Eighteen months ago, if someone told me AI would change how I run sprint planning, decompose Jira tickets, and review pull requests, I would have been sceptical. I had been watching LLMs evolve closely but engineering leadership felt like the last place automation would actually bite.
I was wrong. The shift did not come from replacing engineers. It came from augmenting the connective tissue of the team. The planning, the communication, the context-switching that eats leadership time alive.
The problem I was trying to solve
As Tech Lead at TEG.tech, I oversee the mobile ecosystem for a logistics platform that serves thousands of freight businesses across the UK and Europe. The team is cross-functional. Mobile engineers, backend developers, QA, and product, all spread across multiple time zones.
My days were consumed by three things: ticket decomposition (turning vague requirements into precise, codebase-grounded tasks), coordinating dependencies between mobile and backend, and reviewing security remediation work from pen test reports.
Each of these required deep context about our Flutter codebase, our Jira project structure, and the domain itself. I was the bottleneck.
Enter the AI-augmented workflow
The turning point was completing the Claude Code certification and the agentic programming tools training. These were not theoretical. They were immediately practical. I started using Claude as a thinking partner for ticket decomposition. I would feed it pen test findings, point it at specific files and line numbers in our Flutter repo, and have it draft Jira descriptions with codebase-grounded detail.
The quality of our ticket descriptions went from “fix the auth bug” to precise descriptions referencing exact file paths, line numbers, suggested code changes, and team ownership boundaries.
What I did not automate
AI handles the preparation layer. It gathers context, drafts structured content, and surfaces information. I still make every decision about scope, priority, and team assignments. I still sit with engineers and walk through architecture decisions. I still write the final version of any communication that goes to stakeholders.
The distinction is between augmentation and automation. I am not replacing my judgment. I am freeing up the time I need to exercise it.
Advice for other engineering leads
Start with the boring stuff. Do not try to use AI for your most creative or sensitive leadership work first. Start with ticket writing, meeting summaries, dependency mapping, and codebase documentation. These are high-volume, context-heavy tasks where AI adds immediate value without risk.
And most importantly, tell your team. I was transparent with the Pirates (our mobile squad) about how I was using AI in my leadership workflow. It normalised the practice, and several engineers started adopting similar approaches for their own work.
The best engineering leaders I know do not compete with their tools. They orchestrate them.
AI did not make me a better leader. It gave me back the hours I needed to actually lead.