Estimated reading time: 5 minutes
Key takeaways:
- Your coordination work has moved downstream. Engineers now plan across the stack alone. Your job is keeping the team coherent and setting AI guardrails.
- Get the right experts on AI-guided proposals and check plans against everything in flight.
- Hold the architectural bar: trading quality for speed leads to subtle bugs and 4am pages.
One engineer with a frontier model can now build across the whole stack, and that leaves tech leads with a different job from the one they signed up for.
The shift to AI augmented development has changed the limits of what a single engineer can achieve. Traditional silos (frontend, backend, database, e.t.c) are merging and the boundaries of a developer’s expertise are blurring.
The advancement in AI does not give engineers free rein to build entire products without asking for verification from domain experts, but it does increase the progress an individual engineer can make. This has effects further downstream: technical leads who previously spent a lot of time on bottlenecks and coordination must shift and find ways to maintain impact.
Your inbox, upgraded.
Receive weekly engineering insights to level up your leadership approach.
Up front coordination is less important
Historically, a large part of a tech lead’s time has been taken up by coordination headwinds and overheads. It’s your job to ensure that the team executes efficiently. This means making sure that the engineers on the project are in regular communication, tasks have been clearly defined and divided amongst the teams, and responsibilities are clear.
Up front you might have to reach out to the team’s database expert, the security consultant, and the backend engineer to get their expertise to contribute to your design decision.
Now a single engineer with a capable AI model and access to the codebase is able to do this with little input: a frontier model with a codebase, historical context (version control logs work nicely here), and any relevant documentation can advise on the best implementation.
This plan should still be formulated into a design document and shared with the relevant stakeholders, but the need to coordinate ahead of time is reduced.
As a tech lead, I’m finding this a novel but exciting experience. It’s been fantastic to see my engineers empowered to propose initial implementations vertically with less input.
Your coordination challenge has moved
Individual engineers being able to move faster and spread themselves across the technical slices of your product does not remove the need for judgment, it moves it. Rather than up front coordination, your job is now to ensure team coherence as everyone moves at a new pace.
I’m finding my role in this era akin to a high level architect. I set the shape of the system and our high level goals, and ensure my team is on the right path. I’m noticing that AI makes it very easy to think “I’ll just quickly add that feature” or “I can extend this change with this quick bug fix” because the effort to write code is reduced. However, the lack of friction means the consequences can be skipped.
AI’s ability to go off-course means that you should be looking for opportunities to define the guardrails and systems that it operates in. Tooling to ensure quality and accuracy of code is more important than ever, particularly as the amount of work and changes that land continue to grow exponentially.
This might mean helping iterate on a series of skills your team can give their agents, or building new command line tools for your AI to access data to increase its accuracy. Look for scenarios where an AI went off-track and you have ideas for how to prevent that in the future.
More like this
Review is even more important
This enabling of engineers is fantastic but as a technical lead it means your first and primary involvement is at the design stage and that means that review is a critical step.
As the person with visibility across the business and the relationships cross functionally, you need to make sure that:
- The right stakeholders are involved. If your frontend engineer has proposed backend changes with AI’s guidance, the right person must review that proposal. We’ve all seen AI confidently propose something that the right domain expert can refute.
- The implementation plan is the right one. In isolation the engineer (and/or AI) can produce a plan but only you have visibility into all the projects that are ongoing and can spot areas of potential mismatch or concern.
- AI isn’t an excuse to not involve humans. If the project requires heavy UX it’s OK if the design doc has AI-generated mock-ups, but it’s not OK to proceed without engaging a UX designer to produce the actual mocks that will be implemented.
Failure to adapt here will have serious consequences for your teams, your code, and your users.
If you and your team don’t maintain a high architectural bar in favor of moving quickly with AI, you will pay for this down the line. AI does make writing the code easier, but it will happily make questionable decisions in order to get tests passing and features out the door. These will bite you in the form of subtle bugs and overly complex code. It’s easy to not worry about this when things are working, but you’ll regret it when you’re paged at 4am because your system is down and you’ve lost your understanding of it.
Without a high bar to the features that are shipping the quality of your product will erode. AI does give us the ability to do more – I have shipped many features this year that previously would remain low down the priority list – but AI is not a designer and it is not an excuse to ship the feature and figure out the flow later.
Leading on AI augmented workflows
A good tech lead in this era also uses their overview as an opportunity to think critically about workflows. You have the advantage of seeing how a group of engineers are adapting to and using AI, and you can spot the places where that works well and poorly.
Done in the right way, this can encourage your team to adapt and experiment. A project that goes badly is frustrating but it is a great learning experience for the team and you should be looking for how to learn quickly and level up your team’s AI skills.
Berlin • November 9 & 10, 2026
LeadDev Berlin is two weeks away. Find what helps stretched teams keep going.
The most successful teams will find the ways to use AI to rapidly increase their velocity without paying the costs of more confusion, more bugs, and more overhead of code reviews, deployments, and debugging. This is where you can set the stage and have a huge impact.
Default ownership is shifting and tech leads must shift with it
Five years ago a mid-level engineer in your team could own their area of expertise but would need help in each slice that makes an end-to-end feature. A tech lead’s role revolved around this coordination which naturally constrained a team’s velocity.
With this constraint reduced and an engineer’s scope changing, tech leads must adapt too to ensure their teams remain productive.
