Learn or Die: 11 Steps to a Learning Organization

A Coursera license that stays unopened. A training budget that expires each December with money still in it. I hear about both in mentoring sessions, and both sit under "learning culture" in the org chart. Learning happens somewhere else: in the twenty minutes after something breaks.

In those twenty minutes the team either asks what in the system let this through and changes it, or patches the bug and moves on. Which one your organization does by default is set by structure, and it's set long before the incident. Who owns the retro. Whether the fix is allowed to touch another team's code. Whether the manager who reports the failure upward gets a question or a reprimand. AI makes this urgent, because the skills your teams need now change faster than an annual training plan can be rewritten, so the organization learns on the job or falls behind while the budget sits unspent.

I pulled eleven steps from patterns across 3,611 mentoring sessions with engineering leaders growing their teams. They come in an order, and two of them fail if you skip the first one. You'll leave with the eleven steps, the order, and the one to start on Monday.

AI Engineering Manager in 2027: How to Lead 30+ People With No Burnout

Your teams adopted AI and got faster. Your calendar did not. Jan Vrtiska, Senior Engineering Manager at Everpure, went from 8 people in one team to 26 in three teams between 2025 and 2026, on the same hours. His teams shipped about 30% faster; his double-booked hours went up 2.5x. The bottleneck moved from writing code to coordination, and the manager is the coordination.

The talk walks through what he changed: the four routines he repeated every week (chasing status, collecting context, spotting risk, preparing 1:1s), the question he asked about each one (which part can AI do, which part needs me), and the roughly 100 plain-English rules he wrote down so a script could apply them. The script runs every four hours. AI gets called only where judgement is needed. Anything that touches people waits for his yes. We also cover the afternoon Claude rebuilt his 26 recurring 1:1s and one report's agent complained to his agent.

You walk out with the first AI management workflow to try next week: pick one repetitive task, write how you decide it as 3 to 5 rules, let AI apply them, fix the rules where it is wrong. Plus the 2027 career-fork table, so you can see which manager shape you are already drifting toward.

Delivered as a workshop at Applifting's Actually Intelligent AI, Prague, September 2026. Full page with data and sources: https://www.marian.coach/speaker/ai-engineering-manager/

45-60 min keynote or 90-min hands-on workshop. Story and numbers from Jan Vrtiska's talk at ELC meetup #43, credited on the slides.

5 Levels of AI Adoption: The KPIs You Should Be Measuring

Most CTOs I mentor ask a version of the same question: we bought the AI seats, is it working? The ladder has five rungs above zero. Level 0, still debating. Level 1, licenses bought. Level 2, AI code in production. Level 3, the bill climbs. Level 4, you can say whether it is worth it. Level 5, you ask whether you are the best.

Each rung has its own mistakes: the split team at level 1 (a few already paying out of pocket, most experimenting, the rest scared it replaces them), the review backlog at level 2, the $20k weekend agent at level 3. Navigara's delivery data across 500 open-source projects shows the shape of the climb: a 12% dip in engineering performance during the experiments, +104% per developer once it works, then a flat line while spend keeps rising. We go rung by rung and you place your own company on the ladder, with the two KPIs that prove it.

You walk out with a five-level KPI table: two numbers to track per level, and four vanity numbers to stop reporting (seat utilisation, acceptance rate, share of AI-written code, self-reported speed).

Full page with data and sources: https://www.marian.coach/speaker/5-levels-of-ai-adoption/

45-60 min keynote or 90-min leadership workshop. For CTOs, VPs of Engineering, Heads of AI. Delivery data from Jirka Bachel (Navigara) at ELC meetup #39, credited on the slides.

Human Performance in the AI Era

Leaders and engineers use AI, and most of them don't get back the capacity they expected. Human performance in the AI era is the part of leadership that stays human: judgement, challenge and accountability.

The talk has two parts. AI as a Capacity Multiplier: we redesign leadership work so AI removes the low-value cognitive effort, and the time it frees goes to work that needs a human leader. Human Judgement in the AI Era: we define where AI can support a leadership decision, and where human judgement, challenge and accountability have to stay.

You walk out with the /critical skill, an AI thinking partner that scores any plan or decision from 0 to 10 and names what breaks. AI then challenges your judgement instead of replacing it.

Full page: https://www.marian.coach/speaker/human-performance-ai-era/

45-60 min keynote or 2-session workshop. For CTOs, VPs of Engineering, Engineering Managers, senior engineers.

Mindset Change: From Employees to AI People Delivering Value

Most engineers still price themselves like employees. Hours in, title up, raise once a year. That broke the moment one person with AI started delivering what a team used to.

So run yourself like a portfolio. A skills portfolio is three layers of skills priced together, the way an investor prices a portfolio. Funds: your core craft, say senior backend. ETFs: the supporting skills, product sense, decision-making, communication, network. Crypto: the virtual layer, agentic coding, AI code reviews, AI planning. We sort which of your holdings AI just made cheap, and where to rebalance.

You walk out with the 8-step market-value playbook and the Developer Value Calculator, so you can score your own position before the next salary conversation.

Full page: https://www.marian.coach/speaker/portfolio-of-skills/

45-60 min keynote or 90-min workshop. For senior ICs, Staff+ engineers, Tech Leads. My most requested talk.