Eloelo Group built its business on live entertainment and creator tools for the Indian market. Dolphin AI is a different kind of product: an agentic video production pipeline aimed at creators, brands and studios. The company traces the platform back to its own production arm, Story TV, which has made more than 5,000 original dramas, and to internal pipelines that handle over 100,000 creative assets a month. That is the pitch behind it. A tool built inside a content factory before it was sold as a tool, as reported at launch.
What Dolphin AI does
The core module is called Film Studio. You feed it a script and it returns a structured multi-shot plan instead of a single clip. Assets such as a character's face, wardrobe, props and locations get locked at plan stage, and the same locked set is reused across every shot in the sequence.
That addresses the problem anyone who has tried AI video will recognise. Shot one and shot six look like different productions. Different jacket, different jawline, different room. Dolphin AI's continuity engine keeps context, voices, wardrobe, props, locations and visual style in a consistent state across shots. Around that sit three lighter tools: a natural-language Chat interface, Templates for repeatable formats, and a shot-level Playground for adjusting individual renders.
The launch presentation was given by founder and CEO Saurabh Pandey on 29 September. Coverage from the event did not include independent demonstrations.
The 60-second claim
Most current video models fall apart somewhere between eight and ten seconds. Objects morph, faces drift, and a second shot rarely matches the first. Dolphin AI claims up to 60-second multi-shot output in a single pass while holding narrative context.
No benchmark has been published to support that. There is no side-by-side with OpenAI's or Google's video systems, no resolution specification, no count of how many render passes "single pass" allows, and no statement about how much compute one minute of output consumes. The demos shown at the launch were the company's own. At AI Arena we publish measured throughput and latency for the models we run ourselves, because vendor launch videos are not evidence of anything except rendering time on a marketing budget.
MCP support is the part developers will care about
Dolphin AI integrates with the Model Context Protocol. Assistants such as Claude, ChatGPT, Gemini and Cursor can trigger video generation through the platform rather than through a separate dashboard. In practice, an agent could draft a script, request a shot plan, adjust a scene and issue a render inside the same conversation.
Whether that holds up at production scale is unproven. Eloelo has not published rate limits, per-render costs, concurrency caps or the authentication model for MCP clients. For a solo creator, MCP support is a useful shortcut. For a studio wiring this into an existing pipeline, those missing numbers decide whether the integration is usable at all.
Pricing is not public
Dolphin AI plans three subscription tiers: Starter, Pro and Creator. No price has been announced for any of them. Eloelo's FY25 operating revenue was Rs 69.5 crore, which converts to roughly €6.7 million or $7.9 million at approximate current exchange rates. That gives some sense of the company's size, not of what the tiers will cost.
For a European buyer this is the practical blocker. No EUR pricing, no stated VAT handling, no published compute quotas. You cannot put a line in a budget, and you cannot compare it with what you already pay OpenAI, Google or Runway. The company says tier pricing will follow the early access period, without giving a date.
What EU users should check before signing up
The waitlist appears open regardless of region, and Eloelo has not published where inference runs or whether EU data residency is offered. If your scripts or reference footage contain personal data, including employee or customer likenesses, that processing needs a data processing agreement with the vendor.
There is also the transparency obligation under Article 50 of the EU AI Act, which took effect in August 2026 after the Digital Omnibus pushed high-risk compliance deadlines to December 2027. Artificially generated content must be marked in a machine-readable format, and deepfakes must be disclosed. If you publish a Dolphin AI clip in the EU, part of that duty lands on you as the deployer, not only on the vendor. Ask what provenance marking the platform outputs before you build a campaign on it, and ask it in writing.
Where it sits in the market
Video generation is crowded. OpenAI, Google, Runway and several Chinese vendors all have working products, and most of them ship with published API pricing. Europe has no comparable video platform of its own at this scale, so European buyers keep consuming US or Chinese APIs, with the data-handling questions that follow.
Dolphin AI's differentiator is workflow rather than raw model quality, at least on paper. A script-to-shot-plan pipeline with locked assets, continuity across cuts, and agent access through MCP addresses a real gap. It is also a claim set that has not been tested outside Eloelo's own studios, and the launch material does not name a single external studio that has used it.
The platform is in early access, with pricing to follow. Until the company publishes tier costs and an independent test of the 60-second figure exists, the accurate description of Dolphin AI is a well-shaped production workflow with an unverified core promise.
Is Dolphin AI available in the EU?
The early access waitlist is open and Eloelo has not announced regional restrictions. The company has also not published where its models run or whether EU data residency is available, which matters if you process personal data or client footage.
How much will the Starter, Pro and Creator tiers cost?
No figures have been released. Eloelo says pricing will be published after the early access period, so there is currently no EUR or USD rate to compare against existing video tools.
Do I have to label videos made with Dolphin AI?
Under Article 50 of the EU AI Act, artificially generated content published in the EU has to be marked in a machine-readable format, and deepfakes must be disclosed. The obligation applies whether you generate the clip yourself or a contractor does it for you.
