Platform coverage
Agentic register
across multiple platforms
Coverage is subject to platform technological limitations. Any custom coverage is decided on a case-by-case basis and may incur additional charges.
T H E S H I F T I S A L R E A D Y H A P P E N I N G
Harnessing AI is difficult,
and it comes with a cost
Left ungoverned, agents drive unpredictable spend and an elevated risk
profile. The numbers show where it is heading.
W H A T Y O U G E T
One operational view of
every agent
Not another dashboard to staff. A managed operation that hands you the
evidence and the decisions, every month.
W H Y IT M A T T E R S
Agents act. But right now, no
one can answer for them.
Agents are already running in your business, built in Copilot Studio, Power Platform, Foundry and third-party tools, by IT and by your own people. These are the questions they raise, and how the AOC answers each one.
T H E M I S S I N G C E N T R E
Three operations centres,
one estate
You already run this model for infrastructure and for security. The AOC applies the same discipline to your agents.
How they connect. The AOC escalates agent risk to your SOC with owner and context, and flags agent capacity and spend to your NOC. One estate, no gaps, and never a second SOC.
C O V E R A G E
One control plane over every agent
A single governance layer over the Microsoft controls that secure your agents, with a direct line to the third-party agents that sit outside them.
Managed Agents
1:1 Meeting Assistant
GovernedAOC Operator (dev)
★ VIPManager 1:1 Assistant
GovernedPDR Coach
GovernedDrag the active panel, swipe on mobile or use your keyboard arrow keys.
H O W I T W O R K S
One team, the full lifecycle,
every month
Fast time-to-value. EU AI Act & US Federal AI Framework. Fully managed
and SLA-backed. Hover a step to see what it covers.
Discover
Find and inventory every agent across your estate, wherever it was built. Nothing is governed until it is first seen.
Discover
Find and inventory every agent across your estate, wherever it was built.
Classify
Risk-rank each agent and classify it against the EU AI Act and US Federal AI Framework.
Monitor
Watch continuously for spend spikes, alerts and risks, catching problems before they become incidents.
Triage
Investigate what the monitoring surfaces and act on it, with every finding routed to a named owner.
Report
Turn spend, risk and ROI into a board-ready story your stakeholders can act on, every month.
Optimise
Recommend keep, fix, promote or retire, so the estate gets better and cheaper, not just bigger.
B E A T T H E C L O C K
The law already expects things
of your agents
The EU AI Act’s transparency duties bite first, and the duty sits with you, the deployer, agent by agent. You cannot disclose for agents you do not know you have.
Transparency duties apply
Customer-facing agents must disclose they are AI. Article 50 transparency obligations and enforcement powers are live now.
Content marking
AI-generated content needs machine-readable marking. Existing systems get until December; new ones do not.
High-risk obligations
Annex III high-risk duties land, carrying deployer obligations that must be governed, evidenced and signed off.
High-risk is delayed. Transparency is not. The AOC classifies every agent, flags exactly what each rule requires, and keeps your evidence current as the rules change, so you can answer an auditor before they ask.
T H E B U S I N E S S C A S E
Cheaper to run than to build
The real alternative is governance in fragments across Agent 365,
Defender and Purview, with no owner and no evidence.
Illustrative estimate only. The in-house figure assumes a junior and a mid-level governance hire, fully loaded (around £125,000 a year). Actual costs and savings depend on your scope, licensing and estate, and are confirmed on application.
B E Y O N D T H E S A V I N G
The incident that never happened,
and the spend that never ran away
T H E M A N A G E D S E R V I C E
One team running the full governance lifecycle
Every month, on a rhythm you can rely on. Delivered on your tenant, under full change control, with a named service lead.
