satya Nadella put it more cleanly on his blog: "separate the supply of intelligence from the authority over it."

It has invested in both OpenAI and Anthropic, sells their models, counts OpenAI as one of its biggest cloud customers, and runs its own lab. Here's the scorecard. The labs want to sell you agents and the safety story. Microsoft wants to sell you the agents, then sell you the guard watching them. Self-serving? Completely. Wrong? No. Auditors shouldn't work for the company they audit.

The products are already live. Defender now detects up to 60 types of employee-installed agents, including Claude Code, Codex and OpenClaw. New controls scope each agent's access to the task and require human approval when an agent appears to be making risky decisions.

Translation: if you sell agents to enterprises, your product is about to show up on a dashboard labeled "detected." Ship scoped permissions, approval gates and audit logs before the procurement questionnaire asks for them. And if you build oversight tooling, Microsoft just wrote your "why now" slide by arguing that buyers need someone independent. That someone doesn't have to be in Redmond.

Watch the next renewal cycle: "Does it work with Guardian Agents?" is about to become a checkbox.

Scale your growth: Join names like HubSpot, Ahrefs, and Deel. Put your product in front of 85,000+ startup founders. Let's talk.