The Mythical Man-Month strikes back: why throwing bodies at cloud migrations still fails in 2026

Fred Brooks told us in 1975 that adding people to a late software project makes it later. Fifty one years on, we are relearning that lesson, this time with AI copilots and aggressive migration timelines.

I have been leading data migrations long enough to have seen many flavours of this mistake. A programme falls behind. An executive sponsor gets nervous. The response is to add contractors, compress the timeline, and point at an AI code tool (or tools!) as evidence that velocity is now a solved problem. It doesn't end well.

Brooks's Law is not a historical curiosity. It is a structural reality, and cloud migrations are one of the harshest environments in which to rediscover it.

The 2026 anti-pattern

Here is the conversation I keep hearing. A CDO watches a demo: legacy SQL ingested, translated, unit tests generated, documentation drafted, all in minutes. The conclusion is obvious: migration velocity is now an AI problem, not a people problem. Greenlight the aggressive timeline. Add contractors mid-flight to hit the date.

The problem is that the demo shows the easy 60%. It does not show the rest.

The complexity in a legacy migration is not in writing code. It is in the tribal knowledge locked inside fifteen-year-old SSIS packages, the undocumented business rules baked into a SQL Server stored procedure, and the downstream consumer contracts nobody wrote down because the person who understood them left in 2019.

AI accelerates translation. It does not accelerate understanding. And adding five new contractors (or co-pilots) to a programme that is already struggling does not transfer tribal knowledge faster. It distributes confusion wider.

Where AI genuinely helps and where it does not

AI accelerates this Code translation (SQL, SSIS, Informatica) Test case generation from existing logic Documentation of known patterns First-pass data profiling and anomaly flagging Boilerplate pipeline scaffolding

AI does not help with this Semantic reconciliation of ambiguous rules Resolving conflicting stakeholder definitions Consumer contract negotiation Undocumented exception handling logic Organisational change and adoption

The hard 40% is deeply human work. It requires (say) a business analyst, a legacy developer, and a downstream data consumer to sit in the same room and reconcile three different answers to the question: what does this field actually mean? No copilot does that.

Right-sizing the cloud migration squad

The two-pizza rule exists for a reason. Small, stable, deeply context-loaded teams outperform large, fluid ones on complex migration work, every time. When I think about structuring a migration squad today, I prefer to map it to the domain boundaries of the target platform rather than the volume of work. Each squad owns its domain end-to-end: source logic, transformation, tests, documentation, consumer sign-off. They accumulate context. They do not dilute it.

Adding people mid-flight does not add capacity. It adds on-boarding overhead, communication surface area, and the very real risk that someone with partial context makes a decision that will take three months to unpick.

A staged ramp that respects the law

The alternative to the big-bang contractor surge is a staged ramp. Start small and knowledge-dense. Use AI tooling aggressively on the translatable work to generate velocity early. Let that velocity create the headroom for the hard work, not a reason to compress it. Add resource only when domain context can be transferred properly, not because a Gantt chart says a milestone is due.

This is not slower. In my experience it is substantially faster, because you are not spending the back half of the programme unpicking decisions made by people who did not have enough context to make them.

The executives who get this right in 2026 will be the ones who understand that AI has changed what is easy, not what is hard. Brooks's Law still applies. It just applies to a slightly different distribution of the work.

If this resonates, share it. If you are currently staring at a migration programme that is about to onboard fifteen contractors, let's talk first. wldluk.co.uk