Plug real-time conversational video AI into any application. Eclatira gives developers native voice-to-voice, live vision, and full-stack execution across custom APIs, MCPs, and 3,000+ apps. Ship autonomous multimodal video agents fast.
Congrats. What I find unique is the merger of voices, vision and actions into one conversational engine. There's a lot of potential here for turning existing SaaS products into interactive AI experiences. Excited to see the roadmap.
Voice-to-voice plus real-time combined with vision feels like the future of AI assistants. Static text based assistant will start to feel outdated. Nice work. 🔥
The live vision piece is what stands out here, since most conversational agents are voice only and lose all context the moment something visual matters. I build voice AI that calls aging parents every day for Callie Care, and the thing that consistently bites us is latency the moment a tool call lands mid conversation. How are you handling barge in and turn taking when the agent is executing against an external API in the same turn, do you buffer a filler response or let it go quiet? Also curious how you keep the video agent from feeling uncanny on longer sessions.
@igorgurovich since its voice to voice, the barge in mechanisms is very different that tts stt so barge in and turn taking is handled beautifully and external tool calling doesn't make any difference in quality.
we build voice-only agents for phone calls and latency is already the hard constraint before you add any vision. curious how much slower the loop gets once you're also processing live frames, is video handled on a separate track so it doesn't block the voice turn-taking, or does adding vision to the pipeline push response time up across the board
Documentation.AI
Congrats. What I find unique is the merger of voices, vision and actions into one conversational engine. There's a lot of potential here for turning existing SaaS products into interactive AI experiences. Excited to see the roadmap.
Eclatira
@roopreddy yesss lots more to come. Really appreciate the comment!
Eclatira
@hamza_afzal_butt Thank you!! Appreciate it man!
Voice-to-voice plus real-time combined with vision feels like the future of AI assistants. Static text based assistant will start to feel outdated. Nice work. 🔥
Refocus
The live vision piece is what stands out here, since most conversational agents are voice only and lose all context the moment something visual matters. I build voice AI that calls aging parents every day for Callie Care, and the thing that consistently bites us is latency the moment a tool call lands mid conversation. How are you handling barge in and turn taking when the agent is executing against an external API in the same turn, do you buffer a filler response or let it go quiet? Also curious how you keep the video agent from feeling uncanny on longer sessions.
Eclatira
@igorgurovich since its voice to voice, the barge in mechanisms is very different that tts stt so barge in and turn taking is handled beautifully and external tool calling doesn't make any difference in quality.
Dial
we build voice-only agents for phone calls and latency is already the hard constraint before you add any vision. curious how much slower the loop gets once you're also processing live frames, is video handled on a separate track so it doesn't block the voice turn-taking, or does adding vision to the pipeline push response time up across the board
Eclatira
@galdayan yes exactly its handled separately
Dial
@moad_rahali_semlali good to know, that's one less thing to worry about if we ever add vision on our side. nice work
Macaly
plug and play video agents is a real unlock 👏 latency ok for live calls?
Eclatira
@petrkovacik yes absolutely it will help a lot building video solutions