Background agents and the work humans still need to own
Why useful background agents need triggers, queues, reviewable outputs, and clear human ownership more than they need vague autonomy.
Journal topic
ai-agents
Notes tagged with ai-agents. The archive is being shaped around AI systems, data products, market intelligence and founder-led product lessons.
Agent Skills are AI workflows
Why agents need process, training material, and verification steps before their work becomes predictable enough to trust.
Growth analytics should create decisions, not dashboards
How I think about connecting Search Console, PostHog, and Hermes into a weekly review loop for acquisition, product behavior, and follow-up work.
The underrated part of AI agents is cron
Why scheduled AI workflows matter when the goal is steady, reviewable work instead of impressive one-off demos.
Building MCP: when your data becomes part of the agent workflow
A practical note on why MCP matters when agents need real data, bounded tools, auditability, and human judgment instead of another pasted prompt.
How I use Obsidian as operating memory
A practical look at how my Obsidian vault, cron jobs, GitHub issues, and Hermes workflows keep context moving toward decisions and reviewable work.
Hermes, my agent for everything
How I am using Hermes as a practical execution surface across notes, repos, GitHub Projects, cron workers, QA, and publishing without handing over judgment.