Changelog
Voice quality, the agent editor, knowledge base, and the parts of the stack you can hear. Newest first.
Reworked the marketing site around the brand wireframe system, with a new hero chat mock and EarnOS social proof. Cleaned up cross-page typography.
Landing page rebuild with brand wireframe system
iPhone hero chat mock with live waveform
Smoother voice waveform bars on canvas
Anyone can hit a Try A Call agent without signing up. Useful for sharing the product with prospects who want a one-tap demo.
Public Try A Call agent with daily caps
Demo share links with version pinning
Per-org demo rate limiting
Eight industry-specific landing pages with sector-aware compliance chips and four head-to-head competitor comparisons.
Industry pages: healthcare, financial services, insurance, logistics, and four more
Compare pages: vs Retell, vs Bland, vs Vapi, vs Synthflow
Switched TTS to the Cartesia AU endpoint. End-to-end voice latency now sits around 800ms on calls landing in Sydney.
Cartesia Sonic 3 on the AU endpoint
≈80ms TTS round-trip from Sydney
Resolved double-speak on rapid state transitions
Split the runtime from the transports. Voice and text now share one runtime over Pipecat frames, so the same agent behaves identically on every surface.
Transport-agnostic agent runtime over Pipecat frames
Eval harness for repeatable agent testing
Eval suite expanded to multi-turn calls
We send a short note when we ship something that changes call quality. Email subscribe@verticalai.com.au.
