Dealer Dashboard
OCR purchase intake, stock, invoicing, VAT, the SAPS register, documents and a full audit log.
Thirty years running motor retail — then I built the software it actually needed. Six products shipped into production in two years, including a dealer management platform in live daily use.
Thirty years signing the deals and closing the month. I know which features get used and which get ignored.
Eleven modules on one dashboard, designed, built and shipped into daily use at a working dealership.
Anthropic API document reading and agent workflows — with a person confirming anything that reaches a ledger.
VAT, the statutory register and full audit trails. In this trade, being wrong has a regulator at the end of it.
OCR purchase intake, stock, invoicing, VAT, the SAPS register, documents and a full audit log.
Messaging, homework, attendance, fees and medical — priced for schools that can't carry enterprise software.
Email to branded quote and payment link in ninety seconds; a new client's whole workspace provisions itself in under an hour.
Capsules of photo, video or voice, sealed until a server — not the device clock — says the date has come.
Vehicles, property, goods and services with optional escrow, verified users and a facilitator posture.
Name, logo and full site designed and built end to end — the studio brand behind every product on this page.
Secure sign-in and session management
Business logic and service orchestration
Row Level Security enforced at the database
Data integrity, relations and audit trails
I use AI agents to move faster through spec, build and review — and I draw a hard line where being wrong gets expensive. An LLM drafts; a person confirms. Nothing reaches a VAT ledger or a statutory register without a human signing it off. A misread VIN isn't a bug you patch next sprint — it's a compliance problem with a regulator at the end of it.
The four-part framework for working with AI systems in practice.
Where the technology genuinely helps — and where it must not be trusted.
Core capabilities, prompting discipline and operational limits.
Building on the Claude Platform from the ground up.
The full spectrum of working with Anthropic models through the API.
Running Claude in production on AWS infrastructure.
Claude Code in a daily development workflow.
Long hands-off sessions: steering, hooks, scheduled routines, verified runs.
Agentic workflows across files, tools and connected systems.
Packaging repeatable expertise into skills an agent can invoke.
Delegating work to specialised agents and composing the results.
Connecting models to live tools, data sources and business systems.
Advanced MCP — transports, auth and production server design.
Delivering the AI-fluency framework to other people as structured teaching.
Applying AI fluency to teaching practice and institutional strategy.
AI fluency built for the classroom, with Teach For America.
AI collaboration for learning, career planning and academic work.
Organisational impact and efficiency without losing the mission.
AI fluency aimed at people who ship software, not just use it.
Applying AI where a small business actually feels it — time and margin.
Twenty Anthropic Academy certifications completed in August 2026 — the full agents, skills, MCP, API and AI-Fluency track. The AI Fluency curriculum is developed with University College Cork and endorsed by Ireland's Higher Education Authority. Every certificate carries a credential ID and is listed publicly on my LinkedIn profile.
Available now — remote, or on-site with relocation for the right role.