AI Empowered Engineering Teams
AI powered pair programming
As engineers, we should leverage AI as much as possible during our working day. AI in our IDEs is powerful especially with chat and MCP servers. I like to work within my IDE as though my AI partner is another person pairing with me, I ask it questions about how I’m doing things, some inspiration with my TDD setup and ask to review as I’m building with context of what I am doing to keep me efficient. I never solely use AI and take what it gives me without question, but instead use it as a tool, a pair partner to get me to the same destination, faster.
AI Pre Review build step in pipeline
A huge step in increasing efficiency in my past experience has been within improving builds process. Such as having linting, tests running, etc. the next step is having an agent review your PR before anyone else sees it. Similar to my above comments on pair programming, this step will reduce time spent and increase quality on every PR due to a non-human calling out issues, nitpicks, etc. it could also go as deep as calling out an anti pattern to the architecture of the solution due to its known context. It’s a supremely powerful step in every PR that can be autonomous once setup.
https://devblogs.microsoft.com/engineering-at-microsoft/enhancing-code-quality-at-scale-with-ai-powered-code-reviews/
AI assisted backlog building
I spoke about this in a previous blog post on empowering Product in my team. There’s some additional steps you can take to further enhance your capability. One that I’m going to look into, is setting up automation yourself with ADO to create an Epic with information and then allow AI to build features and PBIs for you within ADO APIs. I think this is a hugely powerful feature and one that would serve the majority of companies worldwide. CoPilot4DevOps extension already does this and Microsoft are building capability in ADO for it work out the box. One for me to play with and see what I can find. Possibly a blog post on this in the future on further help for product.
AI data generation for testing
This one talks for itself! Faking data can be time consuming when there are constraints and variables. AI can do a great job with all the context. There are scripts that have been built in the past to aid teams, but with AI, I can see a shift away from that to rely on an agent to do it instead.