iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
The central unresolved question at Madrona’s 2026 IA40 Summit was who keeps the customer relationship—and who may use the data created when an AI agent acts between a business and its customer. The summit’s speakers raised the issue repeatedly, but the event reporting offers no settled answer, legal conclusion, or contract terms.
What was the IA40 Summit, and what changed in the AI conversation?
Madrona’s IA40 Summit took place September 29–30, 2026, at the Four Seasons in Seattle. Its agenda covered agentic data, AI harnesses, pilots and return on investment, collaborative agents, trust, web tools, software, and autonomous systems. GeekWire describes the gathering as Madrona’s annual event for AI startups, investors, and technology executives. GeekWire’s event report and the official IA40 Summit page provide the event context.
A recurring theme was the move from promises about model capability to evidence of applied value: time saved, revenue generated, work completed, and new capabilities operating in production. Madrona’s interpretation is that value is accruing not only to foundation models, but also to agent systems, model aggregation, and the layers that connect AI to customers and deployment. That is Madrona’s reading of its list and market, not an independent measurement of the entire AI sector. Madrona’s 2026 IA40 overview explains that framing.
Who controls the customer relationship and agent-generated data?
An AI agent can act in the space between a company and its customer: it may find information, make a purchase, or complete work through business software. That raises two related but distinct questions: which company gets access to the customer, and who may use the record of what the agent did—including errors and corrections?
#1 Best Overall
GeekWire reports that moderator Raphaëlle d’Ornano asked who owns an agent’s work record and whether it belongs to the customer. Madrona Managing Director Matt McIlwain said the question of who gets to use data generated by people’s engagement with AI systems came up “over and over and over again.” The report says d’Ornano had never received a clear answer. Anthropic CTO Rahul Patil did not answer the ownership question directly; he said providers would use available data to improve agent systems. That response does not establish a contractual right to customer data or settle who owns it.
The distinction matters in practice. An agent’s activity record could be useful for improving a service, auditing a task, correcting an error, or understanding a customer’s needs. The summit reporting does not establish which party has a right to each use. It reports no legal analysis or contract language, so the ownership question should be treated as open rather than as a settled rule.
Could AI agents become the interface to business software?
Microsoft’s Charles Lamanna predicted that most business software will eventually be used by AI assistants acting on a user’s behalf. Town CEO Jean-Denis Greze said assistants can increasingly operate a browser or computer through its interface without an API, and described a future of “thin apps.” These are speaker predictions, not established outcomes.
Recommended Free Tools
Rank #2
Greze put the idea this way: “Everything is going to become a thin app because the better the AI gets at using the computer, the less the app matters as a unit of software.” If agents do become a common way people interact with business tools, the visible app may matter less to the user while the underlying access, permissions, and records of actions matter more to the organization.
What does agent-mediated shopping mean for customers and merchants?
Stripe’s Maia Josebachvili said agent commerce on Stripe had been roughly flat for eight or nine months before rising sharply over the preceding six weeks. That is her observation as reported by GeekWire, not a market-wide statistic. She also argued that when agents handle purchases, merchants can lose opportunities for add-on sales and advertising.
The same report said Amazon had blocked Meta’s Muse assistant from shopping on its site the previous month. The episode illustrates a business tension rather than a general rule: merchants may want to control how automated agents access their storefronts, while customers may value delegating product discovery and purchasing. The summit coverage does not establish that all retailers will permit or block agent shopping.
Why are organizations still struggling to deploy agents?
Speakers pointed to people, processes, and operational controls—not just model capability—as constraints on adoption.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Fix the driver behind crashes, sound loss and screen glitches3Clear out junk files and repair common Windows errorsWorkflows may not fit the technology
Goldman Sachs’ global head of AI product management, Archana Vemulapalli, said, “The bottleneck is actually not AI. The bottleneck is human.” Her point was that roles and processes were designed before AI; deploying an agent may require changing how work is assigned, reviewed, and completed.
Security, identity, and monitoring need to be in place
AWS’s Swami Sivasubramanian said teams building agents still needed to solve security, identity, and monitoring before rollout. These controls help organizations determine what an agent may access, which user or service it acts for, and how its activity can be observed.
Rules need a way to check model outputs
Sivasubramanian also described pairing generative models with separate systems that check outputs against company rules. Carnegie Mellon professor Zico Kolter argued that control systems must keep pace with AI capability and that doing so may require slower development. Together, these comments frame deployment as a governance and verification problem as well as a capability problem.
Should companies preserve the ability to switch AI providers?
Summit speakers disagreed about the trade-off between provider flexibility and deeper dependence on a particular AI lab. Anthropic’s Patil argued that designing to switch providers can lead companies to build around a least common denominator and divert engineering effort from differentiated work. Noeri co-CEO Carlos Guestrin argued that intelligence should not be controlled by one or two model companies and that companies should be able to build and own AI systems. Factory’s Eno Reyes described businesses that see no path forward without ceding control to one AI lab, while former GitHub CEO Thomas Dohmke emphasized developer choice.
These are competing strategic positions, not a single summit recommendation. A company weighing them has to consider whether provider-specific capabilities justify the dependence, and what it would cost to change providers or build more of its own system. The event report does not compare vendors or prescribe a universal choice.
Best Value
What do Madrona’s IA40 funding figures show—and not show?
Madrona reported that the 45 companies on its 2026 IA40 had raised $410 billion since founding. Anthropic, OpenAI, and Databricks accounted for $377 billion, or 92% of that cohort’s total. Madrona categorized the funding data as of August 15, 2026. These figures describe the companies on Madrona’s 2026 list, not all AI companies or the whole AI market. Madrona’s IA40 article is the source for the figures.
Madrona also said 23 of the 40 prior-year winners returned to the 2026 list, a 58% repeat-winner rate. The return rate is another measure of continuity within this particular list, rather than a measure of how many AI companies endure across the sector.
What should a company examine before choosing an agent platform?
The summit raised practical evaluation dimensions without providing comparative product testing or a vendor ranking. For a company assessing platforms, useful questions include:
- Data and context access: What information can the agent reach, and how are its actions and corrections recorded?
- Identity and security: Whose identity does the agent use, what permissions apply, and how are access boundaries enforced?
- Monitoring and auditability: Can the organization inspect what the agent did and identify failures?
- Interoperability: Can the system work with existing software, and what would changing providers require?
- Rule checking: Is there a separate mechanism to check outputs or actions against company rules?
- Demonstrated outcomes: Does a production deployment show completed work, time saved, revenue, or another result that matters to the business?
Madrona also reported that security and privacy ranked among the top three purchase criteria for 78% of surveyed enterprises. That figure comes from Madrona’s proprietary enterprise survey as summarized in its article; it is not a universal measure of buyer priorities.
What the summit left unresolved
The most consequential open issue is not simply whether agents can perform tasks. It is how companies will allocate customer access, responsibility, and rights to the records created when agents perform them. The summit reporting surfaced that uncertainty alongside concrete deployment concerns—organizational redesign, security, identity, monitoring, and verification—but did not resolve it. For now, organizations considering agents need to treat those questions as design and governance decisions to address, not assumptions the event settled.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
