How to Use AI to Improve Teamwork in Engineering Teams
Great teams build great software, not individuals. This is how we can improve teamwork using AI!
This week’s newsletter is sponsored by DX.
If you’re looking to measure the impact of AI tools → the DX AI Measurement Framework includes AI-specific metrics to enable organizations to track AI adoption, measure impact, and make smarter investments.
When combined with the DX Core 4, which measures overall engineering productivity, this framework provides leaders with deep insight into how AI is providing value to their developers and the impact AI is having on organizational performance.
Thanks to DX for sponsoring this newsletter, let’s get back to this week’s thought!
Intro
I’ve seen many failed projects not because of bad tech, but because of bad communication and teamwork.
Conway’s law is very real:
Organizations that design systems are constrained to produce designs that are copies of the communication structures of these organizations.
But many organizations are focusing on improving individual performance, instead of teamwork. Especially when it comes to AI.
And I believe the reason is: Not a lot of resources on how we can leverage AI to improve teamwork.
So, for today’s article, this is going to be our main question that we'll want to answer:
How can we leverage AI to improve teamwork, especially in engineering organizations and teams?
To help us with this, we have Henry Poydar with us today, who has been working closely on this exact topic with his company Steady for close to 2 years now!
P.S. We’ve met with Henry at the University of Maryland earlier this year, where we talked about how we can improve coordination in engineering teams.
You can read all about the event and what we talked about here:
Introducing Henry Poydar
Henry Poydar is Founder and CEO at Steady, with nearly 25 years of experience in software engineering and leadership → leading product and engineering teams.
For nearly 2 years, he has been closely working on how to utilize AI to improve teamwork within teams and across teams. Today, he’s kindly sharing his insights with us!
Everyone’s talking about AI for tackling individual workflows. But what about teamwork?
GenAI has changed how we write code. But it hasn’t yet changed how we work together as engineers.
What if we could take the benefits of AI code assistants, assembling context, surfacing decision points, and removing boilerplate, and apply them to teamwork itself?
In this article, I’ll offer an approach to do just that. But first, let’s ground ourselves in a few first principles.
There is no greater technology than a team of humans
We are fundamentally social creatures. Humans literally cannot survive alone, we die without connection, collaboration, and shared purpose. These are biological facts, not philosophical conjectures.
Throughout history, our greatest achievements have come not from individual genius, but from coordinated human effort.
The Apollo program wasn't just staffed with brilliant engineers, it was about 400,000 people working in perfect sync toward a shared vision.
Linux didn't emerge from Linus Torvalds alone, but from thousands of contributors building on each other's work across decades.
Even today's AI revolution proves this point. OpenAI and Meta aren't using enormous salaries to lure in the best minds so they can work on their own, they pay them to join the team.
And the most valuable AI startups aren't built around individual contributors working in isolation, but around tight-knit groups who can move fast, think together, and coordinate complex technical decisions at speed.
The ancient Greeks had a word for this: synergia.
The idea that the combined effect of a group working together exceeds the sum of their individual efforts. Modern science backs this up: diverse teams consistently outperform even the most talented individuals when tackling complex problems.
The three ingredients of effective teamwork
So what makes a team truly effective?
Over my 25 years working with high-performing engineering teams (plus data and insights from modern management science, like Google's Project Aristotle), three key ingredients show up again and again:
Trust and transparency: confidence in each other through clear communication
A balance of accountability and autonomy: zero micromanagement required
Real-time context: who’s working on what, where, how, and, most importantly, why
You need all three.
But the secret sauce in today's fragmented modern work environment is the third one: context → it’s what makes the first two possible.
Picture your last nightmare project:
Engineers building features that don't fit together, product managers chasing status updates, everyone working hard but nothing clicking.
That's what happens when context is absent and people guess at it. Trust erodes alongside poor decisions based on incomplete information. Accountability devolves into micromanagement because managers can't tell if work is on track.
And autonomy drifts into chaos because teams build the wrong things even with the best intentions. And crucially, we're all unhappy, because shared purpose is unclear.
