Assess Engineering Team Structure Using AI

Teams will increasingly include both humans and AI agents. We need to learn how best to configure them. A new Stanford University paper "ChatCollab: Exploring Collaboration Between Humans and AI Agents in Software Teams" reveals a range of useful insights. A few highlights: 💡 Human-AI Role Differentiation Fosters Collaboration. Assigning distinct roles to AI agents and humans in teams, such as CEO, Product Manager, and Developer, mirrors traditional team dynamics. This structure helps define responsibilities, ensures alignment with workflows, and allows humans to seamlessly integrate by adopting any role. This fosters a peer-like collaboration environment where humans can both guide and learn from AI agents. 🎯 Prompts Shape Team Interaction Styles. The configuration of AI agent prompts significantly influences collaboration dynamics. For example, emphasizing "asking for opinions" in prompts increased such interactions by 600%. This demonstrates that thoughtfully designed role-specific and behavioral prompts can fine-tune team dynamics, enabling targeted improvements in communication and decision-making efficiency. 🔄 Iterative Feedback Mechanisms Improve Team Performance. Human team members in roles such as clients or supervisors can provide real-time feedback to AI agents. This iterative process ensures agents refine their output, ask pertinent questions, and follow expected workflows. Such interaction not only improves project outcomes but also builds trust and adaptability in mixed teams. 🌟 Autonomy Balances Initiative and Dependence. ChatCollab’s AI agents exhibit autonomy by independently deciding when to act or wait based on their roles. For example, developers wait for PRDs before coding, avoiding redundant work. Ensuring that agents understand role-specific dependencies and workflows optimizes productivity while maintaining alignment with human expectations. 📊 Tailored Role Assignments Enhance Human Learning. Humans in teams can act as coaches, mentors, or peers to AI agents. This dynamic enables human participants to refine leadership and communication skills, while AI agents serve as practice partners or mentees. Configuring teams to simulate these dynamics provides dual benefits: skill development for humans and improved agent outputs through feedback. 🔍 Measurable Dynamics Enable Continuous Improvement. Collaboration analysis using frameworks like Bales’ Interaction Process reveals actionable patterns in human-AI interactions. For example, tracking increases in opinion-sharing and other key metrics allows iterative configuration and optimization of combined teams. 💬 Transparent Communication Channels Empower Humans. Using shared platforms like Slack for all human and AI interactions ensures transparency and inclusivity. Humans can easily observe agent reasoning and intervene when necessary, while agents remain responsive to human queries. Link to paper in comments.

We tried every AI team structure. They all failed. AI-first teams. Human-first teams. Hybrid models. Pair programming with GPT-5. Then we stopped thinking about AI as a team member. Here's the structure that finally worked: We organize around problems, not roles. Each "pod" has: - A Problem Owner (human): Defines success - A Solution Explorer (human + AI): Finds approaches   - A Quality Guardian (human): Ensures standards - An Implementation Sprinter (human + AI): Builds fast - A Context Keeper (human): Maintains knowledge Notice what's missing? "AI Engineer" or "Prompt Engineer." AI isn't a role. It's a tool each person uses differently. The Problem Owner uses AI for market research. The Solution Explorer for ideation. The Quality Guardian for automated testing. The Sprinter for code generation. The Context Keeper for documentation. Same GPT-5. Five different applications. The breakthrough: Stop asking "How do we integrate AI into our team?" Start asking "What problems need solving, and who's best equipped to use which tools?" Our velocity doubled when we stopped treating AI as a separate thing. Your team structure should mirror your problems, not your tools. What organizational antibodies are you fighting while implementing AI?

Great AI-assisted development does not start with prompts. It starts with structure. This “Claude Code Project Structure” visual highlights something many teams overlook when adopting AI for engineering workflows: If your repository is messy, your AI output will be messy too. What stands out here is the intentional design: - a clear project context layer (CLAUDE.md) - reusable skills for repeated workflows like code review, refactoring, and release support - hooks for guardrails and automation - dedicated docs for architecture, decisions, and runbooks - modular src/ ownership for focused implementation context This is bigger than just repo hygiene. It is about building an environment where AI can operate with: clarity, consistency, safety, and scale. As AI becomes part of the software delivery lifecycle, the winning teams will be the ones that treat: - context as infrastructure - prompts as reusable assets - governance as a built-in capability - modularity as an accelerator That is how you move from one-off AI experiments to repeatable engineering systems. I especially like the reminder around best practices: keep context minimal, prompts modular, decisions documented, and workflows reusable. That is not just good for Claude or any coding assistant. That is good software engineering discipline, period. The future of AI-enabled development will belong to teams that know how to combine: architecture + workflows + governance + developer experience How are you structuring AI context and reusable workflows inside your engineering projects today?

