Start with the work people repeat
Map where information arrives, what needs checking, who makes a decision, and which system needs updating. AI can help interpret and prepare unstructured information; rules and APIs handle predictable steps. The aim is a workflow that fits the people using it.
Practical experience with AI tools
My work combines third-party AI tools with custom MCP servers, plugins, and agent skills. I connect assistants to proprietary applications, email, CRM, and business data so they can support email triage, research preparation, task management, and coordinated workflows. Existing tools and services, including Linear, are part of that wider operation.
Workflows I engineer
At Planet of the Vapes, I developed an assistant connected to the platform, email, HubSpot CRM, and task management for planning, vendor operations, and advertising workflows. At Drum & Bass UK, an assistant connects publishing and administration tools with email and business workflows. This experience informs the integrations and scoped automations I build for clients.
Make review part of the process
We agree which steps can run automatically and which need a person to review or approve them. The implementation records progress and reports failures so that your team can see what happened and what needs attention.
One defined workflow to begin
Start with a paid workflow assessment and one scoped implementation. You receive a process map, agreed inputs and actions, review steps, and acceptance checks. We can assess the result before extending automation to other processes.