Runs 24/7 on your server

AI Agent
of a new generation
for your business

Personal assistant in Telegram and Max. Any LLM — Claude, GPT, Gemini, GigaChat. Long-term memory, automation, knowledge base. Runs on your server.

Get in touch Learn more
8+
LLM supported
2
Messengers
24/7
Agent uptime
1
Your infra
Telegram — aiClaw
you> Show today's tasks and make a plan
🤖 Searching knowledge base...
✓ Found 7 tasks in CRM
📋 Here are today's priorities:
1. Client meeting — 14:00
2. Review commercial proposal
3. Sales report for May

you> /model gemini
✓ Switched to Gemini 2.0 Flash

you> Remind me in 2 hours about the meeting
✓ Reminder set for 14:00
_
Features

Everything you need — built in

aiClaw is not just a chatbot. It's a full AI agent with memory, tools and integrations.

Long-term memory

Two-level memory: SQLite for dialog history and pgvector for semantic fact search. The agent remembers everything important.

Hot LLM switching

Switch model mid-chat without restart. Claude, GPT-4o, Gemini, GigaChat and local models.

Task scheduler

Built-in cron scheduler. Reminders, recurring reports, automated digests — all configured by voice.

Knowledge base (Vault)

Structured wiki of your projects. The agent auto-logs decisions and updates the base. Always in context.

Integrations

Cloudflare, SWeb, Hestia, n8n, Perplexity, Playwright. Manage infrastructure directly from Telegram.

Image generation

Flux Dev, Kontext, Schnell, Ultra via fal.ai. Create visuals with one command right in the messenger.

Privacy & security

The agent runs only on your server. No external storage, no SaaS. Your data stays with you.

Extensibility

BotModule plugin architecture. Any custom commands and integrations added without touching the core.

Group chats

Full support for Telegram groups and topics. Flexible session isolation: shared, per-user, per-topic.

Architecture

How it works

Simple and reliable chain: from a messenger message to the result.

Messenger

Telegram or Max. Text, voice, documents, photos — the agent understands everything.

aiClaw Core

Memory, scheduler, tools. Context from the knowledge base loads automatically.

LLM of choice

Claude, GPT, Gemini, GigaChat or any local model. Switch when needed.

Result

Answer, action, reminder. The agent remembers decisions and learns from your context.

Compatibility

Any LLM.
Any messenger.

Don't lock in to one provider. Connect the latest models as they release. Full support for OpenAI-compatible API.

Messengers

Telegram
Max (VK Teams)
Ваш канал

Поддерживаемые модели

✦ Claude 3.x
◎ GPT-4o
◈ Gemini 2.0
⬡ GigaChat
◉ OpenRouter
▣ Ollama
◆ LiteLLM
⬢ Codex CLI
+ Любой OpenAI API
For teams

One agent.
The whole team.

aiClaw works in corporate Telegram and Max chats. Choose the mode for your scenario — from a shared duty assistant to a personal assistant for every employee.

Групповые чаты
Топики Telegram
Режим per-user
Изоляция сессий

The mode is set in config and can be changed without restart. Different rules can be set for different groups — e.g. passive mode in a work chat, personal in DMs with the manager.

shared

Shared chat

One dialog for the whole group

  • All participants' messages go into one shared session
  • The agent sees the full chat context and knows who wrote what
  • Good for duty assistant, digests, briefings
  • Shared memory — facts from conversations available to all
Ivan: what's the status of task #42?
Maria: and who's responsible?
🤖 Task #42 is with Peter,
deadline — June 3.
per_user

Personal

Separate dialog for each

  • Each group member gets their own isolated session
  • The agent remembers each employee's history and context separately
  • Privacy: others' messages don't get into your context
  • Ideal for teams where everyone has their own tasks and projects
Ivan: show my tasks
🤖 Ivan, you have 3 tasks:
1. Report · 2. Meeting · 3. Proposal
— Maria sees only hers —
mention_only

Passive

Responds only when addressed

  • Stays silent in the background — doesn't react to regular messages
  • Activates on @mention or a keyword
  • Doesn't interrupt team conversation, but always at hand
  • Optimal for active work chats
Ivan: let's meet at 15:00
Maria: ok, sounds good
Peter: @bot remind me at 14:45
🤖 Reminder set ✓
Pricing

From $2,500 — on your server

One-time setup. Then only your infrastructure costs — no subscriptions from us.

See full pricing
Open Source

We build tools.
We share with the community.

In building aiClaw we create useful tools and publish them open source. Two projects are already on GitHub.

Open projects
agent-runner
Agentic loop for any OpenAI-compatible LLM. 15 tools, MCP, memory.
deploy-sh/agent-runner
giga-cli
Fork of agent-runner for GigaChat (Sber). OAuth, compatibility adapters, free tier.
deploy-sh/agent-runner · ветка giga_cli
repio.dev
Record and replay macros. Browser action automation without code.
repio.dev
Contacts

Tell us about your tasks

We'll discuss how aiClaw can help your business. Answer any questions about integration and setup.

Email
[email protected]
Telegram
@mylandtg