Study Session: How Long Will This Take? Estimating in the AI Era

Are you a software engineer who enjoys digging into new ideas, practices, and patterns that strengthen your craft? Join us for the next OKTech Study Session, a gathering for engineers who don’t just code, but want to understand the why and how behind solid software design.

Session Summary: Estimation has always been one of the hardest parts of software engineering. AI was supposed to make it easier. Instead, it made it harder. A task that took eight hours last year might take twenty minutes with AI, or thirteen hours of debugging and cleaning up confidently wrong code.

Recent research shows that developers using AI tools believed they were 20% faster while actually measuring 19% slower. The gap between perception and reality is now wider than most estimates themselves and this influences how we estimate our work.

In this session, we will run a small experiment together. You will estimate the same piece of work three ways: how long it would take you with AI, how long it would take you without AI, and what an AI itself predicts. We will compare results across the room, and try to figure out where the gaps come from and what that means for how we plan our work.

The session is not about whether AI is good or bad for software development or how to use it. It is about whether we can still trust the estimates we give to stakeholders, to our teams, and to ourselves.

Goal: Leave with a clearer understanding of how AI changes your way of estimating.

How do we work?

• Small group to encourage discussion and participation
• Concept- and practice-focused, not tool-specific
• Interactive discussions and hands-on exercises

Who is it for?

• Software engineers using AI in their daily work, or expecting to soon
• Tech leads responsible for sprint planning or forecasting
• Anyone who has ever had to tell a stakeholder how long something will take

Session difficulty: Low (everyone with some tech experience can join)