Build Once, Deploy On
Any Platform, Any Channel
A customer calls Monday about a refund. Chats on your website Wednesday for an update. Texts Friday to check status. Your agent remembers every conversation, takes action on every channel, and never asks them to repeat themselves.
One conversation,
no matter how they reach you.
Customers Don’t Think in Channels
They call, then chat, then text. To them it’s one conversation. But behind the scenes, you’re running separate agents that don’t talk to each other.
“Please Describe Your Issue Again”
Nothing kills trust faster. When your voice agent forgets what the chat agent learned yesterday, customers feel it.
Prompts Aren’t Enough
Your agent needs to look up orders, check CRM records, process refunds, and search your docs. On every channel. That takes real infrastructure.
One agent across voice, chat, and SMS
Configure your agent once. Choose a model, set up voice and transcription, assign tools, and connect a knowledge base. Chanl automatically translates that config for each platform and channel. Deploy to VAPI for voice, embed a widget for chat, connect Twilio for SMS. Same agent, same memory, every channel.
- Configure model, voice, tools, and memory in one place
- Chanl translates your config for each platform automatically
- Test across VAPI, ElevenLabs, OpenAI Realtime, Pipecat, and LiveKit
- Add new channels without reconfiguring your agent
Bring your agents, then make them better
Import agents you’ve already built on VAPI, ElevenLabs, or Bland with one API call. Chanl pulls in the config and layers on tools, memory, and knowledge. Expand to new channels or test on a different framework without disrupting what’s live.
- One API call imports your agent with config preserved
- Layer on tools, memory, and knowledge your current platform doesn’t offer
- Compare frameworks: test the same agent on VAPI vs Pipecat vs LiveKit
- Expand to web chat, SMS, or messaging from the same agent
VAPI
SyncedVoice agent synced with full tool and memory access
ElevenLabs
SyncedVoice synthesis with knowledge base integration
Bland
PendingOutbound calling with CRM tool connections
Test with real tools and memory before going live
Quick-test by chatting with your agent in the dashboard using real tools and memory. Then run full scenario-based testing. Throw 25+ AI personas at your agent, score conversations with automated scorecards, and catch failures systematically before deploying to any channel.
- Chat with your agent in the dashboard using production tools and memory
- Run AI-powered test scenarios with diverse personas and edge cases
- Score conversations automatically with customizable scorecards
- Switch between OpenAI, Anthropic, and Google models in one click
100 batch runs
Look up the order. Process the refund. Text the confirmation.
Connect tools via MCP protocol, import them from OpenAPI specs, or write custom logic in JavaScript or Python. Group tools into reusable toolsets and assign them to agents. Add a knowledge base so your agent answers from your actual docs. Chanl handles tool execution on every channel.
- Import entire API surfaces from OpenAPI specs in one upload
- Write custom tools in JavaScript or Python, fully sandboxed
- Group tools into toolsets and share across multiple agents
- Knowledge base search and persistent memory work on every channel
Connected Tools
Salesforce
Slack
HubSpot
Stripe
OpenAI
Anthropic
Intercom
Cohere
Zapier
Google
GitHub
ShopifyValidate before any channel goes live
Every agent goes through a draft → validate → publish workflow. Chanl checks for missing prompts, broken tool configs, and required fields automatically. Clone any agent to test a new model or prompt strategy without touching what’s live. Track every deployment from one dashboard.
- Automated checks catch missing tools, bad configs, and empty prompts on every publish
- Clone agents to A/B test models or prompts without touching production
- Deployment dashboard shows status per channel and platform
- Roll back to a previous version if something goes wrong
Draft
CompletedConfigure model, voice, tools, prompts, and memory
Validate
PassedAutomated checks for missing config and readiness
Deploy
LiveLive on VAPI, ElevenLabs, and Bland simultaneously
Your voice agent just got a website.
Add one script tag to your site and call ChanlWidget.init() with your agent ID. Your agent is live on the web with the same tools, memory, and knowledge it uses on voice. Customers who called yesterday are recognized when they chat today.
- One script tag and one line of config to go live
- Shared memory means customers are recognized across channels
- Customize theme, position, and branding to match your site
- No framework, no build step, no dependencies
Frequently Asked Questions
“Only 18% of organizations have successfully deployed an AI agent in production. The gap isn't building — it's testing, observability, and trust.”
Gartner, 2025 — AI Agent Readiness Report