AI-first engineering is not vibe coding with Copilot.
Best in class engineering organizations start with AI, keep AI at the core of the entire organization identity, and through acceleration and performance gain, influence the business to adopt AI at scale in every organization. I get struggling organizations unstuck with a proven framework for rapidly transitioning traditional engineering processes and workflows to AI-first business influencing operational units.
What AI-first actually means.
AI-first engineering is a structural redesign of how software gets built. The center of gravity moves from writing code to directing autonomous agentic systems against well-formed specifications.
The role of the engineer evolves with it. Today, engineers write implementation code, manually author tests, open PRs, and wait for human review. Tomorrow, they author and accept the spec, curate layered guardrails, direct an ensemble of agents, judge attestations at every gate, and own deploy timing and risk. I call this role the Software Development Orchestrator — the SDO.
Tools are downstream of this redesign, not the cause of it. Copilot is a tool. Vibe-coding with an AI is not a method. AI-first is the architectural and organizational shift that assumes agents do the work — and humans remain accountable for the outcomes.
The SDO pipeline.
Software moves from spec to production through six stages. At each stage, specialized agents do the work in parallel; when something falls short of the bar, it loops back to where it can be fixed. Every action leaves a verifiable receipt, and those receipts assemble into the Release Dossier — a complete, auditable record of how a change reached production.
The Surgical Team Model.
The Surgical Team Model puts a single accountable person in charge of an outcome from idea to production. Around them, an ensemble of skilled, intelligent agents does the work that — in a traditional engineering organization — would require many coordinated contributors.
That single person is the SDO — the surgeon. They understand the circumstance, know the right path to success, and coordinate the agent ensemble to deliver it. Their job is not to type code; it is to direct the work, judge its quality at every checkpoint, and own the outcome.
This model is the delivery on the promise of AI in engineering. It strips waste from traditional engineering frameworks, puts the engineer squarely in the driver's seat of ownership, eliminates ceremonial meetings, and gives every contributor the power to ship more, ship better, and ship faster. Early studies put per-engineer throughput gains in the range of 50% to 300% — and for the business, that translates to meaningful savings across go-to-market and customer support.
My take on AI and the future of engineering.
AI is a multiplier in the hands of the right person. It makes good engineers better and exposes significant gaps in weaker ones. The first move for any leader navigating this shift is to identify the leaders, champions, and stalwarts in your engineering organization — and, where the bench isn't deep enough, to hire the contributors who will help bring an AI-first mindset with them.
The shape of the team matters more than the tools they use. Small senior teams running properly architected autonomous agentic systems will outship large teams with weak processes every time. The fastest-shipping, highest-quality AI-native engineering organizations are the ones where engineers continue to contribute — just in a different way, playing an end-to-end role in building the systems that scale the business. The leader's job, and where I contribute most, is to build both: the key contributors, and the systems they'll use to accelerate your business.
Let's talk AI in your org.
If you're trying to figure out what AI-first actually looks like for your team — not in a keynote, but in practice — let's have the conversation.