Atoms Food
CloudKitchens, Otter, Picnic, Pro Food Properties, Lab 37
Strategic intelligence brief / March 2026
I spent 24 hours researching your portfolio from primary sources. Here's what I'd tell the board.
Five deals worth pursuing right now, six questions only the parent can answer, all built from primary sources in 24 hours. This is the same system I'd run inside your S&P function.
01
The Thesis
This page is the work sample, not the wrapper around it.
Travis Kalanick described his ideal hire: simulate what it's like working together so that day one already feels like week two. This brief is that simulation. Not the research itself, but the system that produced it.
Within 24 hours of the Atoms launch, I deployed six parallel research agents, each investigating a portfolio company from primary sources. Company websites scraped, product pages read, claims cross-checked against filings. Then a second wave: fact-verification agents that caught errors in every single initial research file. Then synthesis: cross-company patterns that no individual subsidiary employee could see.
The question this site poses: what would this system produce with access to internal data, customer conversations, and the live portfolio roadmap?
I'm applying for Strategy & Planning or Corporate & Business Development. I build the systems that produce this kind of output. Hire me and this runs on internal data.
02
The Portfolio Map
Six companies, three divisions, one physical-AI thesis.
Parent view
Atoms
Food generates the revenue. Mining is the furthest along in robotics. Transport is the long-range bet.
Atoms Mining
Pronto
Atoms Transport
Thesis stage / no public product yet
03
Company Deep Dives
Each company has a real business. The question is where the parent should press hardest next.
04
What Only the Parent Can See
The highest-value questions do not belong to any one subsidiary.
05
Five Deals I'd Pursue
Concrete names, concrete timing, concrete reasons to move now.
06
How This Was Built
The job is not just being right. It is being right quickly enough to matter.
This brief wasn't produced by one person reading articles. It was produced by a system: five specialized AI agents, each with a defined role, running in parallel, with me deciding what to keep, what to kill, and what to investigate further.
The first wave, six parallel research agents one per company, built initial briefs from press, market data, and industry coverage. The second wave scraped every company's actual website and flagged every discrepancy. Every single company had errors in the initial research. The fact-check sprint caught them all. Then synthesis: a cross-domain agent that held all six companies simultaneously to find the connections specialists in one lane would miss.
This is the system I'd install at Atoms. Point it at your next board question and have the answer before lunch. Not a one-off report, but a permanent function.
The Agent System
Working files behind this page
07
Let's Talk
This brief is the application for Strategy & Planning or Corporate & Business Development.
Dawson Smith
This was built from public sources in 24 hours. Internal data makes it sharper. Customer conversations make it dangerous.