My overriding thought going into 2026:
Problems you thought were out of reach no longer are.
Pick a problem 100x harder than you thought possible to tackle alone, and run with it.
Start now. AI will catch up and make it possible.
"Most people see empty UI as a bug and immediately reach for a spinner. I see it as feedback."
"If part of the page pops in after navigation, that's the app telling me I forgot to preload something."
Love this framing!
A spinner is an anti-pattern, preloading (sync) is the fix.
More convergences on the same ideas behind Electric Agents
"Session: Durable identity. Append-only event log"
"Make the agent multiplayer by construction"
Two new little features dropped in @tan_stack DB yesterday:
- unionAll() for constructing a live query that is a union of other queries or collections.
- caseWhen() for building queries that have conditions in them.
Both enable a bunch of exciting things 🎉🚀
A database == query engine + page manager + write ahead log.
AI runtimes have a lot to lean from what's been built before.
When you introduce a durable, observable, serverless, write ahead log to your agents, you get distributed coordination.
Electric Agents gives you this!
In the next version of Electric Agents:
You can directly run agents from the CLI or from a TS API
electric agents run <type> [id] [message]
e.g. in CI when a new issue is created — `electric agents run horton "triage this issue"`
It's a lot easier to build a coding harness then people think. 2 days in I'd say we're at ~60% of Claude Code / Codex / Open Code.
Models do most of the work. And @DurableStreams and TanStack DB make quick work of building an amazing app around the coding agent.