The Pulse
Fulcrum Ships MCP Tools for AI-Built Field Workflows
Fulcrum has launched an MCP server and AI Toolkit that let field teams create and manage workflows through natural-language instructions. The early-release tools target utilities, engineering firms, construction companies, and inspection te

AI.info Team ·
Fulcrum launched an MCP server and AI Toolkit on September 15, giving field organizations a way to create and manage digital workflows through natural-language instructions rather than waiting for a developer to build each application.
The tools are available through Fulcrum Labs, the company’s early-release program. Fulcrum says customers can describe a process and produce an end-to-end workflow that runs on the devices their crews already use, while inheriting the platform’s permissions, data structures, offline capabilities, integrations, and audit trails. The company says nearly 3,000 organizations use its platform.
“Since Fulcrum already provides the core services the field needs, it’s the natural place to apply the incredible power of AI to build these workflows,” said Mike Lambert, chief product officer at Fulcrum, in the company’s announcement.
Fulcrum targets the workflows companies leave unfinished
Field organizations often operate with a mixture of paper forms, narrow-purpose mobile applications, spreadsheets, and procedures that live in employees’ heads. Fulcrum is positioning the MCP server and AI Toolkit at the unfinished work that does not fit neatly into a standard form but still needs a repeatable digital process.
In a separate product post, Fulcrum gives examples including confined-space entry permits, soil boring logs that represent subsurface layers, and weld maps tracking testing and coating status along a pipeline. Those workflows can involve conditional logic, drawings, safety gates, strict sequences, and custom reports.
The company says simple forms can be generated in minutes, while more involved workflows that previously took weeks may be completed in hours or days. Fulcrum does not claim that every process can be built without technical help; workflows centered on drawings, safety controls, or strict operational sequences may still require a technical review.
One connector supplies the actions, another layer supplies the guidance
Fulcrum describes the two products as complementary. The MCP server connects compatible AI clients to Fulcrum so they can create and update applications, manage shared choice lists and webhooks, generate app extensions and report templates, and query live records. The AI Toolkit supplies platform-specific guidance about field types, workflow structure, automation patterns, and safety checks.
Fulcrum’s Labs documentation lists 56 MCP tools across 17 categories. Fifty-three tools handle application building and management, while three support read-only queries against live data. The AI Toolkit includes 16 skills covering activities such as workflow discovery, field selection, safe automation, gap checking, and data queries.
The products can connect with AI clients including Claude, ChatGPT, Copilot, Cursor, Codex, Gemini, and other MCP-compatible systems. Fulcrum says the toolkit is a library of skills rather than a server, while MCP provides the live connection to the customer’s Fulcrum organization.
Security and offline operation stay inside Fulcrum
Fulcrum’s pitch depends on keeping AI-generated workflows inside the same operating model as applications built through conventional methods. A workflow created through MCP uses Fulcrum’s existing security model, permissions, data structure, and administrative controls rather than creating a separate application that an IT team must secure and maintain.
That distinction matters for field work, where a workflow may need to function without a reliable connection and may produce records used for compliance, asset management, safety reviews, or emergency response. Fulcrum also points to platform features such as GIS data, offline synchronization, lidar integration, analytics, and audit trails as services that the AI builder does not need to recreate.
“Our customers demand infrastructure they can trust, but they’re also under a mandate to modernize,” said Jim Grady, Fulcrum’s chief executive officer. “That drove us to make our Field Operations Management platform the place where the most significant drivers of trust are built in, like permissioning, integration, and audit trails.”
Early access comes with limits
Fulcrum Labs is an early release, not a general-availability product. The company’s documentation warns that the tools are still in progress, may contain rough edges, and do not yet offer API stability guarantees. Fulcrum recommends testing them in a non-production environment before wider deployment.
Setup also requires an API token, and the current connection uses bearer-token authentication rather than OAuth sign-in. Fulcrum’s documentation says customers must use the endpoint for their organization’s regional instance. The guidance is available for several AI clients, but the company says Claude Desktop and Cowork cannot yet install the Toolkit directly.
Customers can request access through Fulcrum Labs. The company says early users will shape the next iteration of the connector and toolkit as they build workflows that previously remained on a backlog.
Fraser-Lever tests the model with engineering work
Ben Allwood, director at Fraser-Lever, a multidisciplinary engineering company working primarily in the resources sector, said Fulcrum’s extensibility and offline operation provide a base for AI-assisted workflows.
“Its flexible data architecture and powerful API provide a strong foundation for innovation, enabling us to integrate AI-assisted workflows that streamline processing, improve accuracy, accelerate turnaround times, and ultimately deliver greater value for our clients,” Allwood said in Fulcrum’s announcement.
The product’s immediate proposition is narrower than fully autonomous field operations: let the people who understand a process describe it, then let AI assemble the first working version inside an existing field platform. Fulcrum’s release places that capability behind an early-access program, with 56 tools and 16 guidance skills available for customers willing to test the unfinished system.