OpenAI Partners with Reliance to Power AI Search on JioHotstar

Reliance-owned JioHotstar is adding a conversational, ChatGPT-powered discovery layer to its streaming app, replacing rigid keyword search and endless scrolling with natural-language voice + text prompts that work across multiple languages. The feature is built on OpenAI’s APIs and is being rolled out in phases across live + on-demand experiences.

This is not just an “in-app search upgrade.” The companies are also planning a two-way integration where ChatGPT itself can surface JioHotstar recommendations with deep links into the platform, turning ChatGPT into an external discovery funnel for entertainment.

What exactly is being added to JioHotstar?

1) “Multilingual Cognitive Search” (voice + text)

JioHotstar says viewers will be able to speak or type intent in natural language, mood, situation, context, niche themes, and get context-aware recommendations (not just literal keyword matches).

Examples cited in coverage include prompts like:

“My parents are visiting, suggest something we can all watch together.”

“Show me movies about identical twins.”

2) Live sports become “ask while you watch”.

The assistant is expected to extend beyond movies/series into live sports, letting viewers ask for scores, highlights, match moments, and player info (conversationally, in multiple Indian languages).

3) Phased rollout (not a one-shot launch)

Multiple reports emphasize this will start with select experiences and expand in phases across live and on-demand content.

The scale: why this move is strategically big

JioHotstar is operating at a scale where discovery is the product:

450+ million monthly average users

300,000+ hours of catalogue

19 languages

At that size, “what to watch” friction isn’t a UX nit—it’s a retention and engagement lever. JioHotstar’s bet is that conversational discovery can cut time-to-content and increase session depth by letting users describe intent naturally rather than browsing menus.

Why now: the India AI Impact Summit + OpenAI’s India push

The partnership was announced around the India AI Impact Summit in New Delhi, where a wave of AI infrastructure and partnership announcements were made.

Two macro drivers are obvious:

1) Reliance’s “AI at population scale” thesis

Reliance leadership has publicly framed AI as the next “cost curve” to crash (similar to mobile data), alongside major investment signals around compute and infrastructure.

2) OpenAI’s accelerated India expansion

TechCrunch reports India has 100M+ weekly ChatGPT users, and OpenAI plans additional offices in Mumbai and Bengaluru (beyond its Delhi presence).

That combination, massive distribution (JioHotstar) + a fast-growing AI layer (OpenAI/ChatGPT), creates a practical path to mainstream conversational interfaces in everyday entertainment.

Competitive context

This partnership lands in a broader industry shift:

Netflix has tested conversational discovery experiences using OpenAI/ChatGPT-style capability.

Google TV has pushed Gemini-based content discovery features.

In India specifically, where discovery spans languages, genres, and wildly different audience segments, conversational interfaces may be more than a novelty—especially when voice input is natural and typing is a barrier.

The product implications

If executed well, this feature can change four core streaming mechanics:

Discovery becomes intent-led, not catalogue-led
Users start from “what I feel like” (mood/context) rather than “what I remember exists.”

Regional + multilingual navigation becomes frictionless
A voice-first interface reduces dependency on exact spellings, titles, and English-first discovery flows.

Live sports engagement becomes interactive
Asking questions mid-stream shifts viewing from passive consumption to “second-screen inside the same screen.”

ChatGPT becomes a distribution channel
The planned two-way integration means the discovery journey might start outside JioHotstar—inside ChatGPT, then jump into a stream via deep links.

What’s still unknown

Even with lots of announcements, several practical details haven’t been disclosed publicly:

Which exact Indian languages at launch (and whether voice quality is uniform across them)

Latency for voice → understanding → recommendations (critical for live sports)

Safety filters / kids mode controls for conversational prompts

Data handling + personalization boundaries (what signals are used, how “history” influences outputs, opt-outs)

A clean rollout will likely include visible user controls (“use watch history for recommendations,” “clear conversation,” language toggles, etc.). Until those UI details ship broadly, treat performance claims as directional rather than proven.