👋 Hey, Egemen here.
Something I heard from a founder is exactly in Carta's data this week: seed teams are down to four people, but grants for AI engineers at the smallest startups climbed 64% in two years.
Here’s a snapshot of what’s on the menu today:
💡 Spotlight: Stop letting busywork get in the way of selling
🧠 Deep-Dive: The Four-Person Seed Round
🗺️ Method: Test the workflow first
⚾️ Catch: Hacker News is back
☝️ Scaled This Past Week: Rillet
💡 Spotlight
Stop letting busywork get in the way of selling
Researching accounts. Building lists. Writing sequences.
There's a better use of your team's time.
Apollo is the AI revenue engine that handles the busywork, so you can stay focused on selling.
Plus, everything you need is in one place:
230M+ verified contacts
AI-powered outreach
Data enrichment
Inbound lead capture
Meeting scheduler
And more
Stop doing busywork and start building pipeline, faster.
With Apollo — the AI revenue engine powering 4M+ users.
Presented by
🧠 Deep-Dive: The Four-Person Seed Round
Crunchbase published its weekly roundup on Friday. The ten biggest US rounds for 15 to 21 August ran from $800M down to $75M, and most of the list went to AI companies. Capital is not the thing in short supply.
Building got cheaper and teams got smaller. Where founders keep guessing wrong is what happened to the money that came free.
It did not become runway. Between January 2024 and February 2026, the median initial equity grant for AI/ML engineers rose 31%, against roughly 11% across all individual contributors. At startups valued between $1M and $10M, grants for that same role climbed 64%.
Read that 64% as a seed founder. The smallest companies on Carta are paying the steepest premium for the exact hire that makes a four-person team work.
You skip three engineers, hand the difference to one person who can run agents at scale, and the market has already priced that person for you.
👉 Seed teams are down to four people while AI grants at the smallest startups climbed 64%. Count your automatable hours before you open another role, because the grant you sign this quarter is the one you keep.
🗺️ Method: Test the workflow first
Taku is a desktop workspace for AI tools you did not build. You browse a marketplace of agents, apps, skills and workflows that other people already built and tested, you click run, and it runs. No clone, no API keys, no config file.
Pick a painful task you have: Find the job on your team that eats the most hours and needs the least judgment. Weekly reporting and inbound triage are the usual winners.
Search, do not build: Look for the closest thing somebody has already published to the marketplace.
Run it on real data: Not a sample file. Your actual inputs, once.
Fork the winner: Point it at your Notion and your Drive, rewrite the prompt in your voice, save it as yours.
Chain a few forks together and Taku calls it a Stack.
The company reports 12,000 apps and skills on the marketplace and 1.2M runs a month. Free to start, Mac for now.
⚾️ Catch: Hacker News is back!
☝️ Scaled This Past Week: Rillet
Rillet raised $100M in Series C funding led by Iconiq Capital, at a $1B valuation - it’s the scale of the week!
Three rounds in a year tells you investors think the replacement cycle is happening now instead of over a decade. That matters beyond finance software.
Every category with an incumbent built for manual workflows is a category where a small team with agents can undercut on both price and headcount, which is the same argument the deep-dive makes from the hiring side.