Highlights

User lists

Previously you could only "ignore" individual users. Now there are full user lists: ignore lists, whitelists, and your own custom categories — each with a name and a color.

• Manage them on the new "User lists" page in settings.
• In any conversation, add the person to a list in one click; list badges show right on conversation cards.
• Conversations can be filtered by list.
• The bot can work with lists too: for example, following your instruction it will assign the person to the right category on its own.

Existing ignore settings are migrated automatically — nothing to reconfigure.

Voice calls — now for everyone

The voice channel no longer needs to be enabled on request: calls are available to every organization right away. Settings live where they did before — in the "Calls" section of your bot's settings.

Call campaign builder

A new call campaign builder: describe the task in plain language — "call clients with open tickets tomorrow between 10 and 12" — and get a ready calling schedule. The campaign runs as a routine, with results visible in the fire history.

The bot picks up new settings immediately

Previously, in long-lived conversations (for example in Telegram or Bitrix24) the bot could keep replying with the old model or old instruction after you changed its settings. Now changes to the model, instruction, and tools take effect immediately — in all active conversations.

New models

Refreshed the model catalog. Added:

• Anthropic — Claude Fable 5
• NVIDIA — Nemotron 3 Ultra
• Qwen — Qwen 3.7 Plus

Pick a model on your bot's settings page.

Also in this release

• Routines: when fired by a CRM event, the bot now always receives the text of your instruction — reactions are more accurate.
• Security: strengthened data isolation between organizations and hardened public connections — no impact on how your bots work.

Retired models

A few outdated models were removed from the catalog. If your bot was bound to one of them, pick a current one in settings:

• Gemini 2.0 Flash / Flash Lite → current Google models
• GPT-4 (0314 / 1106 Preview) → current OpenAI models
• GLM-4 32B / GLM-5V Turbo → current Z-AI models
• Euryale 70B (L3 / L3.1), Nemotron Nano 9B → current catalog models

No breaking changes

Existing bots and integrations keep working as before — no action required.