ELMA BPM and ELMA365 in 2026: what the platform can do, how the market is structured, and when to migrate from 3.x

ELMA BPM is the historical name of the ELMA platform; the current product is ELMA365, a low-code BPM platform: according to the vendor, it has 4,000+ customer companies, has been in the CIS software registry since 2016, and is No. 1 among BPM systems in CIS according to the TAdviser ranking (elma365.com). In 2026, the platform gained ELMA Cortex - AI agents embedded in processes - and a new release cycle: 4 major releases per year for On-Premises and 1 LTS.

The CIS BPMS market at the end of 2025 was 33-34 billion RUB (Skolkovo Foundation and TAdviser study); the most mature trio is BPMSoft, ELMA365, and Directum RX. In the article: what ELMA365 can do today, the limits of standard 1C integration, what costs more - licenses or integrations, and when to migrate from ELMA BPM 3.x. Watch on YouTube Watch on Rutube

The licensed built-in 1C Connector synchronizes 1C reference data and documents on a schedule or triggered from processes; 1C reference data appears in ELMA365 as native applications (elma365.com documentation). In May 2025, ELMA and partner Onellect released Connector 2.0 - bidirectional ELMA365 ↔ 1C exchange without code, including document cards and transaction registers. For document management, approvals, and reference synchronization, this is enough.

The connector's limit is end-to-end event-driven processes and master data: when a process must react to events in 1C, WMS, and EDR simultaneously, and reference data must have a single source of truth across 3+ systems, an integration layer (bus) is needed. Without it, ELMA365 risks becoming another silo with point-to-point exchanges. How such a layer is built - ESB integrations; what the project includes and how much it costs - ELMA365 implementation and 1C integration.

The winter 2026 ELMA365 release introduced ELMA Cortex, a platform of enterprise AI agents embedded in processes; at the same time, scenario development, document management, and 1C integration in KEDO were strengthened (elma365.com). In June 2026, Cortex gained sensitive data masking for LLM work, a Guardrails builder, isolated projects, and improved support for Yandex and Sber models (CNews). The practical takeaway: an agent is only as useful as the data landscape connected to it.

An AI agent in procurement approval works only if it can see up-to-date 1C reference data, warehouse stock, and document statuses - so Cortex depends on the same integration layer as ordinary processes. How the BPM layer connects people, robots, and agents - in the overview BPM in an AI-native company.