If you’ve just joined a new engineering team and have an approved internal AI agent (where sharing code and sensitive info is allowed), here’s a smart way to ramp up and add value: [1] Start by reading through the main codebase, internal docs, and onboarding materials, don’t skip the groundwork yourself. [2] Dig up old bug reports, incident reviews, and support tickets to see the team’s real-world pain points. [3] Feed summaries of recurring issues (never raw code or sensitive data!) to your internal AI agent, and ask it to surface patterns, compare with open-source practices, or suggest possible root causes. [4] Use AI to help you spot where your team’s codebase genuinely struggles, be it tech debt, recurring outages, security, or reliability gaps. But before you take action: [1] If something looks “broken” or odd, always double-check with teammates first. Sometimes what seems like a mistake is a conscious tradeoff or legacy choice. [2] Once you’re sure, draft a short note or 1-pager about the gap and possible improvements, then invite feedback. Make it a conversation, not a lecture. This approach helps you onboard faster and show that you respect the team’s history and process. All the best!

The New Job of Engineering Managers When AI Joins the Team AI has quietly become the newest member of every engineering team. The question is no longer “should we use AI” but “how do we lead when AI is part of the team’s workflow.” Engineering managers now have a different job than even two years ago. It’s not about replacing engineers with models. It’s about building a team that works with AI the same way they work with testing tools, build systems, or cloud services. Here is what an AI native engineering manager actually does: 1. Shift the team from task output to system thinking Anyone can generate code with AI. Only strong teams can design systems. Your job is to help your engineers zoom out, reason about architecture, tradeoffs, failure modes, and long-term maintainability. AI handles typing. Humans handle thinking. 2. Build workflows where AI removes cognitive load Teams that win are the ones who stop treating AI as a “code machine” and start using it for reviews, scaffolding, debugging, documentation, architecture diagrams, and learning. Managers must set up these workflows so engineers spend their energy on design, not boilerplate. 3. Coach for judgment, clarity, and decision making AI can draft five options. Only an engineer with good instincts can choose the right one. Your role becomes less about unblocking tickets and more about strengthening judgment and reasoning under ambiguity. 4. Redefine collaboration norms AI creates parallel streams of work. Context gets scattered. Good managers create rituals where engineers explain decisions, record assumptions, and keep the team aligned even when AI is moving everything faster. 5. Protect quality and long-term health AI can generate ten times more code. Without stronger review, testing, and standards, you inherit ten times more tech debt. Your job is protecting the codebase from hidden risk while still unlocking speed. 6. Make experimentation normal AI workflows evolve weekly. The best managers create an environment where trial, error, and iteration feel natural. Teams learn together instead of pretending they have it figured out. AI will not replace engineering managers. But managers who ignore AI will slowly lose relevance. The teams that thrive will be the ones where humans design the system and AI accelerates the work. That’s the new job. And it’s a good one. #EngineeringLeadership #AINativeTeams #FutureOfWork #SoftwareEngineering #TechLeadership #AIInEngineering #EngineeringManagement #TeamCulture #AIProductivity #BuildBetterTeams

Your next high-performer might not be human... A 776-person experiment at P&G just showed that an individual working with AI can match the performance of a two-person team working without it. Let that sink in. Researchers from Harvard, Wharton, and P&G studied how GenAI reshapes teamwork, expertise sharing, and even the emotional experience of collaboration. Participants worked on real product innovation challenges across four working models… 🔹 Individual, No AI 🔹 Team (R&D + Commercial), No AI 🔹 Individual + AI 🔹 Team + AI The results challenge how we think about how organizations are designed. Here are the results... 1️⃣ AI elevates individuals to team-level output Individuals using AI performed at the level of cross-functional teams. They worked 16% faster and produced longer, more comprehensive solutions. AI didn’t just help. It recreated the core benefits of teamwork… cognitive diversity, iteration, exploration. 2️⃣ AI dissolves functional silos. Without AI, the pattern was predictable: R&D → technical ideas Commercial → market-facing ideas With AI? Both groups created balanced solutions, regardless of background.   3️⃣ AI improves how work feels. Participants using AI reported: ☑️ More enthusiasm ☑️ More energy ☑️ Less frustration They matched, or exceeded, the emotional benefits of having a human teammate. While I disagree…. The results showed... AI didn’t just improve output. It improved the experience of doing the work.   4️⃣ Breakthrough ideas come from AI + humans together. Teams using AI were 3× more likely to produce top solutions. Meaning... AI raises the floor for everyone. But the ceiling still comes from human–AI collaboration. 5️⃣ Workers underestimate their AI-enabled performance. Despite producing stronger work, AI users felt less confident in their output. The performance boost is real... but not yet internalized.   So, what does this mean for every leader out there? This isn’t “add AI to workflows.” This is re-architect how work happens. For decades, organizations designed around the assumption that only humans could integrate expertise, broaden perspectives, and generate cognitive diversity. That assumption, based on this research shows, that no longer holds. So, if you're a leaders, ask yourself.... --> Team structure: If one person + AI = two-person team, how do you resource work? --> Expertise strategy: How do talent development, hiring, and mobility change when AI lifts non-experts to expert-level output? --> Skill priority: Are you treating AI interaction as a core performance skill? --> Workflow design: What changes when human constraints shift? --> Confidence calibration: How do you help employees trust their AI-enabled output? The research basically tells us that organizations that treat AI as a cybernetic teammate, not a tool added onto old processes, will operate at a fundamentally different speed and scale. Don't look at AI as another tool, think of it as another person.

