Tray.ai unveils Merlin Agent Builder 2.0 for enterprise AI scale

Tray.ai has released Merlin Agent Builder 2.0, offering enterprise teams a platform that delivers AI agents capable of executing tasks beyond merely answering questions.

The new version of Merlin Agent Builder introduces features designed to address persistent challenges in AI agent deployment within enterprises. Industry data indicates that while a majority of enterprises are investing over $500,000 per year on AI agents, many face difficulties in scaling and deriving value from these solutions. Key obstacles highlighted include lack of complete data, session memory limitations, challenges with large language model (LLM) configuration, and rigid deployment options.

Addressing deployment and adoption challenges

Rich Waldron, Co-Founder and Chief Executive Officer at Tray.ai, said: "Enterprise teams aren't short on ambition when it comes to AI agents - but they are short on results. This release clears the path from prototype to production by removing the blockers that stall adoption. We've built the only platform where enterprises can go from idea to working agent - fast - without compromising trust, flexibility or scale. That's how agent-led transformation actually happens."

According to Tray.ai, a significant gap exists between building and actual usage of AI agents in workplace settings. Agents that are not integrated with comprehensive and up-to-date knowledge often lose context, make unreliable decisions, and force users to repeat information, which undermines user trust and can lead to underutilisation. Meanwhile, IT and AI teams find it difficult to align LLMs with appropriate use cases, particularly where multiple agents operate in parallel, and encounter added complexity when deploying agents across various platforms.

To address these issues, Tray.ai's upgraded solution includes advancements in four key areas: integration of smart data sources for rapid knowledge preparation, built-in memory for maintaining context across sessions, multi-LLM support, and streamlined omnichannel deployment.

Smart data sources and session memory

The Merlin Agent Builder 2.0 offers a new smart data sources feature aimed at simplifying the connection and synchronisation of both structured and unstructured enterprise knowledge. Through a single interface, users can link data from sources like file uploads or Google Drive. This data is then automatically prepared and vectorised to ensure agents are informed with relevant and reliable information.

Alistair Russell, Co-Founder and Chief Technology Officer of Tray.ai, commented: "Merlin Agent Builder isn't a services wrapper. It's a fundamental part of our product and built for ease of use and scale. It handles chunking and embedding at the source, ensuring each data source is optimally segmented and vectorized so agents are grounded in high-signal, relevant context. That means fewer retrieval failures, more reliable decisions, and agents that reason and take action. It's how teams move fast - without trade-offs."

Addressing another common shortcoming of AI agents - context loss between interactions - Merlin Agent Builder 2.0 incorporates built-in memory capabilities. The platform enables agents to recall previous sessions, track conversation history, and manage both short-term and long-term memory requirements automatically. This aims to reduce the need for custom solutions and enhances continuity in user exchanges, improving adoption rates.

Flexible large language model support

As organisations deploy multiple agents to handle diverse business processes, the ability to configure each agent with the most suitable LLM becomes increasingly important. Merlin Agent Builder 2.0 supports multiple LLM providers, including OpenAI, Gemini, Bedrock, and Azure. Teams can assign specific models to individual agents with tailored configurations, avoiding proprietary lock-in and supporting privacy-driven workflows where necessary.

Unified deployment across channels

The updated release allows teams to build an agent once and deploy it seamlessly across communication and application environments such as Slack, web applications, and APIs, or for autonomous operations. The delivery configuration is incorporated directly into the agent setup process, which eliminates the need for repeated setup and technical adjustments for different channels.

With these updates, Tray.ai targets what it identifies as critical needs for enterprises: simplified data onboarding, session-aware agents, flexible modelling, and consistent deployment experiences. The company states that by providing these features in a unified platform, both IT and business teams are better positioned to transition from pilot projects to production-ready AI agents that are actively used by employees and customers alike.