AI agents are running wild in enterprises, and security is lagging behind. A staggering 69% still share credentials across agents. That’s an open invitation to breaches and confusion about who did what.
Shared credentials don’t just invite security incidents—they blur accountability. Fixing identity management helps but isn’t enough. Enterprises must enforce action-level authorization and tamper-proof audit trails.
“These organizations need to prove to auditors in a very tamper-resistant fashion that records actually reflect their actions,” says Mukesh Karki, CTO of NTT DATA AIVista. He stresses governance must be external and cover every agent action.
Karki likens AI agents to interns. They mean well but need close oversight and trust built over time. Enterprises can’t realistically vet thousands of agents individually, so governance layers must cover agents, models, and data.
Scoped credentials are just the start. Action-based and rules-based constraints must shape how agents operate. Governance needs to be baked in from day one. Retrofitting governance after deployment is complicated and risky.
Groundcover, a newcomer in observability, raised $100 million this year, pushing total funding to $160 million. It has over 250 paying customers and tripled annual recurring revenue in the past year.
Groundcover’s tech uses eBPF to monitor systems. Its architecture follows a bring-your-own-cloud model with pricing based on infrastructure size, not data volume. CEO Shahar Azulay says, “We’re seeing telemetry exploding. Users are frustrated by not getting all value from existing platforms.”
He adds observability must evolve because AI systems generate far more telemetry. Future platforms will serve both AI agents and human operators.
Meanwhile, Asana’s AI Work Management (AWM) leverages its 18-year-old Work Graph to let AI agents access company goals and share memory company-wide. Arnab Bose, Asana’s Chief Product Officer, explains, “It’s not just looking at a prompt or a file. It works off a shared ledger for the whole company.”
AWM includes data governance to block confidential data leaks and features dynamic model routing to simplify prompt engineering. It also charges a flat fee per task to keep pricing predictable. Cloud provider CoreWeave uses AWM to automate product launches and task assignments.
Rimini Street jumped into AI governance too, launching Rimini Govern™ for AI. CEO Seth Ravin promises 24/7/365 managed governance, security, and interoperability. “Organizations can now confidently deploy AI agents and scale operations with needed oversight and ROI measurement,” he says.
R “Ray” Wang, CEO of Constellation Research, highlights their global managed service and deep AI expertise. He says, “This lets organizations operationalize AI with speed, cost-effectiveness, and confidence.”
Gartner’s recent forecast isn’t optimistic. By 2027, 40% of enterprises will demote or decommission autonomous AI agents due to governance gaps. The message is clear: without proper controls, AI agents risk becoming liabilities.
In short, AI agents are here and multiplying. Enterprises that keep sharing credentials and skip governance will pay dearly. The future belongs to those who build trust, enforce action-level controls, and audit every move. Otherwise, expect chaos and costly rollbacks.
Based on
- AI agents can create database sprawl issues. YugabyteDB’s solution is more agents! — thenewstack.io
- NTT DATA AIVista and Snowflake: Identity alone won’t secure enterprise AI agents | VentureBeat — venturebeat.com
- How is your enterprise tracking AI agent telemetry? Groundcover thinks it should never leave your cloud | VentureBeat — venturebeat.com
- Asana’s AI agents share memory across your company — but not your secrets | VentureBeat — venturebeat.com
- Rimini Street Launches Rimini Govern™ for AI to Deliver Comprehensive AI Agent Governance, Security & Interoperability as a Service | National Post — nationalpost.com
