What are the alternatives to big consulting for AI?

Governance, org-wide change management, procurement comfort, and a counterparty large enough to carry enterprise risk. At the scale those things matter, they are worth real money and nobody else supplies them.

What you pay for it: months of discovery before anything is built, six- to seven-figure engagements, partners who sell and juniors who deliver, and knowledge that tends to stay with the vendor on a retainer. None of that is a scandal. It is simply the shape of the product.

A boutique automation consultancy — one process into production in weeks, published pricing, handover to your team. Fits when the goal is a specific result this quarter. A sector specialist — deep experience in one regulated industry; Neurons Lab, for instance, works almost entirely in financial services. Fits when the rules of your sector are the hard part.

A software house — design, mobile, QA, and engineering in one team; Codebridge is built that shape. Fits when you are building a whole product, not automating an existing process. An in-house hire — full control and permanent capability, at hiring speed and payroll cost. Fits when automation is going to be continuous rather than a project.

Size the problem before shopping for a vendor. One process losing hours → boutique. A whole product to build → software house. Regulated-sector rules doing the damage → specialist. Permanent, continuous capability → hire. An organisation-wide transformation with board reporting → yes, actually a big consultancy.

The expensive mistake is buying the programme shape for a one-process problem. You get governance you did not need, on a timeline you could not afford, and the process is still manual in month four.