The startup Waniwani has just raised $8 million from Seedcamp, Redstone, Plug & Play and several business angels in order to develop what it presents as a revenue and compliance infrastructure for the agentic distribution of financial services. Behind this relatively modest funding round, however, lies a much larger ambition. The company is not looking to build a new conversational agent. It is betting on the emergence of a new distribution channel in which purchasing decisions are no longer made directly by consumers but increasingly influenced, prepared or executed by artificial intelligence systems.
Since the appearance of the commercial web, each major technological development has given birth to a new generation of intermediaries. Search engines have captured access to information. Marketplaces have captured access to products. Platforms captured access to audiences. AI agents could now seek to capture decision access.
For a long time, digital technology has been presented as a tool for disintermediation. The reality has often been different, the Internet has not eliminated intermediaries, but displaced them. Travel agencies have given way to Booking, classified ads to Leboncoin, traditional brokers to specialized comparators. In each case, value has gradually become concentrated in the hands of actors capable of organizing the meeting between supply and demand.
Financial services perfectly illustrate this evolution. Insurers, banks and credit companies have seen the emergence over the years of a multitude of digital intermediaries responsible for guiding consumer choices. Comparators, online brokers, aggregators and affiliate platforms have become essential players in distribution. Their role is based on a simple promise: to simplify a complex decision.
It is precisely in this area that AI agents are beginning to appear. When a user today asks ChatGPT or Claude to explain the differences between several insurance policies, to identify the best accounting software or to compare different credit offers, they are already delegating part of their decision-making process. Capacity remains imperfect and transactions are still largely executed elsewhere. But the economic logic is already visible. The more agents become able to understand user needs, compare available offers and support a transaction, the more they will occupy a central position in the value chain.
For twenty years, companies have learned to optimize their visibility for search engines. An increasing proportion of their marketing budgets have been devoted to SEO, advertising and traffic acquisition, except in an agentic environment, it is no longer just about being visible, but being recommended.
The consequence could be particularly important for traditional intermediaries. The historical value proposition of a comparator consists of aggregating information, comparing offers and guiding a choice. These are precisely the tasks that artificial intelligence models are becoming capable of performing at scale. If a conversational agent can analyze hundreds of insurance offers, filter the most relevant ones and directly present a personalized recommendation, the place occupied today by certain comparators could be weakened.
This evolution explains the emergence of a new category of companies of which Waniwani is one of the first representatives, with the objective of becoming the infrastructure that will allow companies to exist in these new distribution environments.
Concretely, this means making products accessible to agents, managing regulatory constraints, ensuring the traceability of recommendations, allocating revenue and measuring commercial performance. A function which in certain aspects recalls the role played by Salesforce in customer relationship management or Stripe in payments, but this time applied to distribution driven by artificial intelligence systems.
The choice of financial services as the first market is not by chance. Insurance, credit and real estate combine several particularly attractive characteristics: complex products, high commissions, significant regulation and relatively long decision cycles. Each recommendation can generate significant economic value. Each error can also have major regulatory consequences. These sectors therefore constitute an ideal laboratory for experimenting with the first agentic distribution models.
But the issue goes far beyond finance; the same mechanisms could gradually extend to B2B software, recruitment, legal services, home services and even health. Wherever a decision today requires a qualification, comparison or recommendation phase, AI agents can potentially insert themselves into the value chain.
The question then becomes less technological than economic. Who will control these new crossing points tomorrow? Laboratories like OpenAI or Anthropic? A new generation of specialized infrastructure? Historical comparators who will succeed in their transition? Or the suppliers themselves who will seek to maintain a direct relationship with their customers?