Now picture your last breakthrough project:
That magical team flow state where the backend engineer anticipates frontend needs, the QA lead writes tests before seeing tickets, and someone catches a critical but obscure bug in code review.
Everyone moves like they're reading each other's minds because they're all working from the same rich understanding of what's happening and why it matters.
That's context in action.
When teams have real-time visibility into who's doing what and why, transparency becomes natural, trust builds organically, and people can own their outcomes without constant oversight.
And crucially: the right information, at the right time, in the right hands doesn't just make teams more efficient → it makes them happier.
A shared brain for teamwork
If context is the foundation for trust, autonomy, and accountability, the next two question are:
how do we deliver it at scale?
And if we're going to leverage genAI, what's the foundational blueprint for doing so?
The answer isn’t yet another set of dashboards or KPIs, it’s something fundamentally different: a shared brain for teamwork.
Every high-performing team eventually develops one → a collective sense of what’s happening, why it matters, and how the pieces fit together.
But this mostly breaks down at scale. What works for a 5-person team doesn’t translate across teams of teams, time zones, and functions. Context gets siloed, decisions get lost, and alignment turns into overhead.
What if you could build that shared understanding systematically and scale context across the entire org?
A true shared brain operates on three levels:
System of context → Situational awareness: who’s working on what, why, how, and with whom. Real-time risks and opportunities.
System of record → Institutional memory: what decisions were made, why they were made, and what we learned. Slices of context over time that we can mine over time for insights.
System of action → Intelligent execution: constantly collecting, distilling, and proactively delivering relevant context to the right people, at the right time.
Together, these systems form a coordination intelligence layer that strengthens human judgment instead of replacing it.
The shared brain doesn’t make decisions for you, it keeps everyone aligned on what’s happening, why it matters, and where things are going.
Not so fast: humans are the loop
Before we look at ways to implement a shared brain, it’s tempting to imagine a future where AI simply runs the team: flagging blockers, assigning tasks, keeping everything humming.
But that fantasy misses something essential:
Teamwork and coordination isn’t an algorithmic problem. It’s a human one.
AI can help. It can assemble context, surface anomalies, and suggest next steps (see below). But it can’t read between the lines.
It doesn’t know that a late PR reflects burnout, not laziness.
It can’t sense when alignment is performative.
It doesn’t feel the stakes in a strategic tradeoff. Only humans do.
So we don’t automate decisions. We automate context assembly.
In other words, a shared brain for teamwork should make it easier for humans to make good calls, not take the calls out of our hands.
And importantly, it should introduce just the right amount of friction.
Not the drag of endless meetings or noisy dashboards, but deliberate moments of reflection and clarity. Writing. Thinking. Enough to prompt good questions, sharpen assumptions, and keep everyone aligned on reality.
Practical ways to build your team's shared brain
Let me be direct: there's competitive pressure on you to figure this out. Teams that master AI-enhanced coordination will move faster, ship better products, and attract better talent. Teams that don't will feel increasingly left behind.
Here's how you can start building your shared brain today with intention and the AI tools you (probably) already have access to.
1. Designate your context broker
Every team needs someone whose explicit job is collecting and distributing relevant context. This is usually your engineering manager or team lead, but it could be a senior engineer or PM who has the trust of the team.
This person (maybe it's you?) isn't a scribe or note-taker. They're a curator. Their job is to make sure the right information reaches the right people at the right time to inform decisions, and orchestrate how AI is used for context assembly.
This person is the facilitator and gateway to source-of-truth context.
2. Collect forward-looking context
Here's where most teams get it wrong:
They focus entirely on what already happened instead of what's about to happen.
Intent is the most powerful way to balance accountability and autonomy, which we know is key to team performance.
Set up a simple async cadence for capturing forward-looking insight:
Weekly to monthly: The responsible person for each project writes a sentence or two about their next steps and any blockers they see coming.
Daily to weekly: Each engineer jots down their intentions for the next day or two, not what they did, but what they plan to do and why.
