AI & Agentic Consulting
Most AI consulting gives you a slide deck. I give you a working pipeline, a trained team, and a clear picture of where AI agents create real value, where they introduce risk — and the controls that make them safe.
The problem
There is a widening gap between what AI tools can genuinely do today and what most consultancies will show you. On one side, a senior engineer with Claude or Cursor can deliver in days what used to take a team a month. On the other, businesses are buying PowerPoint decks from people who have never actually shipped agentic software into production.
Yellow Melon exists to close that gap for companies that are serious about getting AI and agentic workflows right — practical, implemented, measured, not slideware.
Engagements
A structured review of how your team currently builds software — and where AI and agentic workflows would create the biggest wins. Delivered as a prioritised roadmap with cost/benefit estimates.
Building real agentic pipelines into your development, support, or operations workflows. Practical implementation, not slide decks.
Working sessions and async guidance to help your existing engineers use Claude, Cursor, Copilot and the broader AI ecosystem effectively — including the sharp edges most articles skip.
Ongoing embedded advisory for teams adopting AI tooling who want senior judgment without a full-time hire.
Why me
I have particular depth in fintech, accounting software, and legal tech — industries where AI needs to meet a high bar on correctness, auditability and compliance.
Common questions
Yes — the focus is narrowly on software engineering and business workflow automation, delivered by a developer who ships production code with these tools every day. You're not getting a strategy deck from a generalist; you're getting implementation advice from someone in the trenches.
North American companies are my primary focus, but I also work with UK and European clients. I work from Tulum, Mexico, on US Eastern time, and bill in USD.
My strongest experience is in enterprise accounting systems and financial services. For legal or other specialist workflows, I work with your domain experts to define requirements, review risks and validate the result.
A structured foundation in how agentic systems actually behave — the failure modes, evaluation frameworks, and design patterns that distinguish production-grade agent workflows from demos that fall over the first time a user does something unexpected.
A short email describing what your team is working on is the fastest way to start.
Working with teams across the US, Canada, UK and Europe
From the resource library
Buyer’s guide · 3 min read
Compare AI agents and workflow automation with business examples, approval controls and a decision checklist. Choose the simplest approach that fits the task.
Read the guide: AI agents vs workflow automation: which does your business need? →
Engineering notes · 4 min read
James Drummond explains an approach to AI coding agents built around clear requirements, architecture, human approval, testing and production ownership.
Read the article: How I use AI agents without handing over engineering judgment →
Buyer’s guide · 3 min read
Compare Zapier, Make, n8n and custom integrations by workflow fit, error recovery, maintenance and ownership. A practical guide for business decision-makers.
Read the guide: Zapier, Make, n8n or custom integration: how to choose →