{"ok":true,"case":{"id":6,"category_id":1,"slug":"deepseek","brand":"DeepSeek","title":"DeepSeek: Making advanced reasoning feel useful","summary":"A creator education system that turns DeepSeek’s reasoning and coding strengths into clear, repeatable everyday workflows.","markets":"North America · Europe · Southeast Asia","platforms":"TikTok · Instagram · YouTube · X","challenge_text":"The DeepSeek concept centres on a familiar problem for people comparing AI tools: a strong-looking answer is easy to circulate, while the task, inputs and checking behind it are much harder to see. Developers, students and knowledge workers may ask different questions, but all need a way to distinguish a compelling demonstration from a workflow they can meaningfully evaluate.\n\nThe challenge is to explain reasoning and coding scenarios without treating a fluent response or a selected benchmark as proof of broad superiority. A useful creator story makes the task and its limits legible. It also keeps the creator's assessment visible, so the audience can understand why an output was accepted, revised or left unresolved.","strategy_text":"Use screen-led demonstrations as the proposed core, supported by practical prompt discussions and carefully bounded benchmark explainers. Each piece should start with a question the intended audience recognises. A coding example needs a defined expected behaviour; a knowledge-work example needs a clear source and purpose. The creative promise is an understandable method of evaluation, not an unqualified claim about the tool.\n\nChoose creators who can narrate their checking as clearly as their interaction. A developer's value is not only producing a snippet but explaining how it is examined. An educator can make the same discipline accessible by showing where an answer fits the question and where additional work remains.\n\nWhen a comparison is useful, define its scope before showing the result. Keep task conditions visible and avoid translating a narrow example into a ranking of every use case. Short formats can isolate a single decision, while longer material carries the context needed to interpret it fairly.","execution_text":"Prepare the demonstration around a reproducible task and material suitable for public use. Record the relevant environment and inputs, then keep the actual interaction and the subsequent checks connected. If an output needs correction, the correction belongs in the story when it explains the workflow; removing it merely to create a seamless reveal weakens the teaching value.\n\nFor a coding scenario, show the expected behaviour and the observed verification rather than stop when code appears on screen. For another reasoning task, explain the basis on which the creator judges the response. Any benchmark discussion should identify what the chosen example can and cannot support, with source material checked before scripting a comparative statement. Make the expected result understandable before presenting the output. This lets the viewer follow the assessment rather than rely entirely on the creator's declaration that something worked.\n\nEdit for readability and sequence. A viewer needs time to see the input, understand the response and follow the evaluation, even if not every line appears in a short clip. Follow-up content can revisit a confusing step with a different task, helping separate a reusable practice from an answer that happened to work in one recorded example.","results_text":"Evaluation should focus on the quality of understanding around the workflow. Detailed questions about assumptions, attempts to reproduce a bounded example and requests for clarification can be more informative than praise for an impressive answer. Authorised traffic or usage information can add context, but cannot by itself verify output quality or suitability for a different task.\n\nIf viewers treat the example as a universal ranking, the next version needs a more explicit scope. If they reproduce the prompt but omit the checking, the editorial balance should shift towards evaluation. The proposed outcome is a clearer basis for comparing and testing tools in context. A follow-up can deliberately change the task while retaining the same evaluation method, making it easier to see which parts of the practice transfer and which assumptions need revision.","metric_1_value":"58432000","metric_1_label":"Views","metric_2_value":"128","metric_2_label":"Creators","metric_3_value":"62%","metric_3_label":"Code demo completion rate","logo_path":"/brand-logos/deepseek.png","cover_path":"","is_demo":0,"case_type":"real","reporting_scope":"2025-07-01 to 2026-10-10. Platforms: TikTok, Instagram, YouTube, X (Twitter). Total views of project creator content, including paid promotion and organic distribution. Unique creators who participated and published; a creator publishing on multiple platforms is counted once. Recorded totals: 58,432,000 views; 128 creators.","category_name":"AI Tech","category_slug":"ai-tech","translation_incomplete":0,"metrics":[{"value":"58.432M","label":"Views"},{"value":"128","label":"Creators"},{"value":"62%","label":"Code demo completion rate"}]},"related":[{"id":5,"category_id":1,"slug":"dola","brand":"Dola","title":"Dola: An AI companion built into daily rhythm","summary":"Creator stories position Dola as an approachable assistant for planning, conversation and everyday problem-solving—not another abstract AI promise.","markets":"North America · Europe · Southeast Asia","platforms":"TikTok · Instagram","logo_path":"/brand-logos/dola.jpg","cover_path":"","is_demo":0,"case_type":"real","reporting_scope":"2026-04-01 to 2026-10-10. Platforms: TikTok, Instagram. Total views of project creator content, including paid promotion. Unique creators who participated and published lifestyle or daily planning content. Recorded totals: 7,280,000 views; 32 creators.","category_name":"AI Tech","category_slug":"ai-tech","translation_incomplete":0,"metrics":[{"value":"7.28M","label":"Views"},{"value":"32","label":"Creators"},{"value":"12.4K","label":"App download conversions"}]},{"id":4,"category_id":1,"slug":"perplexity","brand":"Perplexity","title":"Perplexity: Turning search into an answer journey","summary":"Expert-led content shows how Perplexity connects sourced answers, research depth and follow-up discovery for people who need clarity quickly.","markets":"United States · Europe","platforms":"TikTok · Instagram · YouTube · LinkedIn","logo_path":"/brand-logos/perplexity.png","cover_path":"","is_demo":0,"case_type":"real","reporting_scope":"2025-09-01 to 2026-10-10. Platforms: TikTok, Instagram, YouTube, LinkedIn. Total views of project creator content, including paid promotion and organic distribution. Unique creators who participated and published research challenges, source comparisons or expert workflow demos. Recorded totals: 27,120,000 views; 68 creators.","category_name":"AI Tech","category_slug":"ai-tech","translation_incomplete":0,"metrics":[{"value":"27.12M","label":"Views"},{"value":"68","label":"Creators"},{"value":"29%","label":"Product search growth"}]},{"id":3,"category_id":1,"slug":"chatgpt","brand":"ChatGPT","title":"ChatGPT: Showing the range behind one conversation","summary":"A creator matrix demonstrates how ChatGPT supports writing, learning, ideation and productivity across radically different audience needs.","markets":"United States · Europe","platforms":"TikTok · Instagram · YouTube · X","logo_path":"/brand-logos/chatgpt.svg","cover_path":"","is_demo":0,"case_type":"real","reporting_scope":"2025-08-01 to 2026-09-25. Platforms: TikTok, Instagram, YouTube, X (Twitter). Total views of project creator content, including paid promotion and organic distribution. Views are not deduplicated across platforms. Unique technology, education, productivity and creative creators. Recorded totals: 52,347,800 views; 135 creators.","category_name":"AI Tech","category_slug":"ai-tech","translation_incomplete":0,"metrics":[{"value":"52.3478M","label":"Views"},{"value":"135","label":"Creators"},{"value":"76%","label":"Product demo completion rate"}]}]}