This forces agency in a good way → it makes people think about whether they really understand the vision and context. And crucially, if someone's planning to build the wrong thing, you can catch it before they start.
Keep this part focused, don't collect everything.
Sometimes meeting summaries have good next steps, but most of the time, long email chains, slack threads, and full meeting transcripts are situational or backward looking.
A sentence from a human about what's going to happen slices through all of that noise.
3. Distribute context systematically
Take all that forward-looking context and feed it into a shared LLM project as context (a GPT in chatGPT, a project in Claude, or a "gem" in Gemini).
Ask it to create a succinct team or cross-team brief that highlights connections, flags potential conflicts, and surfaces opportunities for collaboration. If you're wondering what this prompt should look like, ask it to create the prompt!
Distribute these briefs via email or share them in your weekly sync meetings.
Again, the goal isn't more communication → it's better communication, laden with context, that actually helps people make decisions.
4. Build your team's pattern recognition
Now that you're building up a corpus of relevant context, set up another LLM project specifically to house it over time.
Feed it the slices of context from point 3, plus your team's goals, common challenges, and institutional knowledge. Use it to mine this growing system of record for insights about what works and what doesn't.
Again, I'll leave the prompts to you, but if you're stuck, ask your project for a prompt.
Caution for both 3 and 4: make sure you understand your company's data requirements. Rarely do any companies want any outside LLMs to training on company data, so it's important to know your interactions are scoped before using these tools, just like you would with coding assistants.
5. Set the tone as a leader
Write your own forward-looking updates alongside your engineers. Show them what good context looks like.
Crucial: read everything people write, or they'll stop writing it.
Your attention is what makes the system work. When people see their updates being used to make better decisions, they'll invest more effort in making them useful → it's a form of recognition and respect.
6. Have a point of view about tools
Everyone on your team is experimenting with individual AI workflows → different prompts, different tools, different approaches. Don't let this happen in isolation.
Document what's working and what isn't. That's context too! Share successful patterns across the team. Build a shared understanding of how AI tools can and should be used for coordination, not just individual productivity.
Wrap up
The teams that figure this out first won't just ship faster, they'll fundamentally change what's possible when humans work together.
That's a big idea, but then again, humans working together is a historically proven big idea, even in the AI era.
While everyone else debates whether AI will replace engineers (it won't, by the way), you'll be using it to unlock something far more powerful: the collective intelligence of your team.
Start small, stay human-centered, and build your shared brain one context slice at a time. Your future self, and your team will thank you.
Last words
Special thanks to Henry for sharing his insights on this important topic with us! Make sure to check him out on LinkedIn and also check out Steady, they are doing a lot of cool stuff!
We are not over yet!
This Is Holding Most Engineers Back from Lead Roles
Through coaching and mentoring many engineers who wanted to grow to lead roles, I’ve seen this repeating pattern: A lot of these engineers were good technically, but when it was time to delegate, trust others and let go of control, that was the hard part.
Learn why that is the case and how to make this shift in this video!
New video every Sunday. Subscribe to not miss it here:
Liked this article? Make sure to 💙 click the like button.
Feedback or addition? Make sure to 💬 comment.
Know someone that would find this helpful? Make sure to 🔁 share this post.
Whenever you are ready, here is how I can help you further
Join the Cohort course Senior Engineer to Lead: Grow and thrive in the role here.
Interested in sponsoring this newsletter? Check the sponsorship options here.
Take a look at the cool swag in the Engineering Leadership Store here.
Want to work with me? You can see all the options here.
Get in touch
You can find me on LinkedIn, X, YouTube, Bluesky, Instagram or Threads.
If you wish to make a request on particular topic you would like to read, you can send me an email to info@gregorojstersek.com.
This newsletter is funded by paid subscriptions from readers like yourself.
If you aren’t already, consider becoming a paid subscriber to receive the full experience!
You are more than welcome to find whatever interests you here and try it out in your particular case. Let me know how it went! Topics are normally about all things engineering related, leadership, management, developing scalable products, building teams etc.