Mark Sear

AI Maximalist · Enterprise AI Operator · Investor-Operator

I build production AI systems for enterprises.

Director of AI Forward Engineering at Maersk. Building GenAI systems that move beyond pilots and into the operating fabric of global businesses.

I work at the point where frontier models meet real organisations: architecture, workflows, governance, evaluation, human approval, and the messy operational layer where AI either compounds or dies.

Read the thinkingGet in touch

Still shipping code.

Receipts

Proof, not theatre.

Enterprise AI is full of demos. The hard part is production.

  • Leading AI Forward Engineering at Maersk

  • Building CX AI systems that turn human-readable SOPs into executable workflows

  • Live across 50+ countries in Ocean freight processing

  • Processing 400k emails/month — 1M by end of June

  • Claude Code and Codex in daily use

  • 30+ years across enterprise technology cycles

Point of view

AI systems, not AI pilots.

Most enterprise AI does not fail because the models are weak. It fails because the operating system around the models is missing.

01

Encode the organisation

The valuable asset is how work actually gets done: SOPs, exceptions, approvals, playbooks, judgment, and institutional memory.

02

Build the operating layer

Durable AI requires routing, tools, retrieval, evaluation, observability, governance, and human-in-the-loop design.

03

Govern to scale

Governance is not the brake. Done properly, it is what lets AI move safely from experiment to production.

04

Stay close to the work

Credible AI leadership now requires proximity to the tools. Strategy without hands-on fluency decays quickly.

Under dark skies

View the gallery →

Landscape, street, and travel photography from hikes, city breaks, and the road between the UK and Puglia.