- Shadow AI Discovery management*
- Full Agent Inventory (unlimited)
- Agentic FinOps Reporting
- Service Report & Call – Monthly
- EU AI Act Classification (Annex III & Art. 50)
- Agent Performance Review – Annually
- Agent Value & Outcome Tracking
- Monthly proactive checks + Standard SLA
- Standard Changes – 5 / month
- Extended Agentic FinOps Insights
- Quarterly Exec Briefing
- Shadow Agent Discovery*
- Agent Owner Dashboards
- Agent Performance Review – Quarterly
- EU AI Act Obligation Notices issued
- Agent Policy Management*
- Weekly proactive checks + Premier SLA
- Standard Changes – 10 / month
- Balanced with your licensing
- Tuned to your situation
- Customisable level of service
- Volume discounts on Managed Agents
- Network control management for agents*
- Agent-level web filtering management*
- Multi-vendor & multi-region estates
* Availability depends on your Microsoft licensing and tenant configuration.
Free discovery session, onboarding included, minimum 12-month term. Add-ons on any tier: additional managed agents (volume-priced), Hyper-care daily checks, and a governance response retainer. Minimum Microsoft licensing prerequisites apply; enterprise volumes priced on application.
O U T C O M E S F O R Y O U R O R G A N I S A T I O N
Scale AI with confidence,
not guesswork
Everything you need to answer the three questions an auditor asks: what runs, what the law expects, and when you were told.
Illustrative. In-house estimate: a junior and a mid-level governance hire, fully loaded, about £125,000 a year. On Advanced (£7,000/mo) the hard saving is around £41,000. Breach data: IBM Cost of a Data Breach 2026.
THE CHANGING SOCIAL FAQ
Here Is The Most Frequently Asked Questions.
We know that investing in a comprehensive Agent Operation Centre is a significant decision for your business. That’s why we’ve put together a list of frequently asked questions about AI and Agents. We’ve got your queries covered. If you have further questions, feel free to reach out to us — we’re here to help!
Agentic Operations Centre
Govern every AI agent in your business.
Discovered, governed and accounted for, run for you.
A fully managed AI agent governance service for your Microsoft estate. Our team discovers, classifies and controls every agent, and reports it back to your stakeholders, so you can scale AI with confidence.
The shift is already happening
Harnessing AI is difficult, and it comes with a cost
Left ungoverned, agents drive unpredictable spend and an elevated risk profile. The numbers show where it is heading.
of enterprise software will be agentic by 2028, up from under 1% in 2024. Gartner
What you get
One operational view of every agent
Not another dashboard to staff. A managed operation that hands you the evidence and the decisions, every month.
Why it matters
Agents act. But right now, no one can answer for them.
Agents are already running in your business, built in Copilot Studio, Power Platform, Foundry and third-party tools, by IT and by your own people. These are the questions they raise, and how the AOC answers each one.
What is running?
No one can name every agent, its owner, or the data it can reach.
What is it costing?
There is no mature FinOps for AI. Spend climbs, unnoticed and unowned.
Who reports on it?
Your board, CISO and auditors have no owner and no evidence to show.
Who is accountable?
Orphaned and rogue agents drift with no named owner behind them.
Is it compliant?
EU AI Act duties sit with you, the deployer, agent by agent.
Is it safe?
Agents are a new attack surface, and most go unwatched.
The missing centre
Three operations centres, one estate
You already run this model for infrastructure and for security. The AOC applies the same discipline to your agents.
- Watches
- Identities, endpoints, data and threats.
- Answers
- “Are we under attack?”
- Run by
- Your security team or MSSP, 24/7.
- Watches
- Every AI agent: Copilot Studio, Foundry, Power Platform and supported third-party.
- Answers
- “What are our agents doing, costing and risking?”
- Run by
- Changing Social, as a fully managed service, in business hours.
- Watches
- Infrastructure, networks and devices.
- Answers
- “Is it up and performing?”
- Run by
- Your IT team or MSP, 24/7.
How they connect. The AOC escalates agent risk to your SOC with owner and context, and flags agent capacity and spend to your NOC. One estate, no gaps, and never a second SOC.
Coverage
One control plane over every agent
A single governance layer over the Microsoft controls that secure your agents, with a direct line to the third-party agents that sit outside them.
Swipe through the four dashboard panels →
1:1 Meeting Assistant
GovernedAOC Operator (dev)
★ VIPManager 1:1 Assistant
GovernedPDR Coach
GovernedData over time
MB uploaded by product
↔ swipe chart to see all products
Manus
Under reviewBaseten
Under reviewPolyAI
Under reviewDistribution by band
Managed agents grouped by spend rating
Score vs observed monthly spend
Validation against real cost history
As data arrives, the scorecard appears here.
