AI implementation and security consultant

01 / Implementation and integration

AI implementation and integration

A useful answer in a chat window is a start. The work is turning it into a process your team can repeat and check — connected to the documents, calendars, databases and business tools you already use.

• Workflow integration. We map one workflow from start to finish, decide where a model helps and where a person stays in charge, and I build the connection between the AI tool and your systems.
• MCP server development. Model Context Protocol servers give an AI assistant a defined set of actions in one system, and nothing beyond them. I write them for the systems your work depends on.

Public MCP servers I maintain

You can read the code before you talk to me. These are on github.com/aplaceforallmystuff, with 11 MCP servers in all:

mcp-arr · 227 ★

Runs the *arr media management suite (Sonarr, Radarr and the rest) from an AI assistant.

mcp-threatintel · 9 ★

One interface to threat intelligence feeds: AlienVault OTX, AbuseIPDB, GreyNoise and abuse.ch.

mcp-cti-glossary

Cyber threat intelligence lookups: MITRE ATT&CK, OFAC sanctions entries and threat-actor aliases across vendors.

spain-ai-kit · 15 ★

MCP servers for Spanish government open data and legal information.

mcp-wp-abilities · 3 ★

Exposes a WordPress site’s Abilities API to AI assistants as tools.

mcp-local-llm · 13 ★

Hands mechanical tasks from a cloud model to a local model running on Ollama.

Star counts taken 2026-10-15.

02 / Security review

AI security review

Before an assistant gets access to your systems, someone should check what it can read, what it can do, and what happens when the text it reads was written by an attacker. I review three things:

• Prompt injection. Where untrusted text — emails, web pages, uploaded files, support tickets — reaches a model that can take actions, and what that text could make it do.
• Data handling. What goes to which AI provider, what they keep, and whether that matches your obligations to your clients and regulators.
• Agent and tool permissions. Which credentials and actions each AI tool or agent holds, and cutting them down to what the job needs.

The result is a short list of changes, in order of risk, that you can make before rollout or apply to tools you already run.

03 / Enablement and training

Team enablement and training

A tool changes nothing until people know how to use it on their own work. I train teams to describe a task to a model, check what comes back, and keep confidential information out of it.

• Sessions are built on your team’s real tasks.
• Ground rules for which tools to use, for what, and how output gets reviewed before anyone relies on it.

Education is one of the three things I work across. I wrote How to Think Like a Coder (Batsford; shortlisted for the ALCS Educational Writers’ Award 2018) and How to Code in Minecraft, and I write Signal Over Noise, on practical AI for people doing real business work.

04 / Who it’s for

Who I work with

Small and mid-sized businesses, and teams inside larger organisations, that want AI doing useful work without building an AI team to run it.

I’m based in Valencia, Spain, on Central European Time, and work remotely, in English. I take on work for teams in the UK, the US, Australia and Canada, and we set meeting times that suit both time zones.