01General note · Thesis
Software used to need a company to build it.
I built Pulse to find out how much one person working at every layer can build with agents. So far: thirteen products on one core.
1builder
308MCP tools
7layers on one ontology
0services headcount
Scroll through the eight notes
ViewDrag to orbit · shift-drag to pan · pinch to zoom
Sheet I-001 · Investing in Pulse
General notes
Applies to the whole set. Read before any other sheet.
01 · ThesisSoftware used to need a company to build it.I built Pulse to find out how much one person working at every layer can build with agents. So far: thirteen products on one core.
02 · The builderI work at every layer, and the record is public.
03 · ProblemCompanies rent a dozen tools and hire people to glue them together.At the top end, Palantir and C3.ai solve integration with forward-deployed engineers. It works, and it is why they are services-heavy and gated behind seven-figure contracts.
04 · ProductOne graph under every product, and agents that operate it.Every product, from Docs and Grids to Forge and Script, shares one ontology. Agents do the forward-deployed engineer’s job on it: onboarding data, building workflows and running them, with approvals, canaries and evals.
05 · ModelProduct-led, with no services arm.The free tier runs on on-device and light models, which is what lets one person afford a free tier at all.
06 · PositionPlatform depth without the services bill.
07 · Why nowModels can operate software now, and MCP gave them a standard way in.That changes who can build a company-sized platform. Pulse is my evidence: the surface area of a much larger company, built and run by one person working with agents.
08 · Where it standsI’m talking with a few investors about Pulse.Capital will go to compute, distribution and the first packs. The team stays small on purpose.
What I show: a live walkthrough over MCP, and the One Organism (42 slides), Agent Storms (38) and Solar Seeds (80) decks.