AI and the Work Around Coding
AI is often discussed as if it's going to replace developers, or at least write most of their code. But if you ask engineers where AI is actually helping today, the story looks very different.
The gains most developers are seeing are in the "surrounding work." The messy, unglamorous, and time-consuming aspects of building software have always impacted team productivity.
A recent Atlassian survey backs this up. Among more than 3,000 developers, the top reported productivity boost from AI wasn't coding. It was everything around it: searching for information, testing, documentation, automating workflows, and refining ideas.
If you're tired of AI hype, that's worth paying attention to. Because it means we're not talking about replacement—we're talking about relief. Relief from the drag of everyday friction that gets in the way of actually solving problems.
Here are three areas where AI is already making a meaningful impact.
1. Reduce Context-Hunting
Developers spend a surprising amount of time searching for answers. Atlassian's survey showed "finding information" is the single biggest productivity drain. Think about the last time you had to track down why a decision was made—you probably checked Slack, Confluence, Jira, and still weren't sure if you had the whole story. That's hours gone before you write a line of code.
AI changes the equation by making context easier to surface. Instead of digging through scattered tools, you can:
- Ask an AI trained on your repos to explain a past pull request.
- Summarize a hundred-message Slack thread into three key points.
- Create a brief outline of related Jira tickets to identify dependencies.
AI doesn't eliminate the need for human judgment, but it clears away the clutter. Developers can focus on solving problems instead of piecing together fragmented information.
2. Shorten Feedback Loops
Every developer knows the pain of waiting for feedback—on a pull request, a test run, or a question for another teammate. The longer that loop, the more flow you lose.
AI can compress those loops. Linting tools augmented with AI can flag potential bugs while you're writing, rather than hours later in CI. Test generation can catch edge cases you might have missed, saving you from a round of QA ping-pong. Even something as simple as drafting a first pass at documentation gives reviewers something concrete to react to, speeding up the cycle.
The effect isn't that AI replaces reviewers, testers, or teammates. It's that it keeps work moving instead of stalling. A faster loop means fewer context switches, less rework, and more energy spent on forward progress. This is the same principle I unpacked in When Is Your Work Actually Complete: the sooner you surface issues, the less painful they are to fix. Shorter cycles consistently drive better outcomes.
|
|
Wondering If a Startup Is Right for You?
Big Creek Growth Company shares what it’s really like to work in a startup and what founders are looking for when hiring.
|
|
3. Ease the Weight of "Glue Work"
Software projects require a lot of invisible effort: writing handoffs, keeping documentation current, updating tickets, and nudging teammates on dependencies. This "glue work" often falls on the most conscientious people, and it eats into the time they could spend on higher-value contributions.
AI is showing promise here, too. A meeting transcript can be distilled into a shareable summary with action items. Out-of-date docs can be refreshed with a first draft that captures recent changes. Workflow automations can close stale tickets or remind the right person to update the status.
None of these tasks disappear entirely. You still need human oversight, but the burden lightens. The result is more balanced workloads and less frustration for the people who usually end up carrying this hidden labor. In Are You Creating Lift or Drag, I described how invisible work can quietly slow a team down. AI won't remove that work altogether, but it can keep it from becoming drag.
AI doesnt change the fundamentals of software development. Teams still succeed or fail based on how well they can deliver, review, and collaborate. What's shifting is the cost of the friction around those activities.
Searching for context, waiting on feedback, and carrying the weight of glue work have always slowed teams down. AI doesn't replace developers in those areas, but it does make the work lighter. That's why the most impactful uses so far aren't about writing code—they're about removing the drag that surrounds it.
If you're experimenting with AI, start there. Identify where your team loses time and energy outside of coding, and assess whether an assistant can assume some of this overhead. Even small gains in these areas compound, and the payoff is more focus on the work that actually moves your product forward.