Price: 25000
Number of applications: 0
17.11.26 (inclusive)
Payment under the contract in stages: MVP development (information gathering → plot → generation → assembly for one format), then a pilot with real customers, followed by finalization into a finished product. Joint development with equity participation in the product is possible.
MVP
ICT tasks
Media sphere
Neurotechnology and artificial Intelligence
Purpose and description of task (project)
The goal is to create a system that automatically produces a finished promotional video based on minimal input data about the company, comparable in quality to the work of a production studio. The system should sequentially go through four stages: 1. Information gathering. Brief questionnaire (purpose of the video, target audience, UTP, placement platform, duration, language); automatic analysis of the company's website and social media; uploading of brand guidelines, logo, product photos. 2. Plot construction. Generation of several creative concepts to choose from, script, frame-by-frame storyboard with a description of each scene. User approval and edits at each step. 3. Video generation. For each scene, the system automatically selects the optimal model and provider (e.g., Runway, Kling, Google Veo, Seedance, etc.) based on the type of scene, quality, cost, and duration. It maintains uniformity of characters, product, and style between scenes using references. Automatically evaluates the result and regenerates flawed parts. 4. Final assembly. Editing, voiceover (in Russian and Kazakh), music, subtitles, logo, and call to action. Export in 9:16, 1:1, and 16:9 formats for Instagram, TikTok, YouTube, and TV. Additional features: web interface, manual refinement at any stage, a single integration layer with provider APIs (new models can be connected without system redesign), cost tracking and control for each project.
Software/ IS
Producing advertising videos remains expensive and time-consuming: filming, actors, editing, and approvals take weeks and require a budget that is unavailable to most small and medium-sized businesses. As a result, companies either do not use video advertising or do it rarely and poorly, even though video is now the main format on social media. Generative video models could solve this problem, but they are currently difficult for businesses to use: — the market is fragmented: each model has its own strengths (realistic people, product shots, animation), and only an expert can choose the right one for a specific scene; — expertise in prompting, scripting, and storyboarding is required; — models generate separate clips lasting a few seconds, and characters and products look different from scene to scene; — assembling the video (editing, voiceover, music, formats) is done manually in other programs; — the cost of generation is difficult to predict and control; — there are few tools with high-quality voiceovers and texts in the Kazakh language. There is no solution that integrates the entire process — from brief to finished video — in one accessible workflow for someone without video production experience.
— the time required to create an advertisement video is reduced from several weeks to several hours; — production costs are reduced by [5–10] times compared to traditional filming; — video advertising becomes accessible to small and medium-sized businesses; — it is possible to quickly create multiple versions of a video for A/B testing and different platforms; — content is localized into Russian and Kazakh; — generation costs are optimized through automatic model selection for each scene; — the system is not dependent on a single provider and easily transitions to new, higher-quality models.
Нурмаканов Ануар Максатович
Note
Only about MVP, full product development will be planned later
Whatsapp
