ETB Annual Meeting, Seattle 2026
Focus
The future of agents in the enterprise. We have circled this topic all year, and this is the meeting where we go deeper on the learnings that matter: P&L returns and capital allocation, organizing for success and the hybrid workforce, architecting the stack across both AI and IT, and trust in agents and the risk surface.
Format
Closed-door sessions are candid, off-the-record exchanges among a highly curated set of cross-industry peers, all of whom carry P&L and the regulatory exposure for their organizations. These are the honest conversations that demos and sales pitches do not allow.
Open-door sessions bring in selected industry and provider executives to deliberate on what comes next: the challenges they foresee at the frontier of models and technology, and a range of perspectives on the questions we are holding today.
Closed-door sessions
The answers are in the room. Come prepared to intensely debate and bring examples and lessons learned. This is where collective intelligence is best.
1. ROI is an organizational achievement, not a technical one. Capital is flowing; proof is not.
- Confidence has outrun measurement. What separates the projects that show real P&L impact from the ones that quietly stall, and what does disciplined capital allocation look like in practice?
- Returns come from redesigning how work flows, not from layering agents onto existing processes. Where has that paid off without boiling the ocean, and how?
- Our people, performance, and learning systems were built for human teams, not hybrid ones. How are we re-planning our workforce to capture value?
2. The bottleneck has moved from the model to our own foundations and architecture, and the bet underneath it.
- Most pilots stall before production, what are we learning on data, integration, evaluation, adoption and auditability that delivered outcomes?
- The architecture choice, whether a single integrated stack, an open and portable one, or a sovereign build, is hard to reverse once agents are wired in. How are we making the call and keeping optionality at scale?
- Tokenomics has become a board-level concern, and value is settling in the orchestration layer that routes work to the right model at the right cost. What routing and cost controls actually work?
3. Trust, control, and crossing borders.
- Governance is better understood as the thing that lets us move faster than the brake that slows us. What does a working control stack that grants autonomy at pace look like today?
- Identity and lifecycle management are the missing primitives. What controls, in identity, oversight, and the ability to revoke authority, are we requiring before we widen an agent's remit?
- The rules are fragmenting by jurisdiction faster than most global operating models can absorb. How do we stay compliant without building a separate stack for every region?
Inputs Requested
Please help sharpen the few areas we can chose to spend time on.
Open-door sessions
With selected provider and industry guests. These go wider than the closed room, into the macro forces, the frontier, and the implications. Come with a point of view, and stay open to being moved.
A. Owning the learning loop.
The frontier debate has shifted from model capability to who captures the value. The argument gaining ground, voiced most prominently by Satya Nadella in mid-June, is that the firm that wins builds and owns a learning loop on top of rented models, turning its people's judgment into a proprietary asset rather than ceding it to a few large models. We are seeking outside perspectives on how value actually flows between models, platforms, and enterprises, and whether owning your own intelligence is realistic or rhetoric.
B. The frontier of autonomy.
Single agents are giving way to multi-agent systems, agents that transact, and agents that supervise other agents, with more autonomy over longer time horizons and emerging agent-to-agent ecosystems across vendors. We are seeking the labs' and startups' read on what is real, what is hype, and what we should be architecting, staffing, and governing for over the next eighteen to twenty-four months.
C. The external risk surface: threats and rules.
Two outside forces are moving faster than most operating models can absorb. On threats, prompt injection is becoming the new malware, agents are emerging as both an attack surface and an insider-threat vector, and the shift is already reshaping security practice and cyber-insurance underwriting. On rules, the EU Act, China's new agent framework, the Five Eyes guidance, and Singapore's graduated-autonomy and liability models are converging and diverging at the same time. We are seeking outside perspectives on where attack and defense are heading, and where regulation is actually going, so we can build for it rather than react to it.
D. The political economy of agents.
The broadest question, and the one with the least consensus. As agents absorb more work, the debate is turning to societal permission, the concentration of value in a few models, and the effect on jobs, industries, and the social contract. Senior voices across the industry now warn that there is no societal license for an AI future that hollows out whole sectors. We are seeking outside voices to push our thinking beyond the walls of the enterprise, on what this means for how we lead, hire, and position our organizations.
Outside Guests
We would love recommendations on which of these we can cover (likely not all) and any suggestions on whom we may invite.
Executive Technology Board (c)