Take the repeated work off your team, and keep the exceptions visible.

Where the problem shows up

When a task is slow, most businesses look for a tool that can do it. The approach is natural, because slow tasks are easy to see and the savings are easy to imagine. What it takes for granted is the task itself, which often exists only because of a gap somewhere else in the process.

Automation repeats whatever it is given, so a flawed step is carried out faster, more often and with fewer people watching. People learn to work around it, exceptions are handled in private spreadsheets, and the one person who understood why it behaves as it does eventually leaves. What happens when pricing, tools or structure change, and nobody can say which parts of the automation depend on the old arrangement? And how much of what is automated would still exist if someone had first asked whether each step was necessary?

The better principle is to trace the process before touching it, and to automate only the steps that still deserve to exist, with logging, alerts and a manual recovery path built in. That is the work I do.

What I have changed for others

HIPAA-regulated platform: 70% less operational workload

A nationwide addiction-treatment platform worked technically, but outdated content and repetitive administration were increasing cost and compliance risk. I led the platform end to end and built medical review workflows, SEO automation, more than 25 custom admin modules, and role-based access with audit logging. Outdated content fell by 80%, operational workload by 70% and facility publishing time by 50%. Read the case study.

100+ people: a weekly report written for each employee

Context sat in Slack, Gmail and meetings, and nobody could read all of it. I built a system with several specialised agents for extraction, interpretation, retrospection, trajectory analysis and summarisation, which produces a Weekly Trajectory Report for each employee and a company-level summary. An organisation of more than 100 people now has a recurring view of plans, progress and blockers. Read the case study.

26 ledgers: one close process with an exception tracker

Twenty-six company ledgers closed their months in different ways, and unresolved items collected until year-end. I designed a standardised close, reconciliation procedures, review steps and an exception tracker, documented in written SOPs. The target is one consistent control framework across all 26 ledgers within a 60-day implementation window. Read the case study.

What changes for your business

Repeated work leaves your team's week, and what remains is work that needs judgment. Exceptions are visible to a named person instead of disappearing into spreadsheets, and the automation can be understood and maintained by someone who did not build it. The work can include process mapping, workflow automation between existing systems, AI agents for extraction and reporting with defined limits, exception handling and documentation.

Where this leads next

Automation is dependable when the data beneath it is consistent, so many engagements move from information systems to automation and then to robotic process automation for rule-based tasks. If the process is already clear, we can begin with the automation itself.