Artificial intelligence (AI) is shaping a new model of the digital economy in which AI agents independently choose products, connect to services, and make purchases on users’ behalf. As a result, companies that used to restrict automated systems’ access to their websites will now have to compete for their attention.
Analysts at venture capital firm Andreessen Horowitz (a16z) released a study on the emergence of the Business-to-Agent (B2A) economic model, in which businesses interact with agentic systems. In their view, AI agents are gradually becoming a new category of customers that can compare offers, choose providers, and pay for services on their own.
HUMAN Security data cited in the report show that AI agents’ browser traffic rose 8-fold from August to October 2026. Meta products accounted for about 60% of that activity and OpenAI tools for around 30%. U.S. shoppers turn to AI before visiting a website to make a purchase in about 11% of cases.
In the first quarter of 2026, traffic to Shopify stores from AI-related sources rose 8-fold year over year. The number of orders placed after an AI-assisted search grew nearly 13-fold.
Autonomous systems are gradually changing how goods and services are bought. Most AI agents currently interact with websites through a browser, but new tools let them connect directly to third-party platforms, sign up for subscriptions, and initiate payments.
As an example, the a16z researchers cited an experiment in which ChatGPT independently coordinated Krea, Descript, and ElevenLabs to produce a promotional video. While completing the task, the system suggested signing up for a paid Descript subscription. Scenarios like this show how AI agents can move from making recommendations to taking a direct part in commercial transactions.
For more on how agentic commerce is taking shape, see CP Media’s special report.
Software development already shows the potential scale of the change. By September 2026, more than 40% of pull requests on GitHub started as drafts created by AI agents, up from 17% 6 months earlier. Other tech companies also report growing automation:
- Supabase says more than 60% of new databases are created with AI tools.
- On Neon, automated agents deploy 80% of new databases.
- AI agents account for nearly 40% of visits to Stripe’s documentation and 70% of command-line requests to its API resources.
- Deployments initiated by AI agents on Vercel rose 28-fold over the year.
The a16z analysts expect similar changes in other industries. One of the most promising areas will be specialized infrastructure that lets AI agents interact with digital services and carry out transactions on users’ behalf.
The market map a16z presented singles out infrastructure for agentic financial systems as a separate segment that includes several categories of solutions:
- delegated payment tools that let AI agents initiate payments on users’ behalf
- account and spend management systems that control agents’ permissions and the financial transactions they’re allowed to make
- identity and trust solutions that verify the parties involved and reduce fraud risk
- checkout technologies that complete orders and process payments
Companies in this market mentioned in the study include PayOS, Crossmint, Prava, Skyfire, Sardine, Prove, and Rye, among others.
Fraud and chargeback risks still hold back widespread adoption of agentic finance. Many large companies continue to block automated traffic for fear of unauthorized transactions.
Andreessen Horowitz believes the rise of the B2A economy could be a technology shift comparable to the spread of the internet and mobile apps. As AI agents develop, competition will shift from winning users’ attention to being chosen by autonomous systems that can find the best offers and make purchases on their own. In that environment, the ability to interact with AI agents safely and conveniently will become an important competitive advantage for businesses.
A few days earlier, Deutsche Bank and Ant Digital Technologies presented a joint study on using AI agents in corporate payments and financial operations.