Managed agents (4)
Sorted by score
AOC Operator (dev)
MEDIUMManager 1:1 Assistant
LOW1:1 Meeting Assistant
LOWPDR Coach
LOWAgent outcome ledger
Attainment vs targetHow it works
One team, the full lifecycle, every month
Fast time-to-value. EU AI Act & US Federal AI Framework. Fully managed and SLA-backed.
Discover
Find and inventory every agent across your estate, wherever it was built. Nothing is governed until it is first seen.
Classify
Risk-rank each agent and classify it against the EU AI Act and US Federal AI Framework.
Monitor
Watch continuously for spend spikes, alerts and risks, catching problems before they become incidents.
Triage
Investigate what the monitoring surfaces and act on it, with every finding routed to a named owner.
Report
Turn spend, risk and ROI into a board-ready story your stakeholders can act on, every month.
Optimise
Recommend keep, fix, promote or retire, so the estate gets better and cheaper, not just bigger.
Beat the clock
The law already expects things of your agents
The EU AI Act’s transparency duties bite first, and the duty sits with you, the deployer, agent by agent. You cannot disclose for agents you do not know you have.
Transparency duties apply
Customer-facing agents must disclose they are AI. Article 50 transparency obligations and enforcement powers are live now.
Content marking
AI-generated content needs machine-readable marking. Existing systems get until December; new ones do not.
High-risk obligations
Annex III high-risk duties land, carrying deployer obligations that must be governed, evidenced and signed off.
High-risk is delayed. Transparency is not. The AOC classifies every agent, flags exactly what each rule requires, and keeps your evidence current as the rules change, so you can answer an auditor before they ask.
The business case
Cheaper to run than to build
The real alternative is governance in fragments across Agent 365, Defender and Purview, with no owner and no evidence.
Illustrative estimate only. The in-house figure assumes a junior and a mid-level governance hire, fully loaded (around £125,000 a year). Actual costs and savings depend on your scope, licensing and estate, and are confirmed on application.
Beyond the saving
The incident that never happened, and the spend that never ran away
Shadow AI shows up in 43% of breaches, and the global average breach now costs $4.99m. We find and govern unsanctioned AI before it becomes an incident.
IBM Cost of a Data Breach, 2026EU AI Act transparency breaches carry fines up to €15m or 3% of global turnover. Every agent is classified and evidenced before anyone asks.
Low-value agents retired, and overnight token spikes caught, flagged and blocked before they run away with your budget.
Around 3,700 hours a year of standing governance work absorbed, so your team stays on the build roadmap instead of the admin.
Customer Zero
We run it on ourselves, first.
The AOC runs live on Changing Social’s own tenant, under full change control. We operate exactly what we sell, and our founding cohort helps shape it. A real snapshot from our own estate:
The managed service
One team running the full governance lifecycle
Every month, on a rhythm you can rely on. Delivered on your tenant, under full change control, with a named service lead.
- Shadow AI Discovery management*
- Full Agent Inventory (unlimited)
- Agentic FinOps Reporting
- Service Report & Call — Monthly
- EU AI Act Classification (Annex III & Art. 50)
- Agent Performance Review — Annually
- Agent Value & Outcome Tracking
- Monthly proactive checks + Standard SLA
- Standard Changes — 5 / month
- Extended Agentic FinOps Insights
- Quarterly Exec Briefing
- Shadow Agent Discovery*
- Agent Owner Dashboards
- Agent Performance Review — Quarterly
- EU AI Act Obligation Notices issued
- Agent Policy Management*
- Weekly proactive checks + Premier SLA
- Standard Changes — 10 / month
- Balanced with your licensing
- Tuned to your situation
- Customisable level of service
- Volume discounts on Managed Agents
- Network control management for agents*
- Agent-level web filtering management*
- Multi-vendor & multi-region estates
* Availability depends on your Microsoft licensing and tenant configuration.
