Hatz AI lets MSPs bring AI-as-a-service to enterprise customers

Startup Hatz AI has launched with the goal of bringing managed service providers (MSP) the tools to build and support AI applications they can quickly offer to SMB customers.

The Hatz AI service is anchored by a Large Language Model (LLM) Ops engine called Mido and includes multi-tenant management through an MSP admin dashboard. MSPs will be able to integrate AI into their services, developing focused AI applications and custom workflows managed by AI assistants.

AI as a service (AIaaS) is slowly emerging as an option offered by cloud services providers directly to large enterprises. But most current offerings aren’t geared for small or midsized businesses. And while Amazon and Google have the resources to offer AIaaS, the estimated 50,000 MSPs in the U.S. alone did not, according to Hatz AI cofounder and CEO Jimmy Hatzell.

“In order for small businesses to compete as AI comes up over the next couple of years, they’re going to need help, and they’re going to need tools and human help. So we are offering everything that would be included in an MSP AI-as-a-service offering to keep your business alive and make it through this next digital transformation,” Hatzell said. 

Hatz AI will launch its first products in March. The initial release will include an AI Application Builder and what it calls Organizationally Managed AI Assistants that MSPs can use to build custom applications for customers.

Hatzell describes the software offerings as a system of engagement and system of record. The system of engagement is an AI application that MSPs can customize for their end customers. The first application is an organizationally managed chat bot that is model agnostic, can be managed on a multi-tenant basis, and can be branded for an MSP.

MSPs also get an AI app builder, which lets them do prompt engineering and publish applications. Another product, which will be released in the summer, is an AI customer service agent that can be used to field and manage inbound phone calls.

The system of record part will come later this year and will allow MSPs to manage data sets and LLMs inside of Hatz’s platform and help train them and power the platform for the customers.

Hatz does not provide the hardware for processing models or chat bots. All of the processing is done either on the customer’s premises or on the MSP’s systems. “I think there will be a lot of companies that want their data in a private cloud, or in their own hosted data racks,” said Hatzell.