By Benjamin Guggenheim

Not a subscriber? Sign up here to get this newsletter in your inbox.

• In order for agentic commerce to truly take off, sellers of agentic products need to overcome a trust deficit with consumers and convince skeptical merchants to provide access to their online marketplaces, according to a new WP Intelligence report from Lead AI & Tech Analyst Dan Gallagher.
• Meta and OpenAI’s recent releases of agentic products bring new, wide-ranging capabilities.
• Existing regulation around complex transactions and competitive dynamics in the tech industry have blocked the diffusion of these new agentic products.

This is Benjamin Guggenheim. Welcome to WP Intelligence’s AI & Tech Brief, where we examine the transformative technology of artificial intelligence at the intersection of innovation, policy and power. Get in touch at Benjamin.Guggenheim@washpost.com or on Signal at ben_guggenheim.88.

Programming note: The AI & Tech Brief will be running only two editions this week so I can catch up on some reporting and help lead conversations around the future of AI and finance for DC Fintech Week. Keep your eyes out for another edition this Thursday.

On Wednesday, I’ll be moderating a conversation on agentic money with Tom Mutton of the Bank of England, Nima Elmi of Circle Internet Group, Nilesh Dusane of Amazon Web Services and Jack McDonald of Ripple. (Amazon co-founder Jeff Bezos owns The Washington Post.) You can register here.

In addition, please join us for a reception Tuesday evening with autonomous drone company HavocAI. WP Intelligence Editorial Director Luiza Savage will be discussing the future of all-domain drone warfare with the company’s CEO, Paul Lwin. You can register here.

The Lead Brief

It’s DC Fintech Week, so WP Intelligence is taking a deep dive into how the financial services industry is adopting AI. According to a new report by WP Intelligence Lead AI & Tech Analyst Dan Gallagher, there are plenty of reasons to anticipate roadblocks. And consumers will be slow to take up agentic AI for purchases and financial services.

1. The first reason is that the financial services industry is heavily regulated and traditionally has been quite slow to take up new technologies.

Dan makes the point that, decades into the existence of digital payments, a surprising number of businesses — just over one-quarter — still made business-to-business payments by paper check in 2025, according to an annual survey by the Association for Financial Professionals.

Many Americans still make mortgage payments by paper check, and Bernstein Research estimates that, overall, cash and check accounted for over one-quarter of consumer-to-business payments in 2025.

2. AI needs to overcome a trust deficit.

According to an EY AI Sentiment Study, only 14 percent of consumers are willing to give agentic AI permission to make investments, compared to 36 percent that would allow AI to claim discounts online for them.

“Trust, data access, and regulations are the biggest obstacles to agentic adoption in financial services, in our view,” said payments industry analyst Harshita Rawat of Bernstein Research.

3. “Agentic commerce” needs to have buy-in from both the consumer and the merchant — and early products suggest that sellers want to maintain control of their platforms.

Agents could take on complex financial tasks such as managing consumers’ retirement portfolios or shopping for and comparing retirement policies —which could replace some of the intermediary steps off of which the financial services industry makes money.

In a sign that more AI agents are indeed roaming the internet, Adobe found in a survey that AI-driven traffic had surged 393 percent in the first quarter from a year earlier.

But there’s evidence to suggest that merchants want to keep control of their relationship with the consumer: For instance, OpenAI killed its Instant Checkout service in March, with the company’s explanation suggesting that it was partly due to limited merchant adoption. (The Washington Post has a content partnership with OpenAI.)

A report from Forrester Research found that there’s “lukewarm” interest from consumers in having agents make purchases for them — though an AI assistant launched by a start-up called Instinct has garnered some success in Silicon Valley with early tech adopters.

Agentic products

The launch of Meta’s Muse agent and OpenAI’s agentic AI tools, called “dots,” bring new, wide-ranging capabilities.

As OpenAI detailed at its DevDay 2026 two weeks ago, the company’s dot agent has access to its own computer and browser in the cloud — and users can access dots on their phones and computers to assign them tasks. Dots will then perform the tasks even when consumers are offline.

“Dots are powered by Astra, our most aligned models, and they have built-in protections and safeguards,” said OpenAI CEO Sam Altman during the DevDay keynote. “You can set boundaries on app and computer use, and you can customize instructions for any actions your dot takes.”

Separately, Meta launched its Muse agent on Sept. 8, which is similar architecturally to OpenAI’s agentic product but is targeted toward everyday consumers rather than enterprise businesses. (Users can access Muse through WhatsApp, whereas users get access to dots through Slack and Teams.)

Mobile downloads of the app have so far totaled around 7 million since the launch, according to data from Sensor Tower.

While that number is relatively small, Meta has a significant advantage over other agentic AI offerings in terms of the scale of the social platforms such as Instagram that it’s able to plug into, Dan argues.

Indeed, the market reacted positively to Muse’s launch, with Meta jumping 17 percent in the month afterward. In a sign that investors potentially took the launch as a bad omen for the financial industries, the S&P Americas BMI Financial Services Index dropped more than 6 percent over the same period.

“In the coming years, I expect that Muse is going to grow into the personal superintelligence that billions of people around the world are going to use to accomplish their goals and improve their lives,” Meta CEO Mark Zuckerberg said at a Meta Connect event last month, promising that “Muse will make you money.”

The Strategic View

Jefferies software analyst Brent Thill says “many chief information officers believe Meta can build a durable enterprise AI business.”

Indeed, Meta is making massive bets — driving its free cash flow into the negative — that its agentic products will be a hit.

But while Walmart, Best Buy, Gap, Sephora and other major U.S. businesses have agreed to connect with Meta, Amazon, the world’s largest retailer, has blocked Muse from accessing its site, Dan notes.

Amazon said that it needs to “ensure a safe, secure and reliable customer experience” — but the company, of course, also offers competitive AI agents.

Similarly, there are reasons to believe that other kinds of entities will balk at connecting to agents such as Muse: Banks and credit card companies typically rely on identity and authorization, while complex transactions frequently require licensing (think about how insurance agents need to be licensed in their respective states).

In another example of how major incumbents might feel about connecting to other companies’ agentic AI, JPMorgan Chase began charging financial data aggregators such as Plaid — citing significant costs incurred when the aggregators access the bank’s networks.

The outlook: Businesses will need to get ready for agentic commerce and develop rules around liability and authorization for when agents purchase items from online marketplaces.

But universal adoption of agentic commerce may be a long way off given competitive dynamics in the tech industry and existing regulations around complex transactions.

This newsletter is published by WP Intelligence, The Washington Post’s subscription service for professionals that provides business, policy and thought leaders with actionable insights. WP Intelligence operates independently from The Washington Post newsroom. Learn more about WP Intelligence.