If you’re in leadership, you need to understand *how* genAI will transform your organization, and what that means for restructuring teams. Here's what we're learning: BREAKTHROUGH IN AI IDEATION OpenAI is getting ready to launch new AI models (o3 and o4-mini) that can connect concepts across different disciplines ranging from nuclear fusion to pathogen detection. (Reporting from The Information's Stephanie Palazzolo and Amir Efrati). Molecular biologist Sarah Owens used the system to design a study applying ecological techniques to pathogen detection and said doing this without AI "would have taken days." THE NEW TEAMMATE EMERGES Remember the HBS study with 776 Procter & Gamble professionals? It showed that genAI functioned as an actual teammate. Individuals using AI performed at levels comparable to traditional human teams, achieving a 37% performance improvement over solo workers without AI. Teams using AI were three times more likely to produce top-quality solutions while completing tasks 12.7% faster and producing more detailed outputs. BREAKING DOWN SILOS That study showed that AI also dissolves professional boundaries. Without AI, R&D specialists created technical solutions while Commercial specialists developed market-focused ideas. With AI, both types of specialists produced balanced solutions integrating technical and commercial perspectives. A NEW KIND OF TEAM AI users reported higher levels of excitement and enthusiasm while experiencing less anxiety and frustration. Individuals working alone with AI reported emotional experiences comparable to those in human teams. That's wild. RESTRUCTURING FOR ADVANTAGE The HBS study showed that AI reduces dominance effects in team collaboration. When genAI translates between roles, it accelerates iteration at a pace that there’s no way traditional teams could match. ++++++++++++++++++++ THREE THINGS YOU SHOULD BE DOING NOW: 1. Upskill your entire workforce: Develop a fundamental behavioral shift in how teams interact with AI across every task. This only works if everyone is doing it. (We work with enterprise to upskill at scale - more below.) 2. Experiment with new team structures: Test different AI-team combinations. Try individuals with AI for routine tasks and small teams with AI for complex challenges. Find what works best for your specific needs. 3. Redefine success metrics: Set new standards for what good work looks like with AI. Track not just productivity but also idea quality, knowledge sharing across departments, and team satisfaction—all areas where AI shows major benefits. ++++++++++++++++++++ UPSKILL YOUR ORGANIZATION: When your company is ready, we are ready to upskill your workforce at scale. Our Generative AI for Professionals course is tailored to enterprise and highly effective in driving AI adoption through a unique, proven behavioral transformation. It's pretty awesome. Check out our website or shoot me a DM.

