Notes from the playground.
What we're building, what we're learning, and where AI playgrounds go next.
More from the team
Engineering deep-dives, product notes, and field reports from running agents in the wild.
How the memory graph works
Every conversation indexed, linked, and searchable. A look under the hood at the structure that keeps your agent oriented over months.
Why Froots is model-agnostic by default
Locking yourself into one model ages badly. We let you swap brains per task — here's the architecture that makes it boring to do.
Designing the Skills library
Capabilities as apps you toggle on. How we kept the model honest about what it can and can't do at any given moment.
Routines and the heartbeat loop
Scheduled agent runs sound boring until you let one wake up every morning and tee up your day. Here's how heartbeat picks what matters.
Going local-first with Ollama
Run a real agent fully offline on a laptop. What works, what doesn't, and the trade-offs you actually feel.
Best note-taking app for ADHD in 2026: capture speed, friction, and the AI that finds what you forgot
Most 'best note-taking app' lists fail ADHD users because they optimise for organisation, not capture. This is the honest comparison — Apple Notes, Obsidian, Notion, Reflect, Mem, Logseq, Capacities, and Froots — sorted by the dimensions that actually matter for an ADHD brain: time-to-capture, tolerance for mess, and an AI that can find what you wrote when you can't remember writing it.
Best Notion alternatives in 2026: free, open-source, AI-native, and honest tradeoffs
People are leaving Notion in 2026 over the AI add-on price, the sluggish editor, and the data-portability problem. Here are the nine best Notion alternatives — free, open-source, self-hosted, AI-native — with honest tradeoffs and a migration plan that doesn't pretend Notion's databases are easy to replace.
What is an agent harness? (and why your AI workflow needs one in 2026)
An agent harness is the runtime layer between an LLM and the real world — tools, files, browsers, memory, approvals. It's the missing piece between 'I have an API key' and 'the agent does things autonomously.' Here's what an agent harness actually is, why every serious AI workflow has one, and how to choose between Claude Code, Cursor agents, custom rigs, and Froots.