Agentic AI: investors are betting big on agent platforms

For more than twenty years, the digital transformation of companies has been structured around a relatively stable model: ERP for financial management, CRM for customer relations, analytical tools for data exploitation. This software architecture, gradually enriched by the cloud and SaaS, today constitutes the IT backbone of most organizations.

The emergence of artificial intelligence agents is stimulating a new generation of players who are trying to build platforms capable of orchestrating these agents at the heart of companies’ information systems. The promise is no longer just to improve existing software, but to enable autonomous agents to directly execute certain operational tasks.

From software to agents

The growing interest in AI agents is due to their ability to go beyond the traditional role of digital assistant. Unlike the first generative artificial intelligence applications, essentially designed to produce content or analyze data, agents are capable of interacting with the company’s computer systems and acting on these systems.

In a professional environment, these agents can respond to customer requests, analyze contractual documents, coordinate internal workflows or automate certain administrative operations. Their specificity lies in their ability to navigate between several applications, access internal databases and trigger actions in company systems.

This development is part of a broader dynamic around so-called agentic AI, in which artificial intelligence becomes an operational layer capable of executing processes.

The challenge of scaling up

Despite the growing interest in these technologies, the transition from experimentation to large-scale deployment remains one of the main challenges for companies. Prototypes are multiplying, but their integration into often complex or fragmented IT infrastructures slows their diffusion.

Agent platforms seek to respond precisely to this difficulty. Their role is to provide an architecture capable of orchestrating agents, monitoring their performance and ensuring their reliability in demanding operational environments.

These new platforms become a kind of agent operating systemresponsible for orchestrating their operation in the company’s daily processes.

A horizontal architecture

Agent platforms are also distinguished by their horizontal architecture. Rather than offering a specialized tool for a single profession, they seek to build a common infrastructure capable of accommodating different types of agents and workflows.

Once this architecture has been deployed in an organization, new use cases can be gradually activated: automation of customer support, documentary analysis, coordination of internal operations or even management of certain administrative processes.

This cumulative logic constitutes one of the arguments put forward by startups in the sector. In some pilot deployments, agents helped reduce processing times for certain operations by almost 60%with automation rates exceeding 80% in certain interactions.

A global race for agent platforms

The agent platform market is now attracting considerable funding, particularly in the United States, where several startups are trying to build the infrastructure that could structure the next generation of enterprise software.

The startup Cognition AI, for example, made itself known with Devin, an agent presented as capable of carrying out complete software development tasks. The company recently raised 400 million dollarsreaching a valuation higher than 10 billion dollars.

In a similar vein, Adept AI Labs is developing agents capable of interacting directly with the interfaces of existing software. Founded by former researchers from Google and OpenAI, the company has raised more than $415 million to automate workflows in professional applications.

Another closely watched player, Sierra AI, founded in 2023 by Bret Taylor, former manager of Salesforce, develops agents specialized in customer relations. The company has raised more than $685 millionfor an estimated valuation around 10 billion dollars.

Alongside these highly capitalized companies, other startups are exploring more specific uses of autonomous agents. The company TinyFish, created in 2024, develops, for example, agents capable of carrying out complex tasks on the internet, such as automated data collection or monitoring of online markets. The startup has already raised more than $40 million.

This week, Wonderfulfounded by Bar Winkler And Roey Lalazarannounces a lifting of 129.8 million euros in series Bvaluing the startup approximately 1.7 billion euros. The funding round is led by Insight Partners with the participation of existing investors, including Index Ventures, IVP, Bessemer Venture Partners and Vine Ventures. This operation comes after two raisings carried out in 2025: a round of 29.4 million euros in seedfollowed a few months later by a Series A of 86.5 million euros led by Index Ventures.

Since its release less than a year ago, Wonderful has been deployed in more than thirty countries in Europe, the Middle East, Asia-Pacific and Latin America. The company plans to use this funding to continue the development of its agent platform and accelerate its international expansion, with the aim of increasing its headcount by approximately 350 to nearly 900 employees by the end of the year.

The multiplication of these investments illustrates the speed at which this segment is structuring itself. Startups specializing in AI agents have raised several billion dollars in recent years, even though the market is still emerging, money that is no longer flowing to classic SaaS startups, many of which are running out of steam due to lack of financing.