Some of the best AI breakthroughs we’ve seen came from small, focused teams working hands-on, with structured inputs and the right prompting. Here’s how we help clients unlock AI value in days, not months: 1. Start with a small, cross-functional team (4–8 people) 1–2 subject matter experts (e.g., supply chain, claims, marketing ops) 1–2 technical leads (e.g., SWE, data scientist, architect) 1 facilitator to guide, capture, and translate ideas Optional: an AI strategist or business sponsor 2. Context before prompting - Capture SME and tech lead deep dives (recorded and transcribed) - Pull in recent internal reports, KPIs, dashboards, and documentation - Enrich with external context using Deep Research tools: Use OpenAI’s Deep Research (ChatGPT Pro) to scan for relevant AI use cases, competitor moves, innovation trends, and regulatory updates. Summarize into structured bullets that can prime your AI. This is context engineering: assembling high-signal input before prompting. 3. Prompt strategically, not just creatively Prompts that work well in this format: - “Based on this context [paste or refer to doc], generate 100 AI use cases tailored to [company/industry/problem].” - “Score each idea by ROI, implementation time, required team size, and impact breadth.” - “Cluster the ideas into strategic themes (e.g., cost savings, customer experience, risk reduction).” - “Give a 5-step execution plan for the top 5. What’s missing from these plans?” - “Now 10x the ambition: what would a moonshot version of each idea look like?” Bonus tip: Prompt like a strategist (not just a user) Start with a scrappy idea, then ask AI to structure it: - “Rewrite the following as a detailed, high-quality prompt with role, inputs, structure, and output format... I want ideas to improve our supplier onboarding process with AI. Prioritize fast wins.” AI returns something like: “You are an enterprise AI strategist. Based on our internal context [insert], generate 50 AI-driven improvements for supplier onboarding. Prioritize for speed to deploy, measurable ROI, and ease of integration. Present as a ranked table with 3-line summaries, scoring by [criteria].” Now tune that prompt; add industry nuances, internal systems, customer data, or constraints. 4. Real examples we’ve seen work: - Logistics: AI predicts port congestion and auto-adjusts shipping routes - Retail: Forecasting model helps merchandisers optimize promo mix by store cluster 5. Use tools built for context-aware prompting - Use Custom GPTs or Claude’s file-upload capability - Store transcripts and research in Notion, Airtable, or similar - Build lightweight RAG pipelines (if technical support is available) - Small teams. Deep context. Structured prompting. Fast outcomes. This layered technique has been tested by some of the best in the field, including a few sharp voices worth following, including Allie K. Miller!

How do you design an AI Org? Spoiler alert, it’s not an HR exercise, it’s an architectural exercise . The decision to design an AI org across business, technical, ops, and talent functions isn’t just a talent mapping exercise. It’s a near-term architecture exercise with long-range implication mapping. That’s why it’s an architecture discussion more than a head count discussion - here me out. AI orgs orbit around three high-tension interfaces (and if you ignore them, it’s only a matter of time before something explodes): 1. Product ↔ Engineering. Why? Because business viability vs. technical feasibility needs to be aligned - a misaligned build solves nothing 2. Ops ↔ Tech. Why? Because eventually your build needs to get out of POC and into Production - so if a model crashes and bad business decisions are made as a result AND YOU DONT HAVE a feedback loop to course correct, that’s a major fail! 3. Talent ↔ Business. Why? You’re hiring must match the market & your roadmap - late or wrong hires stall initiatives and cost $. This isn’t theoretical. Promise! I once walked into a Fortune 500 boardroom where the CTO said: “We’re launching 11 AI use cases next quarter.” I asked, “Who’s owning the runtime layer, monitoring, retraining, and fine-tuning?” Crickets. They had hired 10 prompt engineers and thought they built an AI org. 😳 So I helped retrain the mindset and the final takeaway was: 🎤Org design ≠ headcount. It’s about leverage and scale 🙌🏼 Start with overlap, then specializations. AI teams wear many hats early on. Recruit for interface thinkers and overlap magic makers because your MVPs sit across many overlapping boundaries: product/tech, infra/ethics, change/adoption. If you’re building your AI team now, ask yourself: 1. Who owns the runtime layer? 2. Who’s translating business value into technical feasibility? 3. Who’s responsible for feedback loops once the model is live? 💡 I’ve built org structures across 6 industries and scaled hybrid human+AI teams globally. I can provide the exact roles, reporting lines, and critical questions you need to ask. If this is of interest, say “YES” below and I’ll drop a full write-up on how to design an AI org that scales and where it should sit! #FutureOfWork #AILeadership #OrgDesign #TechStrategy #AITransformation

Most engineering teams bolt AI onto their existing process. Sprint planning, design review, implementation, code review, QA, deployment. The line gets faster, but it's still the same line. At Abnormal, agents now generate the spec, and the spec is what agents execute. Blog post + claude plugin: https://lnkd.in/gE9Pvz5m Here's what that actually looks like in practice: - Self-updating architecture files -- Every design review is recorded. An agent processes the recordings, Slack threads, and PR comments weekly and proposes updates to the system files. The next spec any engineer generates automatically incorporates the feedback. - Two-audience specs -- The top half is for human reviewers (problem statement, architecture, trade-offs). The bottom half is for agents (function signatures, inline implementation guides, verification steps). - Compliance before code -- Security, legal, and architectural constraints are encoded in markdown files the spec tool reads on every run. Violations get flagged with citations before a line of code is written. - Non-engineers building production features – The system files encode enough organizational knowledge that the spec tool fills in everything they don't know. #AIEngineering #DevTools