The customer service agent company now sells orchestration, and Salesforce just bought in for the first time.
Wonderful was valued at $2 billion in March. Today it is worth $5 billion. The Israeli-Dutch startup closed a $550 million Series C less than six months after its last raise, and it got there while moving away from the product that built its reputation.
Insight Partners led the round for the second time running. Index Ventures, IVP, Vine Ventures, 9Yards and Bessemer Venture Partners all wrote checks again. The new entry is Salesforce, investing in Wonderful for the first time. An incumbent software vendor putting money into an AI orchestration startup is not a neutral act. It is a read on where enterprise buying is heading. The last round, a $150 million Series B, closed in March. The gap between rounds is shorter than most companies take to hire a sales team.
Wonderful was founded in early 2025 and grew fast on a customer service AI agent platform, tailored for markets where English is not the working language. That is no longer the pitch. The company now sells Wonderful AI OS, a layer it says coordinates agents, workflows and AI applications against a company’s own data, context and existing integrations. It claims the platform runs on any model and sits on top of whatever stack a customer already has.
- Advertisement -
CTO and co-founder Roey Lalazar framed it as optional adoption: customers pick the pieces they want, pick the model per workload, and keep ownership of what they build.
The other half of the story is people. Wonderful sends engineers to its customers, sometimes to work on site, to embed the product into live systems. That approach has carried the company into more than 35 countries, and forward-deployed engineers have become one of the most contested hires in the AI industry. The $550 million goes to three places:
- Product speed: faster development across the AI OS platform.
- Engineer headcount: expanding the forward-deployed teams.
- Demand capacity: keeping up with existing customer pull.
The bet is that enterprises want one conductor for their agents, and that they will not build it themselves.
If you are picking AI vendors, the question has shifted. It is no longer which agent to buy, it is who owns the layer that connects them. Wonderful is betting that layer is a product, not internal plumbing. Salesforce agreeing with that is a signal worth pricing in. Also note the delivery model: vendors who send engineers to your office are winning deals that self-serve tools lose.
