Everything you need, nothing you don't.
alChatBot is opinionated where it should be (channels, agents, playbooks) and endlessly configurable where you need it (skills, tools, roles).
Every agent, exactly the shape you need
Pick a model, wire up tools, tune the memory. Test in a sandbox before shipping.
Multi-provider
OpenAI (GPT-5, o-series), Anthropic Claude Opus/Sonnet, Google Gemini 2.5, Z.ai GLM 4.6 — swap providers per agent without rebuilding prompts.
Skills as checkboxes
Toggle memory, CRM, user fields, HTTP tools, live-operator handoff, WhatsApp polls, Google Calendar. Each skill drops in its own tools + guidance.
Voice notes → text
Whisper transcription runs on incoming audio DMs. Agent answers what the customer actually said, not a placeholder.
Vision on media
Send an agent product photos, screenshots, or menu shots — vision-enabled models read them and reply in context.
Visual playbooks that actually branch
Trigger. Filter. Fetch. Reply. Route. Build the whole customer journey on a live canvas.
Rich triggers
WhatsApp keyword, Instagram DM keyword, IG post/comment, click-to-message ad, new contact — each with keyword modes (contains, exact, starts, regex).
Per-button branches
Send a DM with quick-reply buttons or a WhatsApp poll — each option becomes its own output handle on the node. The flow resumes down whichever branch the customer tapped.
HTTP request node
Call any API mid-flow. Response goes into a variable you reference downstream as {{apiResponse.data.name}} — with JSON body editor, headers, query params.
AI-agent handoff
Insert an 'AI Agent Reply' node to swap the responding agent mid-flow. Great for routing 'sales' vs 'support' intents into specialised agents.
One codebase. Every messaging surface.
Same agent, same memory, same tools — no matter where the customer wrote you from.
WhatsApp Business
Multi-number, media handling, native polls with vote decryption, group chats, contact profiles + LID/PN mapping done right.
Instagram DMs + comments
Auto-reply on comments, comment-to-DM bypass of the 24-hour window, quick-reply buttons, image + audio + document media.
Facebook Ads (click-to-message)
Route each ad to the agent trained for that campaign. Track lead → agent → outcome end-to-end in one thread.
One shared inbox for the whole team
Real-time updates, custom fields, tags, statuses — all inline with the conversation.
Realtime updates
Socket.IO events push new messages, delivery ticks, poll votes, and comment replies to the open chat without a refresh.
Custom fields
Define fields (age, city, plan, purpose…) on the Contacts page. Agents read + write them via a first-class tool; humans see them in the CRM panel.
AI-assisted CRM
Agents auto-upsert clients on first contact, tag by intent, and update the status as conversations progress. Sales team gets pre-qualified rows.
Bring your workspace with you
Third-party services your agents can use as tools. Connect once per workspace.
Google Calendar
Agent checks availability with listCalendarEvents, books slots with createCalendarEvent, cancels with cancelCalendarEvent. Google-native invites to attendees.
Custom HTTP tools
Give an agent a Bitrix, Notion, or in-house API tool via reusable HTTP templates — parameter schemas, headers, dynamic values from the conversation.
Data tables
Upload price lists, FAQs, product catalogues as tables. Agents search across turns and answer with the right row inline.
The metrics that pay rent
Response time, resolution rate, ad ROI, agent quality — filtered every way you'd want.
Response time
Median and p95 first-reply time, broken down by channel, agent, and time-of-day. See where the night shift needs help.
Agent quality
Oversight agents review a sample of every agent's replies and flag issues you'd have missed. Learn from your own conversations.
Ad attribution
See which Meta Ad campaign brought each lead, what the agent said, and whether it converted. Close the reporting loop.
Built for real businesses
Workspaces, roles, audit trails, and data handling designed for teams that aren't playing.
Workspaces + roles
Invite teammates, assign owner-defined roles with granular permissions (inbox view/write, agent edit, CRM access…). Every mutation gates through the role matrix.
Live-operator escalation
Agent doesn't know the answer? It pings a real human over WhatsApp, waits for the reply, and forwards it to the customer — with full audit trail.
Your data, your keys
Bring your own AI API keys on higher plans — usage bills your Anthropic/OpenAI account directly. On-prem deployment available for Scale customers.
See it in your workspace.
Try every feature free for 7 days. Connect a channel, build an agent, ship a flow — all before your credit card runs out (there isn't one).