Vibe coding in Bitrix24: how AI builds a client passport

Vibe coding in Bitrix24 let the platform's teams build a "client passport" with AI agents: a summary card for a company that shows contacts, deals, leads and all activities on a single screen of the business portal.

What Vibecode is and why it exists

Vibecode is a development platform inside the Bitrix24 ecosystem, built on the vibe coding principle: you describe a task in natural language, an AI agent generates the interface and logic, and you check the result and refine it through iterations. No hand-written code, and no diving into the API from scratch.

The platform's main goal is to let business analysts and product managers customize the portal on their own, without involving a developer. That is exactly what the authors set out to test on a concrete, practical case.

The client passport: what they built and how

In a typical CRM, company data is scattered across sections: contacts, deals, and call and email history live on different screens. Before every call or meeting, a manager has to piece the picture together by hand.

The "client passport" solves this problem: one page where everything is visible at once. The authors put the following into the card:

• company contacts with job titles and communication channels
• open and closed deals with amounts and statuses
• leads with sources and responsible managers
• an activity timeline: calls, tasks, meetings, emails

They built it iteratively: they described the next block to the agent, looked at the result and refined it. Some tasks, such as the basic display of a contact list, the platform handled without trouble. Connecting all the blocks to live CRM data turned out to be harder.

Where the AI made mistakes and how they were handled

There were errors. The agent periodically built requests to the Bitrix24 API with wrong parameters: fields were named differently from the real structure, or the selection logic did not work as expected. In such cases the authors clarified the task in the dialogue and restarted the iteration.

Some problems were fixed by rephrasing the request to the agent. Others required direct manual edits, either when the agent kept repeating the same error or when the request structure had to be written out precisely by hand.

"We tell our experience step by step: what was easy, how we organized

the work, where the AI made mistakes and whether we had to fix anything manually," the authors write.

This is a standard vibe coding pattern: the AI takes over most of the routine, but precise integrations with real data and final debugging still remain with a human.

Why this matters for business

Vibecode is a telling example of how enterprise platforms build AI development directly into their ecosystems. Previously, any Bitrix24 customization required a developer who knew the API. Now a business analyst can describe a task in Russian and get a working prototype within a few hours.

The main current limitation is the same as for most AI development tools: the more complex the integration with real data and the more specific the business logic, the more often the agent misses and the more manual finishing is needed. Even with that caveat, the barrier to entry for customization drops several times over.

What this means

The client passport case shows both sides of AI development in enterprise systems: real speedups on routine tasks, and a continuing dependence on manual oversight where the data is complex. The more candid write-ups like this appear, the better businesses can judge where vibe coding is justified and where it is better not to skimp on a developer.

What is vibe coding in Bitrix24?

Vibe coding here is a development platform inside the Bitrix24 ecosystem, built on the vibe coding principle: you describe a task in natural language, an AI agent generates the interface and logic, and you check the result and refine it through iterations.