DevOps / SRE / Platform · 06.08.2026, 17:23 UTC
Why Todoist says less AI can deliver more
| Schweregrad | info |
|---|---|
| Kategorie | DevOps / SRE / Platform |
| Quelle | The New Stack ↗ |
| Veröffentlicht | 06.08.2026 UTC |
Sicherheitsmeldung mit Schweregrad noch nicht bewertet. Technische Details im Tab „Originaltext“; empfohlene Schritte in der Checkliste.
Doist CTO Gonçalo Silva doesn’t pretend to know where AI is headed. The pace of change has made it harder to commit to one roadmap, model provider, or way of working.
“I can make predictions, but I feel like I can make 10 of those and they are all very different from each other, and they almost all have a similar probability of being right,” Silva tells The New Stack. “So I’m operating in a space with more unknowns than ever.”
Rather than trying to predict which direction will win, Doist — the company behind products like Todoist and Twist — established three broad principles more than a year ago, Silva says: The first is that AI features should be purposeful; the second is that they should be private and secure, and the third is that they should be built to improve as models change.
Doist has learned that just because a feature is now easy to build with AI, that doesn’t mean it’s worth shipping.
Through its early experiments, the Doist team also learned that just because a feature is now easy to build doesn’t mean it’s worth shipping.
Doist is currently testing Automations, a new service that is meant to take Doist from a company that focuses on helping users remember tasks to one that helps them execute them as well.
Automations uses AI to turn a user’s natural-language request into a manifest. Then, when a trigger fires or a scheduled job begins, ordinary code — not a large language model — executes those steps. Doist plans to launch Automations out of its closed beta in August or early September, Silva says.
“When you create an automation that you expect to run …