From Papercuts to Autopilot: The Future of Feedback

If you haven’t read Papercuts: The Little Fixes That Made a Big Difference at GitHub, start there. It’s the story of how fixing tiny annoyances—those little things everyone grumbles about—can build trust, momentum, and joy.

But what comes next?

I can see a future where AI doesn’t just help collect and sort feedback—it becomes your co-pilot and mechanic, finding friction before your users even notice it. Imagine an AI quietly watching how developers use your product, sensing when they stumble, and opening a pull request to fix it while you’re still making coffee.

We’re already seeing glimmers of this. The slow, painstaking work of sifting through spreadsheets, community posts, and endless support threads is starting to fade. Soon, AI will help us not only organize and summarize feedback, but draft better documentation when confusion spikes, suggest onboarding improvements when drop-offs appear, and even tweak user interfaces that consistently trip people up.

How to Use AI Tools Now to Stay Ahead of Feedback

That future isn’t fully here yet—but there are ways to make AI your secret weapon today:

• NotebookLM: Toss in all your customer feedback—support tickets, forum posts, survey data—and let it help separate signal from noise, surfacing what really matters.

• Audio Overview: Don’t have time to read long reports? Summarize feedback into an audio briefing you can listen to on your Monday commute, so you’re prepped for your weekly planning meeting.

• Gemini for Docs & Summarization: Feed interview transcripts or customer reviews into Gemini and get distilled takeaways and suggestions on what to prioritize.

• ChatGPT or Claude: Treat them as brainstorming partners for turning vague complaints into concrete ideas, or drafting user-friendly release notes.

• Slackbot Integrations: Set up bots to monitor your community channels and bubble up repeated questions or complaints so they don’t slip through the cracks.

Why We Still Need People in the Loop

The best products of the future will anticipate needs and smooth out friction almost invisibly. But we’re not quite at autopilot yet—and honestly, I’m not sure we ever should be.

At GitHub, we once had users asking us to remove the contribution graph—the grid of green squares showing your daily commits. Some developers felt pressured to keep those squares filled, day after day, and it was leading to burnout. If we had taken that feedback at face value, the easy move would’ve been to remove it. But the contribution graph wasn’t just a dashboard—it was part of GitHub’s personality. It was playful and unique.

Instead of scrapping it, we leaned in. Developers started turning those squares into pixel art: snake games, secret messages, even animations like the game of life. It became a canvas for creativity, not just a metric to chase. We didn’t just listen—we interpreted.

And that’s something AI can’t do yet. Because sometimes the loudest feedback points to a real problem, but the solution users ask for isn’t the one that actually solves it. Great product work comes from asking why, reading between the lines, and finding solutions that not only fix the pain, but make the product more joyful and human.

So yes, AI will make it easier to listen, spot patterns, and even suggest fixes. But someone still needs to steer. Someone who can turn tension into delight, and complaints into creativity.

Papercuts may become autopilot one day. But the best product teams? They’ll always have their hands on the wheel.