The nine themes

1. Agent orchestration and enterprise architecture

The question: How do agents coordinate work across applications, platforms and vendors, and who controls the new layers?

Our position: Orchestrating work matters more than adding applications. Enterprise applications are unbundling into platforms and callable capabilities, and integration is the bottleneck that decides who wins.

This quarter we are asking:

• Is the coordination layer a product, a protocol or an open standard, and who owns it?

• When does native agent depth beat cross-platform breadth?

• Does MCP hold up as the default connection for governed, multi-vendor work?

• What happens to the application when vendors expose its capabilities to any agent?

What we want to hear in briefings: How your agents discover, call and coordinate with other vendors' agents. Protocol support, including MCP and A2A. What you expose headlessly. How conflicts between agents are resolved.

Start here: The Future Enterprise report and the Future Enterprise blog series, including the Agent Service Bus analysis.

2. Agent governance and operations

The question: What does it take to trust, control and account for agents in production?

Our position: Governance has to be designed in. Controls belong in the architecture from the start and outside the model, and every agent needs a named owner.

This quarter we are asking:

• What does continuous audit require in practice?

• Who is the single accountable owner of an agent?

• How do enterprises inventory, monitor and stop agents across platforms?

• Can certification stand in for continuous oversight?

What we want to hear in briefings: Agent inventory, monitoring and audit trails. Your identity and permission model for agents. Shutdown and incident notification. What is enforced versus only logged.

Start here: Agentic Governance-by-Design and the Accountable Autonomy series in The Digital Workforce.

3. AI strategy, operating model and readiness

The question: How do leaders tie AI to business strategy and know the organization is ready?

Our position: AI strategy is business strategy. Organizational readiness and agentic capability are two different problems, and they have to advance together.

This quarter we are asking:

• Why do most enterprises use AI without making it core to operations?

• Which readiness dimensions predict success with agents?

• How should boards and CEOs measure strategic AI impact?

• Where do organizations overshoot or undershoot their maturity?

What we want to hear in briefings: How you assess customer readiness. What separates customers who reach production from those who stall. Time to value.

Start here: The AI Strategy is Business Strategy series on the Arion blog, the Dual Maturity Framework, and The Complete Agentic AI Readiness Assessment.

4. Digital workforce and organization

The question: How do roles, talent, leadership and accountability change when agents join the team?

Our position: Humans lead. Agents and people work together, with people setting direction and holding accountability.

This quarter we are asking:

• What new roles does agent oversight create, and who is training people for them?

• How does oversight scale without burning out the people doing it?

• What changes in the org chart when agents take on defined jobs?

• Why does adoption stall on people and not technology?

What we want to hear in briefings: How your product divides work between people and agents. Approval gates. Evidence on adoption and change management.

Start here: Building the Digital Workforce, The Digital Workforce newsletter, and the Disambiguation podcast.

5. Economics of agentic AI

The question: How are agents priced, measured and insured, and who pays when they are wrong?

Our position: Measure outcomes, not activity. Whoever controls metering and attribution shapes the economics of the agentic enterprise.

This quarter we are asking:

• What replaces per-seat pricing, and does outcome pricing work?

• How do enterprises attribute return to agent work?

• How should AI spend be managed as its own category?

• Can insurers price agent liability?

What we want to hear in briefings: Your pricing model and how it is metered. Customer ROI evidence. Cost per agent or conversation at scale. Liability and indemnification terms.

Start here: The pricing analysis in the Future Enterprise series, and our ROI and KPI research reports.

6. Market, vendors and models

The question: Who is building what, how agentic is it, and which vendors will last?

Our position: A map needs a method. We classify products against published tests, not vendor labels.

This quarter we are asking:

• Which products are native AI versus AI embedded, and why does that matter to buyers?

• Where is consolidation heading?

• How should buyers assess model provider viability and concentration risk?

• Which emerging categories deserve a place in the Atlas?

What we want to hear in briefings: Product architecture and AI dependency. Agentic level, with evidence. Roadmap. Customer proof.

Start here: The Enterprise AI Atlas and The Arion Brief: AI at 10.

7. AI in go-to-market and customer experience

The question: How is AI changing how companies market, sell and serve?

Our position: Evidence over claims. We separate what is new from what is repackaged, and judge by measured revenue and service outcomes.

This quarter we are asking:

• Are agent governance and outcome reporting becoming buying criteria in revenue technology?

• What happens to the CRM when the interface moves to AI workspaces?

• Where do agents deliver measured results in sales, marketing and service?

What we want to hear in briefings: Measured customer outcomes. What agents do autonomously versus with approval. Data and context requirements. Pricing.

Start here: In the Hot Seat, and our customer experience, commerce and sales technology research reports.

8. Operational and vertical AI

The question: Where do agents create value in operations, finance, supply chain and industry-specific work?

Our position: Vertical depth beats horizontal breadth. The value is in specific, well-understood processes with controls built in.

This quarter we are asking:

• Which back-office processes are ready for agents today?

• How are mid-market ERP vendors bringing agents to market?

• What controls do agents need in systems that move money and materials?

• Where does vertical depth beat a general platform?

What we want to hear in briefings: Process coverage and pilot results. Approval and audit design. Integration with systems of record. Industry data advantage.

Start here: The use case and industry issues of The Digital Workforce and the vertical AI episodes of Disambiguation.

9. Data foundations and sovereignty

The question: What data quality, governance and control do agents need to work?

Our position: Data quality comes first. An agent reasoning over poor data produces confident wrong actions.

This quarter we are asking:

• Which data quality gaps cause agent projects to fail?

• How do sovereignty and regional rules change where agents and data can run?

• What does governance of agent memory and context require?

What we want to hear in briefings: Data requirements for your agents. Residency and sovereignty options. How context and memory are stored and governed.

Start here: Our AI privacy research reports and the data quality issues of The Digital Workforce.