Broadcaster
A broadcaster or OTT operator whose engagement metrics show that a large share of sessions end without a play
AI & automationTestingBuilt by LabsLast updated 29 July 2026
Content BuddyConversational content discovery instead of catalogue browsing
Viewers leave without watching anything because the catalogue never surfaces what they actually want. Content Buddy replaces browsing with a conversation: users describe what they are in the mood for, typed or spoken, and the assistant draws on their viewing history and the full catalogue to reply with exactly the right titles - from VOD picks to what is on live tonight.
Personalized Content RecommendationsNatural Language ProcessingVoice Interface
Watch the demo
Click to unmute
or try the live version
Content Buddy is password protected
Enter the demo password to launch the interactive demo.
Don't have a password? Ask us for one and we will send it over.
Request demo accessThe problem this solves
A catalogue with thousands of titles is worse than useless to a viewer who does not already know what they want. Scroll-and-browse produces decision fatigue, recommendation carousels recycle the same top titles, and search only works when you already know what to search for. The viewer gives up and opens a competitor app that feels easier.
Personalisation engines improve the situation but they still push rather than converse. They cannot handle 'something like what I watched last week but shorter' or 'something good for the kids tonight' - queries that are completely normal to say aloud but that no browse interface was designed to receive.
Who this is for
This is relevant if you are:
Telco
A telco bundling a video catalogue who needs to make that catalogue feel usable rather than overwhelming
How it works
- 1The user types or speaks a request in their own words - a mood, a genre, a show they half-remember, or a question about what is on tonight. Spoken input is transcribed with OpenAI Whisper, or by the browser's own speech engine where that is available.
- 2The agentic layer interprets the request and decides which tools to call: the viewer profile for history and preference data, and the catalogue for matching VOD and EPG titles. Every candidate is verified against the catalogue before it is offered, so the assistant recommends titles that exist rather than titles that sound plausible.
- 3The catalogue and metadata behind those answers come from the InsysGO API, so the live demo on this page runs against a real OTT platform rather than against a fixture.
- 4The assistant replies conversationally with a ranked shortlist and reads the answer back through Amazon Polly or OpenAI TTS, so the whole exchange works hands-free on a remote or a set-top box - in English or in German.
- 5Session state is kept for the length of the conversation, so the assistant remembers what it has already suggested and narrows down from a follow-up instead of starting over.
What makes it different
Recommendation engines push titles at viewers; Content Buddy lets viewers pull through conversation. The difference matters because a viewer who can describe what they want in their own words is far more likely to find something that satisfies them than one who is scrolling hoping to recognise something. The voice path makes this usable on TV remotes and set-top devices where typing is impractical, and the multilingual support means the same assistant can serve audiences in different markets without a separate build per language.
Current status & next steps
TestingReviewed 29 July 2026
Content Buddy is in controlled testing. The demo on this page is available to selected partners on invitation, which is exactly what the Testing status commits us to - the password gate is the access control in practice.
Language support currently covers English and German, switchable in the chat window settings, although changing it resets the session. Recommendations can take several seconds to appear whenever content descriptions have to be translated for a non-German language. Speech handling has rough edges too: where OpenAI Whisper is unavailable the demo transcribes through the Web Speech API, which is less accurate, and OpenAI text-to-speech is occasionally inaudible or slow in some browsers, which switching the engine to Amazon Polly in the settings resolves.
The next step is to run the assistant against a customer's live catalogue and user profile data, so recommendation quality can be measured on real viewing behaviour.
Want to add conversational discovery to your platform?
Tell us what catalogue you have and how viewers currently navigate it, and we will tell you what a pilot of Content Buddy on your platform would involve.