SnapLogic has introduced AI Gateway and Trusted Agent Identity to its Agentic Integration Platform, targeting organisations moving artificial intelligence from pilot projects into operational use.
The products are part of broader updates across AI connectivity, agent orchestration and governance. SnapLogic is also expanding tools that let businesses connect AI agents to existing applications, data sources and APIs.
The launch comes as many companies struggle to turn early AI trials into working systems embedded in core business processes. SnapLogic cited Gartner research showing that nearly 50% of generative AI projects were abandoned after proof of concept due to poor data quality, weak risk controls, rising costs, or uncertain business value.
At the centre of the update is AI Gateway, a central layer for authentication, authorisation and traffic throttling across AI interactions. SnapLogic has also added an observability dashboard to monitor how agents interact with enterprise systems.
Trusted Agent Identity is designed to ensure an AI agent operates with the identity and permissions of the individual user, rather than a shared service account. Under this token propagation model, user identity passes from the agent through the integration layer into backend systems, making each action traceable to the person who initiated it.
Platform Changes
SnapLogic is also extending its platform so existing enterprise integrations can be used as tools by AI agents, including agents built outside its own environment. An OpenAPI Function Generator converts API specifications into agent-callable functions, while bi-directional support for the Model Context Protocol allows organisations to connect to external MCP servers and expose SnapLogic pipelines as secure tools for outside agents.
These changes add more than 1,000 prebuilt connectors for systems such as ERP, CRM, databases, and software-as-a-service products to the pool of tools available to agents. SnapLogic has also added tool lifecycle management, versioning, metadata and lineage controls to support reuse and oversight.
For orchestration, the company is expanding its AgentCreator product to enable teams to design and deploy custom AI agents visually. Users can inspect reasoning steps, tool calls, and results during design, while workflows can coordinate activity across multiple agents and enterprise systems simultaneously.
The platform supports leading AI model providers, including OpenAI, Azure OpenAI, Google Gemini and Amazon Bedrock. According to SnapLogic, support spans eight capability dimensions, providing a consistent experience without requiring separate logic for each model.
Execution Focus
Jeremiah Stone, Chief Technology Officer at SnapLogic, said the market's main problem is not a lack of AI models. "Enterprises don't have a shortage of AI models or agents. They have a shortage of execution," he said.
Stone said the challenge is embedding AI in business operations rather than keeping it in test environments. "The real challenge is operationalising AI as digital labour inside business processes. With this expansion of our platform, we're enabling organisations to turn AI into real work. The companies that win won't be the ones with the best models. They'll be the ones that harness them first."
SnapLogic has also updated SnapGPT, its in-platform AI assistant for building and managing workflows. The additions include Plan Mode, intended to validate and refine workflow plans before execution, and Preview Data Insights, which highlights data structure, quality and compliance issues during design.
The company counts AstraZeneca, Adobe, Verizon and Sony among its customers. The latest release reflects a broader shift in the enterprise software market as suppliers try to solve a recurring problem in corporate AI adoption: how to govern, connect and audit agent-based systems once they move beyond experimentation.
Other governance features in the update include end-to-end observability for AI workflows, full pipeline lineage and traceability, and controls spanning data, APIs and agent interactions.