Generative AI · AI agents · Intelligent operations

I make AI work as a team, inside real enterprise operations.

I'm Mauricio Abarca, a generative AI and AI agents specialist with 10+ years in enterprise IT service management. I design multi-agent workflows where Claude Code builds, automates and talks to corporate AI such as Microsoft 365 Copilot, turning contracts, data and processes into decisions leadership can trust.

Based in ChileEnglish & Spanish

Enterprise AI levels I deliver

  1. 1Assisted AICopilot in everyday work
  2. 2AI on your documentsMicrosoft 365 Copilot over contracts and procedures
  3. 3AI agentsEnd-to-end process automation
  4. 4AI-driven managementMulti-agent orchestration with human sign-off
Mauricio Abarca
10+years in enterprise IT service management
15+executive dashboards published by AI agents
9management areas run by one AI-built system
70k+lines of AI-generated, human-validated code

See it in action

An Intelligent Management System, Built with AI

Nine management areas, 15+ executive dashboards and automated reporting, built with Claude Code and Microsoft 365 Copilot for a leading bank in Peru. Two minutes, illustrative data.

  • Loaded by AI
  • Reviewed by AI
  • Managed by AI
  • Deployed by agents, signed off by people

What I build

AI that does real work inside the tools a company already uses

Practical systems for operations teams, not demos: agents that read the data, apply the rules and deliver the reports, with a person signing off on what matters.

AI agents for IT operations

Agents that pull data from Jira, SharePoint, Outlook, Teams and Azure DevOps, calculate SLAs and publish reports on schedule, without manual copy-paste.

AI-to-AI integration

Claude Code consults enterprise copilots like Microsoft 365 Copilot about contracts and internal documentation, checks the sources they cite and applies only verified rules.

Multi-agent orchestration

Several AI sessions working in parallel, coordinated by a master session that reviews every deliverable, with persistent memory, skills and the Model Context Protocol (MCP).

Executive dashboards & reporting

Interactive dashboards and committee-ready PowerPoint and Excel reports, with KPIs and SLAs designed from hands-on ITIL 4 practice.

Approach

AI that checks AI, and a person who signs off

Generative AI is only useful in operations if its answers can be trusted. Every workflow I build follows the same four steps.

  1. Ask

    Claude Code asks the corporate AI (Microsoft 365 Copilot) what the contract and the procedures actually say.

  2. Verify

    It cross-checks the sources Copilot cites. Only rules backed by the source text are applied.

  3. Automate

    Parallel agents capture the data, run the SLA calculations and build the dashboards and reports.

  4. Validate

    I review every result before it reaches leadership. Human-in-the-loop, always.

Selected work

Recent projects

Banking · 2026 – present

An AI-built service management system for BCP (Banco de Crédito del Perú)

Challenge
Govern the ITSM processes and contract SLAs of Kyndryl's BCP account, with the data spread across several platforms.
What I built
With Claude Code, an agentic system that captures the data, calculates the SLAs and generates executive reporting automatically, across 9 management areas: incidents, changes, work orders, backlog and more.
Result
15+ executive dashboards and committee reports produced by AI agents, each one validated before it reaches leadership.
AI-to-AI · 2026

Verifying contract rules with Microsoft 365 Copilot

Challenge
SLA targets and calculation criteria had to match the exact wording of the client's contract.
What I built
A workflow in which Claude Code queries Microsoft 365 Copilot about the client's contracts and procedures and cross-checks every source it cites.
Result
SLA targets and calculation criteria corrected to match the contract text, with the source behind each rule.
Operations intelligence · 2023 – 2026

One executive view across multiple ticketing platforms

Challenge
Service data for enterprise clients such as Codelco, Grupo Security and CMPC lived in separate ticketing tools.
What I built
An executive management system in Power BI that unified those platforms, with service metrics based on ITIL 4.
Result
One shared view of service performance for technical teams and clients.

Client data, screens and figures stay confidential. Every example here describes the work, not the client's information.

Experience

10+ years in enterprise IT, now centered on AI

  1. Kyndryl
    • Process Owner, BCP account (Banco de Crédito del Perú)2026 – present
    • Senior ITSM Service Delivery Analyst2023 – 2026
    • Incident & Problem Process Owner2021 – 2023

    ITSM process ownership, SLA governance and executive reporting for enterprise clients, now run with AI agents I built.

  2. SILL
    • IT Project Coordinator2018 – 2020

    Arauco technology refresh: 4,000+ workstations deployed nationwide.

  3. IBM
    • Senior Service Dispatcher & Help Desk Agent2012 – 2015

    Led a 25-agent team for mining operations under ITIL, working daily in English with teams in India, Australia and London.

  4. Unisys
    • IT Support Specialist2011 – 2012

    Support for critical banking infrastructure.

Contact

Let's talk about AI agents and intelligent operations

I'm always glad to exchange ideas on AI agents, AI-to-AI integration and AI adoption in operations. The best way to reach me is by e-mail or LinkedIn.