Overview/MCP/AI PPT

AI PPT

Create AI PPTs and read their document artifacts through MCP.

AI PPT through MCP

AI PPT uses slides.

Call kp_get_catalog, then pass valid arguments to kp_create_creation:

{  "type": "slides",  "prompt": "Create an onboarding presentation that explains MCP to new product managers",  "language": "en",  "aspect_ratio": "16:9",  "scene_count": 8}

Treat AI PPT as a document workflow. Do not send duration_seconds or assume a narration voice is required.

After completion, read artifacts, scenes, and main_url from kp_get_creation. Ready artifacts may include PPT or PDF files.

When to choose AI PPT

Choose AI PPT when the audience needs a presentation they can read, present, export, or revise as a document. Give the prompt a clear audience, the decision or lesson the deck must support, and the logical order of its pages. For example, an onboarding deck should name the role, first-week decisions, and practical takeaways instead of only naming the topic.

Treat it as a document workflow

Use scene_count only from kp_get_catalog to control the presentation scope. Do not send video-only fields such as duration_seconds or assume a voice is required. After kp_create_creation, poll the returned job, then retrieve main_url, scenes, and artifacts through kp_get_creation. Use the returned PPT or PDF asset rather than guessing an export URL.

If the scope or structure is wrong, create a revised presentation with an updated prompt and catalog-valid settings. Check the job status before retrying a failed request so a rate, entitlement, or validation issue is handled deliberately.

Continue from here