In partnership with
Your prompts are leaving out 80% of what you're thinking.
When you type a prompt, you summarize. When you speak one, you explain. Wispr Flow captures your full reasoning — constraints, edge cases, examples, tone — and turns it into clean, structured text you paste into ChatGPT, Claude, or any AI tool. The difference shows up immediately. More context in, fewer follow-ups out.
89% of messages sent with zero edits. Used by teams at OpenAI, Vercel, and Clay. Try Wispr Flow free — works on Mac, Windows, and iPhone.
Quick gut check before you read another "AI is transforming SaaS" take this week: what if it isn't transforming software at all — what if it's replacing it?
That's not my line. Crunchbase made the argument out loud this week: SaaS isn't coming back. I wanted to disagree. Then I lined it up against where capital actually moved in the last seven days — and it's hard to.
The tell is in the term sheets
Look at the rounds that closed this week. Almost none of them are seat-based software:
Skild AI — $1.4B at $14B+, for an "omni-bodied" robot foundation model.
Saronic — $1.75B at $9.25B, for autonomous maritime systems.
Mecka AI — $60M, to train robots on human body-sensor and iPhone data.
Groq — $650M, rebuilding leadership after Nvidia's $20B "not-acqui-hire."
The one classic-software exception — AppsFlyer's $1B+ raise at $2.7B — got bought into by Google, Meta, Unity and Moloco precisely because it sits underneath everyone's AI ad spend. Even the SaaS winner of the week is really an AI-infrastructure play.
The week's biggest rounds. The lone adtech-SaaS raise (gray) is dwarfed by robotics and embodied AI. Source: Value Add Pulse.
The seat was the unit of SaaS. The outcome is the unit of agentic AI. Nobody buys 50 seats of an agent — they buy the work.
Why this is structural, not a hype cycle
The SaaS model priced access: pay per seat, per month, forever, whether or not anyone logs in. Agentic AI prices work done — and that quietly breaks the comp table founders and LPs have leaned on since 2015.
For founders: "build a tool people open" is becoming "build an agent that does the job nobody has to open." Your wedge isn't a better UI anymore — it's a better outcome per dollar.
For GPs: your SaaS multiple benchmarks are measuring a category being repriced in real time. An 8x revenue multiple on a seat-based business and an 8x on an outcome-based agent are not the same asset.
For LPs: the durable revenue you underwrote on net-retention math assumes humans keep renewing seats. Agents don't renew — they get swapped for the next, better agent. Retention curves built for SaaS may not survive contact with this.
What I'm watching
Whether any pure-seat SaaS company posts accelerating net-new logos in H2 2026 — or whether growth quietly migrates to usage and outcome pricing.
How fast this week's ~$3.5B in robotics and agent-foundation rounds turns into revenue, not just valuation.
The OpenAI and Anthropic S-1s — the first time the public market gets to price "AI that does the work" at scale.
The uncomfortable version: if your 2026 thesis still rhymes with "vertical SaaS plus a copilot," you may be underwriting the losing side of this trade.
I keep a live tracker of AI company valuations against the old SaaS multiples — it's the clearest picture I've found of the repricing as it's happening. (And if you want the other side, the SaaS multiples tracker shows just how far the correction has already run.)
— Trace
Reply and tell me which side you're on — "SaaS is fine" or "it's already over." I read every one.
