Platform features
Everything the platform layer does, in one index. One agent, any channel, complete workflows.
Each entry states what it is and links to the page that explains it.
One AI system across voice, SMS and email.
One AI system across voice, SMS and email. One agent, any channel, complete workflows.
Turns conversation into action.
What separates a call that works from a demo that works.
Knows “uh-huh” isn’t an interruption. Classifies raw audio, not energy heuristics.
AgentsWhen interrupted, the agent resumes from what the caller actually heard. Not what the model generated, or what the script assumed.
AgentsEvery sentence is checked. The verdict lands in ~20ms per check, alongside synthesis rather than in front of it.
Alebex Voice Engine
One agent, any channel, complete workflows.
Qualifying leads, in real time as the conversation happens.
Booking meetings, directly into the calendar.
Answering questions and resolving issues.
Updating the CRM as the work is done.
Triggering workflows inside existing business systems.
Handing off to a human when it should.
Canadian data residency, Canadian compliance.
Recordings, transcripts and customer information are held in Canada.
A Vancouver-based Canadian AI company, incubated at the Alexander Innovation Centre.
Designed to support PIPEDA, CASL and applicable CRTC requirements, with configurable consent, disclosure, opt-out and retention controls. No certification implied.
Configurable opt-out controls across channels, with the contact flagged.
Configurable call-recording consent disclosure, with recordings stored for audit.
The six questions an enterprise checklist actually asks.
In place of a certification badge this stack does not hold, here is the one-line answer to each question a buyer’s own risk review raises — every line is one this platform already stands behind elsewhere.
Speech recognition, turn control, the hallucination gate and streaming synthesis run as one system.
Every sentence is checked alongside synthesis, with rejected audio aborted and regenerated.
Barge-in classification separates a real interruption from a backchannel, then playback tracking preserves what was heard.
Voice, SMS and email turns resolve to one ordered thread instead of three disconnected histories.
Recordings, transcripts and customer information are held in Canada. The platform is designed to support PIPEDA, CASL and applicable CRTC requirements, with configurable consent, disclosure, opt-out and retention controls. No certification implied.
The agent hands work to a person when a request falls outside its approved answers or workflow.
The answer check executes while speech is being prepared, not as another serial stage. That placement protects the response without adding a new wait in front of it.
Word-level playback tracking records what the caller actually heard before speaking over the agent. The next turn resumes there, not at the end of text the model generated.
Every capability is one layer of the same owned execution path.
Try the live demo
Drop in your website. Alebex reads it, learns what you do, and calls you in under 30 seconds — fully trained, zero setup.