boost.ai Version 14 Adds LLM Testing and Code-Free Integrations for CX Teams

Comparing models before they reach customers

The Norwegian company, which describes itself as an enterprise conversational AI platform for regulated industries, has extended the Test Studio in its platform with Large Language Model (LLM) testing. Teams can take any use case or guardrail in their AI Agent and run it against multiple AI models at the same time. The results show how each model performs, with quality, response speed and cost compared side by side.

boost.ai argues that public benchmarks only show how a model performs in general. They do not show how it handles a specific organisation's use cases with that organisation's guardrails in place. According to the company, the new testing lets CX teams see where simpler, more cost-effective models perform just as well as premium alternatives, and where a more capable model is genuinely needed, before either choice reaches a customer.

Integrations without a developer queue

The release also upgrades the AI assistant built into the platform's API Connector, now called the boost.ai API Connector Companion. Teams can describe the integration they need in plain language, from CRM systems such as Salesforce to automation platforms such as UiPath. They then point the companion at the relevant API documentation and receive working integration code, step by step.

The company says integrations are where transactional AI use cases have historically hit their limits. Checking a balance, processing a change or retrieving a policy detail all require a connection to a backend system, and building one has traditionally meant writing code and waiting for a development cycle. boost.ai says teams can now build and refine integrations themselves with no coding knowledge required.

A configurable chat panel

Version 14 includes a new chat panel and a redesigned configuration interface. The panel can run as a floating widget, a full-page experience or embedded directly in a page, and all configuration is handled in the platform's UI.

• Teams can keep multiple configurations and apply each to specific pages by URL.
• One AI agent can therefore behave differently across different parts of the customer journey.
• New additions include a landing page view with suggested topics and a fixed area for system messages and disclaimers.

These join existing features that clearly show end users when a response is AI-generated, which boost.ai says supports EU AI Act transparency requirements.

What the company says it means for CX teams

Jerry Haywood, CEO of boost.ai, said the question enterprises ask is no longer whether conversational AI can handle complex, transactional interactions, but "whether the teams responsible for those interactions can move at the speed the business needs."

Clay Cardozo, VP of Product Development, said the release targeted places where customers reported losing time: "Model selection was guesswork, integrations required code, and interface changes required front-end resources." According to Cardozo, a CX team can now handle all three itself.

The company says it has over 650 successful deployments, 600 live AI agents and more than 150 million automated conversations. Its customers include Nordea, Credit Union of Colorado, Sage, DNB and Trading 212, and it says it was named a Leader in Gartner's 2025 Magic Quadrant for Conversational AI Platforms. The press release does not give pricing or exact rollout dates beyond the coming weeks.