PLDT deploys agentic AI to strengthen risk management
PLDT Inc. has rolled out its first agentic AI initiative to tighten enterprise risk management across the group. The company introduced the Enterprise Risk Intelligence Companion Agent (ERICA), an autonomous system built on its platform to help teams assess, identify, and manage risk with more speed and consistency.
"For the PLDT Group, the real value of ERICA goes beyond the technology itself… It reflects how we are embedding innovation into core enterprise functions in a way that is deliberate, disciplined, and aligned with how the business actually works," said PLDT and Smart Communications, Inc. First Vice-President and Group Head of Information Technology and Transformation Office Gilbert O. Gaw.
What ERICA changes for risk and operations leaders
- Standardizes risk assessments across units, using a consistent approach and language.
- Reduces manual effort so teams can spend more time on analysis and decisions.
- Translates complex operational data into clear insights that can be reviewed and validated.
- Adapts to the needs of different business units and flags critical risk areas tied to changes in operations.
ERICA was developed by the PLDT Group's Enterprise Risk Management team with PLDT and Smart's IT team and UiPath, an enterprise automation provider.
"ERICA was introduced to make risk assessments more seamless and consistent across the organization. By reducing the effort required at each stage, the solution allows teams to focus on insights and decision-making, rather than administrative efforts," the company said.
Governance and human oversight stay in the loop
PLDT emphasized that "human oversight remains central to ERICA's design." All outputs go through review and validation by risk professionals, with full traceability through existing governance systems. That supports transparency, accountability, and responsible use of AI aligned with practices such as the NIST AI Risk Management Framework.
Speed now, scale next
The company expects faster assessments and quicker translation of operational signals into actionable insights, improving response times. PLDT is preparing to extend ERICA across more business units and expand agentic AI adoption with UiPath to strengthen enterprise resilience.
What management should watch as you bring agents into risk workflows
- Define clear use cases: high-volume, rule-based assessments and monitoring where consistency matters.
- Keep humans accountable: set explicit review checkpoints and escalation paths for high-impact decisions.
- Traceability by default: require logs of inputs, prompts, versions, and decisions for audit and compliance.
- Data discipline: confirm data lineage, quality checks, and access controls before scaling.
- Change management: train line managers, clarify roles, and align SOPs so agents fit the existing control environment.
- Measure what matters: cycle time, coverage, false positives/negatives, and decision lead time.
- Vendor governance: include security, privacy, and continuity terms; test failover and model fallback.
Practical next steps to replicate this move
- Run a 60-90 day pilot on two processes (e.g., risk registers and control testing summaries) with a clear success metric.
- Integrate with your current GRC tool to maintain one system of record and full audit trails.
- Stand up a small AI risk council with ERM, IT, Legal, and Internal Audit for fast decisions on scope and controls.
- Document model limits and create a "human-in-review" rubric that defines when expert validation is mandatory.
- Publish a short playbook so each business unit can adopt the agent with minimal rework.
For leaders planning similar initiatives, see AI for Executives & Strategy and AI for Finance for governance, adoption, and risk-focused resources.
Your membership also unlocks:
