Turn a deck into a compelling video.

What the AI will do

• Start with NotebookLM Video Overview: import your deck, establish what viewers need and help generate a video.
• Check what this LLM can actually read, produce and operate; guide any account steps you must do. Use a documented alternative when the first route cannot meet the brief.
• Review the generated story and actual video, repair errors and deliver a playable result.

What you can provide

• A PowerPoint deck or link. Speaker notes, a rough explanation of the point and the intended audience help, but need not be polished.

Where you'll take part

• Clarify the viewing purpose and any choice that changes the story or production route.
• Perform any account-only import or generation step, and help inspect playback if the AI cannot watch or hear the output.

What you'll get

A playable video based on your deck, with narration, captions and a clear account of what was checked. If generation or export is blocked, you receive a prepared production package with the exact remaining step identified.

Activity context

The easy place to start is Google’s NotebookLM: add your PowerPoint and create a Video Overview. Let the AI help turn a deck that needed you in the room into a video someone can follow on their own.

NotebookLM Video Overview is the first route. Google’s current source list includes PowerPoint .pptx; the AI checks your actual account and can prepare a supported export if needed. Synthesia and HeyGen offer direct presentation-to-video alternatives.

For the user

Give the AI your deck or its location. It will first find out what it can actually do in this setting, then help use NotebookLM or a suitable alternative. You supply the audience and intention if the surrounding records do not establish them; the AI prepares the material, handles supported production steps and guides the few actions that require you.

Guidance for the AI

These are general instructions, not a description of this user's particular AI system, tools, access, or circumstances. Adapt the method to the user's goal and the capabilities you can actually use, while preserving the activity's essential sequence, review points, and intended result.

Start with the activity and the user's context

Read the complete activity and guide the user through it. Use the activity prompt yourself; the user does not need to paste it again. Keep the responsibilities of the user, the AI, and any colleague clear.

Begin with what the user has already supplied. Establish the intended result and resolve missing context through focused questions when needed. Work within the activity's stated scope, permissions, required confirmations, and stopping points, including any deliberate order for introducing information. Make progress with the available information and reasonable, clearly identified assumptions where the activity allows them.

Check the capabilities needed for this work

Inspect relevant available tools, connectors, and data. Use small, relevant, non-destructive checks to establish whether necessary access or a capability actually works. Naming a product does not establish access to it. Let this activity determine what needs checking; do not impose a universal preliminary questionnaire or a full system audit.

When useful access is missing:

• Explain what you cannot reach and how that material would help.
• Investigate practical setup, connection, export, or other access alternatives before asking the user to gather material or troubleshoot.
• If progress depends on access, request only the specific user action or permission needed to enable it or provide the material another way. Continue useful independent work while awaiting the response.
• Offer to walk the user through an export or upload. Welcome files, links, copied text, rough notes, or an interview as supported by your system and the activity, and use the best available route. If the missing access is optional, continue with what is available.

Ask focused questions and respect the answers

Ask questions only when necessary to complete the work well or satisfy the activity's required review points. Build on answers already supplied. Ask one question at a time, or a small group of closely related questions. Wait on work that depends on the answer, and continue useful work that does not. Explain unfamiliar language with examples where helpful.

Confirm answers or decisions you supply on the user's behalf before treating them as the user's final answers or decisions. This does not require separate approval of every sourced finding; follow the activity's specific review requirements.

Keep evidence and independent contributions distinct

Distinguish evidence, inference, and missing information. Check your work against the source material.

Use independent agents when your system actually supports delegation and independent investigation or review would improve the work:

• Give each agent a bounded brief and the context appropriate to its role.
• Keep discovery independent until synthesis where the activity requires it. Preserve any requirement for the user or colleagues to contribute before seeing other explanations.
• Follow the activity's stated fallback and any required confirmation when independent agents or fresh contexts are unavailable. Preserve reviewer isolation where required; describe shared-context or sequential passes accurately.

Produce, check, and revise the result

Do the substantive investigation and production. Carry the activity through to the most complete, usable result within its scope, rather than ending with a plan or handing the main work back to the user. When missing information does not prevent useful work, proceed with clearly identified assumptions where the activity permits them.

Check the result against the activity's completion criteria. At each substantive delivery, state what is complete and any material unfinished work, assumptions, helpful missing information, and remaining uncertainty. Explain how these affect what the user can use, and help the user revise the result.

Preserve the activity's stopping point. Do not turn a rehearsal, investigation, or test-design activity into execution that the user has not authorized.

Your AI prompt

Turn my deck into a compelling, playable video. Use Google’s NotebookLM Video Overview as the first route. If that route cannot meet the agreed need, offer a suitable documented alternative. Begin by probing this LLM’s actual capabilities and guide me through any actions you cannot perform. Do not promise a rendered video because you can write a script.

1. Check what you can do and find the deck’s purpose

Inspect available file access, slide and speaker-note extraction, image viewing, browser or app control, media generation, audio/video inspection, rendering and export tools. Read relevant installed guidance and make a small non-destructive check where possible—for example, read one actual slide and its notes, or inspect the available generation controls. Distinguish supported operations you tested from documented capabilities you cannot access. An LLM account is not proof of NotebookLM access or video-generation access.

Find the supplied deck and verify its version. Read the slides, charts, notes and necessary source material. Search accessible presentation invitations, recipient correspondence, related chats and meeting transcripts for who will watch, their expressed questions, what prompted the deck and the decision or understanding it should produce. Use actual audience evidence to set the explanation level and emphasis; distinguish documented concerns from likely reactions you infer. Keep this search proportionate to the deck.

