AI agents and MCP

An AI agent is not a chatbot that answers prettily: it is software that understands an intent, queries your systems and executes actions — create a ticket, check availability, register a booking or update a CRM. I build those agents with your business's tools, not with promises.

Cases I solve

• Customer service — agents on the web or on WhatsApp that solve common questions, escalate to a person when needed and leave a trace in your system.
• Bookings and appointments — real availability, confirmation and reminders, integrated with your calendar or your ERP.
• CRM and sales — enrich records, draft follow-ups, classify leads.
• MCP tools — I expose your systems (APIs, databases, files) as tools that any agent or assistant can use in a controlled way.

How I do it

The agent can only do what you allow it: explicit actions, permissions by role, a log of everything it does and a plan B when it gets it wrong. I use local models or APIs (OpenRouter, OpenAI, Qwen, Ollama) depending on the balance you are looking for between cost, privacy and quality.

Related

It fits with applied artificial intelligence and with the APIs and integrations that give it access to your data.

Tired of copying data from the chat to the CRM and the email? Tell me the flow.