Five9 buys Level AI and Sierra buys Anon as customer-service AI consolidates

What happened

Five9 has agreed to acquire Level AI, a provider of conversational AI for customer service teams, with the deal expected to close in 2026. Level AI's platform handles more than one billion customer interactions a year, and the plan is to integrate it into the Five9 Genius CX Platform while continuing to offer it as a standalone product for enterprises on Five9 and other customer-service platforms. The financial terms were not disclosed. m:…1802

In the same week, Sierra acquired Anon, a company that built infrastructure for AI agents to navigate websites, manage credentials, and take actions in systems that have no API. Anon had raised more than $13 million and its product was used by hundreds of customers, including AI startups and publicly traded companies, with deployments in banking, healthcare, and enterprise software. Sierra says the technology will support its Horizon platform on longer-running goals such as loan origination and healthcare prior authorizations. m:…7746

Genesys, meanwhile, expanded its forward-deployed engineering teams, which work directly with customers on integration, governance, and operational requirements to move agentic customer-service use cases from pilot into production. The company points to Banco de Crédito del Perú, eir, and Riachuelo as customers that have already used these teams to bring use cases into production, and says its cloud platform powers more than 33 billion interactions a year for more than 7,500 customers. m:…3420

Why the buyers are moving the same way

The three moves share one diagnosis: the model is no longer the hard part. Level AI's pitch is cost and speed — the company says its Latitude models deliver up to 49x lower cost and 4x lower latency than frontier models — and Five9 gets that efficiency inside a platform that already serves more than 4,000 customers across more than 140 countries. The implication is that cheap, fast models only matter once they sit inside the system that already holds the customer relationship. m:…1802

Sierra's deal fills a different gap. Plenty of enterprise work still lives behind web portals with no API, and Anon's stack — authentication, credential handling, browser infrastructure, an agent harness, and a software kit — is built for exactly that terrain. The two companies have partnered since 2022 in Level AI's case and Sierra is buying the plumbing that lets its agents finish a task rather than stop at the edge of a website. This suggests the next enterprise-agent deals will be about reach into messy systems, not about a smarter single model. m:…7746 m:…1802

Genesys adds the third piece: people. Its forward-deployed teams exist because integration, governance, and operations are what stall deployments, and the company employs more than 2,000 customer-service professionals globally. Parloa's expansion of its Lens observability tool to cover every agent in a service operation, whatever stack it was built on, makes the same point from the measurement side — its 2026 study with Metric Sherpa found that 89.5% of operators cannot see at least one part of the customer experience. The pattern is that buyers now want proof the agent worked, not just a demo that it talked. m:…3420 m:…7794

What it means for founders

For a founder building in this space, the week's message is uncomfortable but clear. A point product that handles one slice of the contact center now has to explain why it will not be absorbed into a platform that already owns the customer, the workflow, and the measurement. The defensible ground is moving toward the parts platforms cannot easily buy: deep access to a specific system of record, a verifiable performance record, or a niche where the workflow is the product. m:…7746 m:…7794

The other lesson is about how deals get priced. Neither Five9 nor Sierra disclosed financial terms, and Anon's funding history — more than $13 million raised — is the only public anchor for its value. Founders should read that as a reminder that in a consolidation wave, the acquirer's distribution is doing much of the valuation work, and a product with hundreds of customers but no distribution of its own is worth more inside a platform than outside it. The question to ask about any customer-service AI startup now is not whether the agent works, but whose customers it reaches and who can see what it did. m:…7746 m:…1802