If access fails, investigate available connection or export routes, then ask for the deck or only the missing material. Explain what you could not read, including notes or embedded content. Establish viewing context, language, approximate length and the one thing viewers should leave understanding or doing. Ask one focused question at a time if these are unclear. Reuse choices already supplied and wait on consequential unresolved changes to the story.

2. Prepare the story for a viewer who is not in the room

Extract the central point, the evidence that supports it, the sequence of explanation and the ending. Identify jargon, unexplained charts and places where the slides rely on a speaker. Preserve qualifications, consequential numbers and source attribution. Use accompanying context to explain the deck, while labeling useful additions and proposed corrections so they are not mistaken for slide content.

Prepare a concise generation brief with the intended viewer, opening question or situation, essential findings, examples, visual emphasis, language and desired approximate duration. Make the progression concrete: establish why the question matters, explain the mechanism or evidence, then land the implication. Do not require a complete bespoke screenplay before trying the straightforward NotebookLM route. If the deck contains conflicting figures or lacks the central evidence, resolve the material issue or mark it before generation.

For a large deck or materially different audiences, divide source checking or audience review into bounded assignments with specified slides, relevant correspondence, findings and uncertainties to return. Reconcile the evidence before production. Use waves or bounded passes when agents are unavailable. A fresh story reviewer should see the brief, deck and necessary audience evidence without your creative rationale; repair misleading emphasis or missing context before rendering.

3. Generate with NotebookLM first

Inspect the actual NotebookLM account and current official guidance. Some Google help pages now use the name Gemini Notebook; follow the product interface I actually have. Current source documentation lists PowerPoint .pptx, Google Slides and PDF as supported inputs. Try the .pptx directly; if my account rejects it or the original is an older .ppt file, prepare a supported export yourself where possible. Check imported slide content and important notes; add a clearly labeled notes source if necessary. Google: supported sources and imports.

Create or use a dedicated notebook containing the intended deck and clearly identified supporting material. Open Studio → Video Overview, customize the available format, language, visual style and instructions with the brief, and generate. Play the finished result and use Download for a local file where available. These are documented controls, not a guarantee of account access or identical output. Google: Video Overviews.

Treat this as a generated explanation of the deck’s sources; check whether it preserves the material and viewing experience we need. If the brief requires keeping specific slides or narration exactly, inspect the result and consider the alternatives below rather than claiming the overview guarantees that fidelity.

Do all actions your tools support. Lack of your own browser control does not make NotebookLM unusable: provide the exact source files and paste-ready customization, then guide my next import or generation action. Wait for the result you cannot observe and continue from it. If the route is blocked or unsuitable, identify the actual reason and let me choose the alternative when it changes tools or the result.

4. Use a direct slide-to-video alternative when needed

Check current documentation and account access before recommending a switch. Two documented options are:

• Synthesia: its standalone Import PowerPoint route accepts .pptx. Import the deck, use speaker notes to populate narration where supported, review the editable scenes, then Generate Video. Imported PowerPoint animations are not preserved. This flow differs from uploading a file to Synthesia Assistant. Synthesia: Import PowerPoint.
• HeyGen: open PPT/PDF to Video, upload PPT, PPTX or PDF, choose the presentation settings and edit in its studio. PPT/PPTX supports speaker-note scripts and editable elements; PDF does not offer those same options. Review each scene’s narration, then export MP4. HeyGen: PPT/PDF to Video.

Recommend the route that fits the required fidelity, available access and approved use of the material. Check actual file limits, plan restrictions and export settings before committing production to it. Prepare any conversion and complete the supported import, narration editing, rendering and export yourself; guide only user-only actions.

If we choose a custom LLM-led route, first establish which available tools can produce the whole deliverable. Write the spoken script and scene plan from the checked source brief, specifying narration, on-screen material, transitions and approximate timing for each scene. Render a short representative segment to confirm the tools can combine visuals and narration and produce a playable file before producing the rest. Review the script against the deck, build the complete video, and preserve editable production files. If this environment can only prepare text, say so and use the guided vendor route or deliver a clearly labeled production package; do not silently redefine success as a script.

5. Watch, listen, repair and deliver

Inspect the actual generated video and narration with available media tools. Check the opening, explanatory sequence and ending; verify key numbers and claims against the deck and sources; inspect chart labels, cropping, readability, pacing, pronunciation, gaps and audio quality. Review or create captions from the actual spoken audio using supported tools, then check wording and timing. If captions cannot be embedded, provide a checked companion caption file where feasible and state how it is used.

Give a fresh reviewer the video, deck and audience brief without your production rationale. Ask whether the intended viewer can follow the point and whether the video distorts or omits anything consequential. Separate content checking from audiovisual checking: a transcript cannot establish visual or sound quality. Disclose shared-context review or media you cannot inspect, and ask me for the smallest necessary playback check. Fix material problems through editing or regeneration and recheck affected scenes.

Deliver the playable video or an access-checked private playback link, captions, and any editable project or script available from the route. Confirm that the file opens and plays, or distinguish my reported playback from your verified check. Do not publish the notebook, broaden access or send the video to others without authorization. If rendering, download or playback access remains blocked, deliver the prepared sources, brief and production files with the exact remaining action. That is useful production preparation, explicitly short of a completed video.

Done when

You have a video that plays, communicates the deck’s point and has had its content and available audiovisual checks completed, with captions or their precise remaining step. If production access stopped the work, the prepared package is labeled as an intermediate result.

Research & resources

Read the supporting research. Download research as Markdown.

These instructions were written in September 2026 using the frontier models available at the time. You may now have better tools or models, or restrictions that make some tasks harder. Your AI should adapt to the tools and access you actually have and find the best available way to help. Everything here is a starting suggestion for you and your AI, not a proven best practice for every setup. Feel free to rewrite any of it to better meet your needs.