BetterCloud AI Governance: Onboard, Authorise and Offboard Agents Like Employees
BetterCloud, a CoreStack company, added AI spend tracking and agent lifecycle management to its IT governance platform on 29 October 2026: IT and FinOps get centralised visibility into AI consumption, while security teams create, audit and offboard AI agents using the same automated workflows they already run for employees
Official Links
Features
- ✓ Spend tracking unifies fixed monthly seat licences with variable consumption such as API tokens in one dashboard, across major model providers, over 30- and 90-day windows
- ✓ Agent lifecycle management: create, update and retire agents, authorise which tools each agent may use, and grant or revoke user access to agents
- ✓ Manages AI data stores and AI projects as well, so the data behind an agent does not become a governance blind spot
- ✓ Offboarding in one run: reassign an employee's agents to their manager, suspend the Google Workspace account, revoke Microsoft Entra ID sessions and remove AI tool access in parallel
- ✓ Every automated run lands in execution history, closing the loop with existing shadow-AI discovery across Cursor, Anthropic, OpenAI and GitHub Copilot
How to Use
- Inventory first: list every AI tool and agent already in use, especially departmental purchases IT never approved — the shadow AI problem
- Connect seat licences and consumption fees into one dashboard and split by user and model; find duplicate subscriptions and accounts nobody actually uses
- Give every agent a lifecycle: who approved its creation, which tools it may call, when it retires — an agent should not outlive its purpose
- Put agents into the offboarding runbook: reassign their agents first, then revoke platform sessions and AI tool access in parallel, and check execution history to confirm each step actually fired
FAQ
What exactly did BetterCloud ship?
Per the release on 29 October 2026, BetterCloud (which describes itself as a CoreStack company) released AI spend tracking and agent lifecycle management, extending its IT governance platform to enterprise AI: centralised visibility into AI consumption for IT and FinOps, plus automated create, audit and offboard workflows for AI agents on the security side. Platform page at https://www.bettercloud.com/platform/ai-governance with the release carried by ACCESS Newswire via Yahoo Finance
What does agent lifecycle management actually cover?
The release lists creating, updating and retiring AI agents; authorising which tools each agent can use; granting and revoking user access to agents; and managing AI data stores and AI projects. The offboarding example is concrete: one workflow reassigns an employee's agents to their manager, then suspends the Google Workspace account, revokes Microsoft Entra ID sessions and removes AI tool access in parallel, with each run logged to execution history
How granular is the spend tracking?
The described capability unifies fixed monthly seat licences and variable consumption such as API tokens into a single dashboard, identifies unused or duplicate paid subscriptions across departments, and tracks consumption across major model providers over 30- and 90-day windows for forecasting and avoiding surprise bills. It does not claim per-call cost attribution, so if you need cost split by business line you still need your own metering layer
How does it relate to the August AI Governance launch?
BetterCloud shipped AI Governance on 28 August 2026, focused on shadow AI discovery, pulling Cursor, Anthropic, OpenAI and GitHub Copilot into one governed inventory. October's release layers spend tracking and agent lifecycle on top: August answered which AI is in use, October answers what it costs, who owns it and what happens when the person leaves
What size company actually needs this?
The trigger is dispersion, not headcount. If AI tools are bought team by team on individual cards, model API keys are scattered across a dozen projects, and nobody reclaims access when staff leave, the governance gap already exists. Conversely, with one or two centrally procured AI accounts and no cross-team agents, a spreadsheet plus an approval step is enough — no platform required