Hooman Labs

Automate Phone Calls with AI

[Hooman Labs @hoomanlabs]

Ask the chat widget. Get nowhere. Call. Explain it all again.

Not a chat problem or a voice problem — two vendors that don't know about each other.

Same agent on chat and calls now, context carried across both.

[Hooman Labs @hoomanlabs]

Language-switch latency: 800ms → 90ms.

When a caller asks for Hindi, our agent now switches in under a tenth of a second. Across 6 Indian languages.

#VoiceAI #ConversationalAI #Latency #Multilingual

[Hooman Labs @hoomanlabs]

Most voice agent QA happens in a spreadsheet. Export transcripts, read a sample, mark them up somewhere else.

You can now review a conversation where the conversation is.

[Hooman Labs @hoomanlabs]

Changing what a voice agent knows shouldn't need an engineer.

Routing lists, screening criteria, and team mappings, ops knowledge that changes weekly, usually buried in a prompt only a developer can access.

Now it's a table. Edit a cell, publish, next call uses it.

[Hooman Labs @hoomanlabs]

Every voice agent prompt has a sentence starting with "always" or "never".

Most of the time the model honours it.

"Most of the time" is fine in a demo and not fine at 4,000 calls a day.

Those instructions can now be expressions instead of prose.

[Hooman Labs @hoomanlabs]

At Hooman Labs, we are building voice agents to automate inbound and outbound calls for e-commerce companies. Our mission is to make handling voice based customer support and engagement super easy.

Check out a demo video of how easy it is to return an order on a @Shopify store.