Power up the north.

The systems a business actually runs on, and the work of joining them to what is already there. Integration is where most of these projects fail, so it is where we start rather than where we discover a problem.

1. 1.1.1Production, inventory and equipment management
2. 1.1.2Custom ERP and MES
3. 1.1.3Commerce, corporate sites, iOS and Android
4. 1.1.4Maintenance after launch

Pipelines first, then models. A prediction built on data nobody trusts is a prediction nobody acts on, which is the expensive way to learn what an unglamorous ingestion job is worth.

1. 1.2.1Data pipelines and warehousing
2. 1.2.2GA4 and Looker Studio reporting
3. 1.2.3Prediction and recommendation models
4. 1.2.4Measurement design

Getting people to the thing we built, and knowing which ones arrived. Paid, organic and the conversion work in between, run against the same numbers the product reports.

1. 1.3.1Google, Meta and paid social
2. 1.3.2SEO, and answer engines that now sit above it
3. 1.3.3Conversion rate optimisation
4. 1.3.4Reporting a finance team will accept

The repetitive work between your systems, and the internal AI workflows people will actually adopt. Adoption is the hard half. A tool nobody opens is indistinguishable from a tool nobody bought.

1. 1.4.1Business process automation
2. 1.4.2Internal generative AI workflows
3. 1.4.3Rollout, training and handover
4. 1.4.4Governance and access control

Advertising, film and brand content produced with AI in the pipeline rather than as the headline. The tools changed what a week of production can hold; they did not change whether the result is worth watching.

1. 1.5.1Advertising and campaign film
2. 1.5.2Brand content
3. 1.5.3Motion graphics
4. 1.5.4Production for paid channels