Complete AI Training
Start for Free

Your job's AI kit

Get your AI kit

Tell us who you are and what you do. We show you your kit right away and email you the link: skills, prompts, AI agents, MCP servers and courses for your job.

500+ jobs ready, and we make a kit for any other job. No payment needed to look.

Share

AI news ·

Informatica bets on AI agents, rolling out Claire, a no-code builder, and an Agent Hub for data teams

Informatica's IDMC update centers Claire Agents and a natural-language Copilot for data work. Expect less grunt work and safer self-serve, with human-in-the-loop approvals.

Informatica puts AI agents at the center of its data strategy

Informatica is pushing agentic AI deeper into data operations with a Fall update to its Intelligent Data Management Cloud (IDMC). The release packages a set of purpose-built "Claire Agents," stronger reasoning, and new engineering tools that let teams build and govern their own autonomous data agents.

The pitch is straightforward: cut routine lift for data teams, give business users safe self-serve options, and move from rules-based workflows to outcome-focused decisions. All of it is accessible through a natural language interface via Informatica's Copilot.

What's new and why it matters

  • Agent-first model: Data management tasks are exposed through agents you can query, chain, or trigger through other agents. Informatica plans to make every IDMC capability agent-accessible over time.
  • Reasoning over reporting: Agents don't just raise flags; they can propose fixes based on context. Human approval remains in the loop, and approved actions can be automated going forward.
  • No-code agent engineering: A canvas to build, orchestrate, and govern custom agents. Bring your own models, vector databases, and domain logic. Support for Model Context Protocol clients and servers opens up interoperability.
  • Reusable components: AI Agent Hub acts as a directory for pre-built agents and automation recipes so teams can validate, share, and reuse.
  • Security updates: Controlled unmasking lets authorized users view protected data under policy. Multifactor authentication is now standard for all native IDMC accounts.

Claire Agents: what's available now

  • General availability: Data Exploration Agent (for MDM), Enterprise Discovery Agent, Extract/Load/Transform (ELT) Agent, and Product Help Agent.
  • Public preview: Data Quality Agent.
  • Private preview: Product Experience Agent.
  • Access: All agents are usable through Copilot's natural language interface.

Where executives should focus

  • Time-to-value: Informatica said the agent engineering toolkit can shrink build time from weeks to minutes. Pressure-test that claim with a narrow, high-value use case first.
  • Business self-serve: With reasoning and approvals, non-technical teams can handle more of their own data tasks. That's throughput without headcount-if governance is tight.
  • Agent sprawl: Preview programs show "triple-digit" agent counts at customers. Without standards, you'll end up with duplication, unknown costs, and policy drift.
  • Security posture: Controlled unmasking is useful for regulated workflows. Validate policies on unstructured data, audit trails, and escalation paths before scaling.

Practical use cases to pilot in 90 days

  • Data quality triage: Let the Data Quality Agent propose fixes on a critical dataset, keep approvals manual, measure rework reduction and cycle time.
  • Discovery and lineage: Use the Enterprise Discovery Agent to map sources feeding a key KPI; lock in a playbook for ongoing governance checks.
  • ELT automation: Have the ELT Agent generate and maintain pipelines for a specific domain, with rollback and cost monitoring in place.
  • MDM exploration: Put the Data Exploration Agent in front of business stewards to speed up matching, enrichment, and exception handling.

Operating model and controls

  • Agent catalog: Centralize ownership, templates, and naming. AI Agent Hub can help, but you still need intake and review gates.
  • Policies by default: Bake in PII handling, controlled unmasking rules, and MFA requirements. Treat agent actions like any privileged operation.
  • Metrics: Track task completion time, reduction in manual work, data issue recurrence, and cost per run. Kill or consolidate low-value agents.
  • Building blocks: Standardize models, prompts, and context stores. If you use vector databases, align on one approach and document usage patterns. A primer on vector databases is here: what they are and why they matter.

What to watch

  • Coverage: Informatica's goal is full agent access across IDMC. Validate gaps against your roadmap-governance, MDM, quality, lineage, ELT, and support.
  • Governance at scale: As agent counts grow, so will policy exceptions. Put audit, versioning, and incident response on day one.
  • Model choice: BYO models and MCP support increase flexibility. Standardize contracts and evaluation criteria for reliability, latency, and cost.

Bottom line

Informatica's Fall release moves data management from manual tickets to agent-driven workflows that propose, act, and learn-while keeping approvals in human hands. If you're spending too much on repetitive data tasks or waiting on backlogs, run a focused pilot with clear guardrails and measurable outcomes. Scale only after you've proven quality, policy compliance, and cost control.

Helpful resources

Share