Pervaya Forma adds MCP so an AI agent can configure BPM processes

Pervaya Forma has added MCP support and shown how a BPM system can become a working interface for AI agents. Part of the setup of categories, forms and processes can now be started not through a technical specification and a queue at an integrator, but through an ordinary natural-language dialogue.

The BPM bottleneck

Almost every BPM platform faces a long-known problem: the business knows which process it needs but cannot quickly build it in the system on its own. Even a simple task, such as a category for approving vacations, usually passes through a chain of the process owner, an administrator, a developer or an integrator. At each step, "I know what I need" gets translated into "this is how it must be configured in the interface and access rights", and with that translation come delays and lost meaning.

As a result, process configuration stops being a management task and becomes a mini-project with a request, a requirements description and a wait in a queue. If the business needs to change form fields, user permissions or routing logic quickly, it depends on the people who know the system's internal structure. A BPM system that is supposed to speed up a company's work ends up as a bottleneck itself.

The Pervaya Forma team decided to close this gap between intent and implementation through MCP.

How MCP was connected

MCP, or Model Context Protocol, lets an AI agent do more than answer in text: it can work with a specific system through described tools and context. In Pervaya Forma's case, the agent gets a controlled way to create and configure objects inside the BPM: categories, forms, processes and related parameters. The user describes the task in plain language, and the system turns it into actions that used to be done by hand through the interface or internal settings.

In practice, this changes how people interact with the platform. Instead of a formal specification, you can phrase the request the way the business understands it: which entity is needed, which fields must be mandatory, who has access and how approval works. If the integration is built carefully, the agent becomes not a "chat next to the system" but a full management layer.

According to the team, they have not yet seen ready-made MCP server implementations in Russian BPM systems, so the case looks less like a hype-driven experiment and more like an infrastructure shift.

What it delivers

The main effect is not that AI writes settings instead of a person, but that the distance between the idea of a process and its launch shrinks. The fewer intermediaries between the process owner and the working configuration, the faster a company tests hypotheses and makes changes. This matters especially for a BPM platform, because business processes rarely stay stable: they are regularly extended, simplified or rebuilt for new rules.

• creating categories from a text description
• configuring fields and forms without manual assembly
• assigning permissions and roles through natural language
• rolling out changes without a long cycle of specifications and queues

It also matters that this is not about generating pretty mockups or drafts, but about access to real actions inside a corporate system. That makes MCP far more practical: the value appears where an agent can safely change the working environment under clear rules.

If this model catches on, BPM vendors will face a new standard of expectations: users will want not only an interface for administrators but also a conversational management layer on top of it.

What it means

The Pervaya Forma story shows that AI agents are gradually moving from text assistants to an interface for corporate software. If MCP really becomes the integration standard, businesses will be able to configure internal processes noticeably faster, and developers' role will shift from manual configuration to oversight, rules and architecture.