Free discovery session, onboarding included, minimum 12-month term. Add-ons on any tier: additional managed agents (volume-priced), Hyper-care daily checks, and a governance response retainer. Minimum Microsoft licensing prerequisites apply; enterprise volumes priced on application.
Outcomes for your organisation
Scale AI with confidence, not guesswork
Everything you need to answer the three questions an auditor asks: what runs, what the law expects, and when you were told.
- A single source of truth for your entire agent estate
- EU AI Act obligations mapped and evidenced, agent by agent
- A predictable monthly cost in place of new headcount
- Spend under control, with an ROI story for the board
- Rogue and orphaned agents found, owned and resolved
- One owner for agent governance, working alongside your SOC
Illustrative. In-house estimate: a junior and a mid-level governance hire, fully loaded, about £125,000 a year. On Advanced (£7,000/mo) the hard saving is around £41,000. Breach data: IBM Cost of a Data Breach 2026.
Ready when you are
See what is really running in your tenant.
Book a discovery session and we will map your agent estate, flag the risks, and show you exactly what the AOC would govern. Scoped, revocable access, and no obligation.
Response within 1 business day. No spam. No obligation.
The Changing Social FAQ
Frequently asked questions
We know that investing in a comprehensive Agent Operation Centre is a significant decision for your business. That’s why we’ve put together a list of frequently asked questions about AI and Agents. If you have further questions, feel free to reach out — we’re here to help.
What is the Agentic Operations Centre?
The AOC is a fully managed AI agent governance service for your Microsoft estate. Our team discovers, classifies, monitors and governs every agent, and reports the outcomes back to your stakeholders. Think of it as the third operations centre alongside your SOC and NOC, run for you as a service rather than another tool for your people to staff.
How is this different from a tool or a dashboard?
Microsoft gives you the levers through Agent 365, Purview, Entra and Power Platform, but pulling them every week, with evidence, is a new operations job. The AOC is the team that does that job. You do not get another dashboard to run; you get the outcomes, the decisions and the audit trail.
How much does it cost and how is it billed?
The AOC is a monthly managed service. We start with a full discovery of your agent estate, then give you a customised quote based on what we find. Billing is annual in advance, which is our preferred option and comes with a discount. Otherwise it is quarterly in advance.
What is the difference between a monitored and a managed agent?
Every agent in your estate is monitored from day one, and inventory is unlimited. An agent becomes managed, receiving the full governance lifecycle, when it is high-risk under the EU AI Act, among your highest agents by spend or risk, or nominated by you. You govern what matters, and pay to deep-govern only what matters.
What access do you need, and is it secure?
We work through least-privilege, read-first access using Microsoft GDAP, never global admin. You grant the access, you see every use of it, and you can revoke it at any time. Every action on your tenant is ticketed, logged and reported back to you. We are Cyber Essentials certified, with ISO 27001 and 9001, and you choose whether we operate entirely inside your tenant or from a dedicated environment in your region.
Do you replace our SOC?
No. The AOC works alongside your SOC and MSP with an agreed escalation path, never as a second one. We provide expert judgment through the working day, and critical alerts still route to your SOC around the clock, with owner and context attached.
What does this mean for the EU AI Act?
The Act’s transparency duties under Article 50 apply from 2 August 2026, and they sit with you, the deployer, agent by agent. High-risk obligations were pushed to December 2027, but transparency was not extended, and you cannot disclose for agents you do not know you have. The AOC classifies every agent, flags what each rule requires, and keeps your evidence current as the rules change.
Which platforms and agents do you cover?
We cover agents built and run across your Microsoft estate: Copilot Studio, Power Platform, Microsoft Foundry and Microsoft 365. Enterprise adds shadow agent discovery, policy enforcement and threat detection through Microsoft Agent 365. Optional coverage is available for ChatGPT Business and Enterprise, and Claude Team and Enterprise, limited by what each platform’s console exposes.
Will this expose your internal methods or tooling?
No. You get the managed service, the reporting and the evidence. The way we run the operation, our internal tooling, playbooks and delivery methods, stays with us. You buy the outcome, not the mechanics.

