Article

Taking Care of To-Dos with AI

Dallas Inman

•

March 16, 2026

•

2 min

Summarize this article with:

I hate to-do lists. So I made AI maintain mine.

I'm a Forward Deployed Engineer at Airbyte. My days are customer calls, sprint planning, business reviews, and support threads, each one producing small commitments that are supposed to live somewhere.

I've tried every task app. I always abandon them. The maintenance burden never felt worth it because manually copying action items from a transcript into a task board is just work about work.

So I stopped. Instead, I built a system that does it for me.

A few weeks in, I ran my morning briefing command and opened my task app to a full board of tasks I hadn't entered. Everything from the day before, with due dates, priorities, and links back to the source. My reaction to each one was just "oh right, I remember that" which is exactly how it's supposed to feel.

Here's how it works. My AI assistant writes tasks to a shared file and a background process syncs that file with my task app every five minutes. Check something off on my phone and the AI sees it. AI extracts an action item from a call transcript and it shows up in my app within minutes. When I process a meeting transcript, the system pulls out my specific commitments even when a call covers five different workstreams. One sprint planning call turned into three discrete tasks that would have completely blurred together in my head otherwise.

Every morning, a single command generates a briefing with yesterday's meetings, today's priorities, and which customer threads are going stale. On Mondays it looks back to Friday.

The thing I find most useful is probably the most boring to explain. Every processed customer call automatically appends a handoff note to a running coverage document. I'm taking two weeks off next month, and that document has been writing itself since February.

The todo list still isn't something I enjoy. What changed is that I'm not the one keeping it alive. My task app is just the view layer now. The shift is from "I need to remember to do X" to "the system already knows about X," and that difference is what actually makes it stick.

At Airbyte, we build infrastructure that connects data sources and makes them useful to the systems that need them. Turns out I needed the same thing for my own workflow. When transcripts, task managers, and customer context are all connected, nothing falls through the gaps.

This is the first post in Deployed with Dallas, a series on how I've been building AI into my day-to-day as an forward deployed engineer at Airbyte. Stay tuned for the next installation!

Limitless data movement with free Alpha and Beta connectors

Introducing: our Free Connector Program ->