Skip to content

script.md

Latest commit

 

History

History
231 lines (148 loc) · 8.33 KB

script.md

File metadata and controls

231 lines (148 loc) · 8.33 KB
Raw

MemContext Public Launch Video Script

Goal

Create a polished public launch video for Twitter/X, LinkedIn, the website, and product launch posts. This is not an investor-targeted video, but investors should understand the product depth by watching it.

The video should make people think:

My AI should not start from zero every time. It should remember what matters, know where that memory came from, and let me control it.

Core Positioning

MemContext is the memory and context layer for AI agents and apps.

It has two connected parts:

  • Persistent User Context: personal memories for users, agents, and apps.
  • Context Vault: a workspace knowledge vault for team documents, URLs, files, and curated facts.

The strongest line to own:

The memory layer your AI, and your team, can actually trust.

Secondary line:

Memory that evolves. Not just stores.

Video Direction

  • Length: 75-90 seconds.
  • Style: polished product launch, not founder talking head.
  • Audience: AI builders, founders, teams, and power users.
  • Structure: one realistic workflow, not a random feature montage.
  • Technical level: plain-language first, with quick proof through UI, MCP, SDK, and API visuals.
  • Competitors: do not name competitors in the public voiceover.

Main Story

Use a launch/product workflow as the anchor.

Scenario:

A founder or team asks an AI assistant:

“Draft our launch post using our latest positioning, match my writing style, and avoid old claims.”

Without MemContext, the AI guesses or needs everything pasted into the prompt again.

With MemContext, the AI retrieves:

  • Personal writing preferences from Persistent User Context.
  • Product positioning from Context Vault.
  • Source documents and extracted facts with citations.
  • Updated memories instead of stale decisions.
  • Feedback and version history when something changes.

The output feels better because the AI is using the right context from the right place.

Voiceover Draft

Use this as a base script. Keep it natural; the editor can trim lines for pacing.

0-8s: Hook

Your AI should not start from zero every time.

It should remember your preferences, your decisions, your projects, and the knowledge your team already has.

8-18s: Problem

Today, most AI tools still forget the context that makes their answers useful.

You repeat your stack. You paste the same docs. You correct the same outdated facts.

18-30s: Introduce MemContext

MemContext is the memory and context layer for AI agents and apps.

It gives AI persistent user context for individuals, and a Context Vault for shared team knowledge.

30-45s: Persistent User Context

Save preferences, facts, decisions, and project context once.

Then retrieve them from your dashboard, your app, your SDK, your API, or any MCP-connected AI tool.

MemContext uses hybrid search, version history, feedback, and temporal expiry so memory can evolve instead of becoming stale clutter.

45-65s: Context Vault

For teams, Context Vault turns PDFs, docs, URLs, images, CSVs, and curated facts into searchable AI context.

It extracts useful memories, keeps source passages connected, and shows citations so your AI can answer from trusted knowledge, not guesswork.

65-78s: Trust + Platform

Everything is scoped by user, tenant, workspace, or project, so the right context stays in the right place.

And because MemContext works through MCP, REST API, and a TypeScript SDK, you can bring the same memory layer to your agents, apps, and internal tools.

78-90s: Close

MemContext is memory that evolves. Not just stores.

Give your AI memory.

Shot List

1. Cold Open

  • Show quick cuts of repeated prompts: “Remember I use pnpm,” “Use our new launch positioning,” “Where did that answer come from?”
  • On-screen text: Your AI should not start from zero every time.

2. Ask The AI

  • Show the prompt: “Draft our launch post using our latest positioning, match my writing style, and avoid old claims.”
  • Show the AI needing context, then cut to MemContext.

3. Persistent User Context

  • Show a memory being saved: “User prefers concise, founder-led launch copy.”
  • Show categories: preference, fact, decision, context.
  • Show search returning the memory through a natural-language question.
  • Show feedback buttons: Helpful, Not helpful, Outdated, Wrong.
  • Show version history: old positioning replaced by updated positioning.
  • Show temporal expiry for temporary facts.
  • Show scope/project picker for separation by user, tenant, or project.
  • Optional quick visual: memory graph as a high-energy UI moment.

4. Context Vault Reveal

  • Show the Add knowledge tabs: Upload, Paste, URL, Fact.
  • Upload or add a product docs URL.
  • Show document status moving through processing.
  • Show extracted memories count.
  • Search Context Vault for product positioning or launch claims.
  • Show hybrid results: source passages plus extracted facts.
  • Show evidence citations linked back to the source document, section, page, or chunk.

5. Works Everywhere

  • Briefly show Claude/MCP, REST API, and TypeScript SDK visuals.
  • Include one short code snippet only, not a long walkthrough.
  • Show dashboard surfaces quickly: Memories, Context Vault, MCP, API keys.

6. Close

  • End on brand visuals and the product name.
  • On-screen text options:
    • Memory that evolves. Not just stores.
    • The memory and context layer for AI agents and apps.
    • Give your AI memory.

Visual Priorities

Show these clearly:

  • Memory save/search flow.
  • Feedback and version history.
  • Scope/project isolation.
  • Context Vault ingestion.
  • Hybrid workspace search.
  • Evidence citations.
  • MCP/API/SDK availability.

Do not over-focus on:

  • Long code walkthroughs.
  • Pricing tables.
  • Admin/billing screens.
  • Dense technical architecture.

Terms To Use

  • Persistent User Context
  • Context Vault
  • workspace knowledge vault
  • memory and context layer
  • trusted context
  • source-linked facts
  • evidence-backed answers
  • hybrid search
  • versioned memory
  • feedback-driven ranking
  • scope isolation
  • cross-tool memory
  • MCP, REST API, and TypeScript SDK

Internal Competitor Notes

Use these only for strategy. Do not name competitors in the public launch video.

  • Some products focus on drop-in memory for agents. MemContext should show that memory also needs human control, versioning, feedback, and a workspace knowledge layer.
  • Some products sell broad context clouds. MemContext should feel simpler and more concrete: personal memory plus Context Vault.
  • Some products focus on enterprise graph systems. MemContext should feel easier to adopt for builders and teams.
  • Some products are full agent runtimes. MemContext should be framed as infrastructure that plugs into any app, agent, model, or workflow.

What Investors Should Notice Without Being Told

Do not mention investors in the video. Let the product imply this:

  • The category is large: every AI app needs persistent context.
  • The wedge is clear: personal memory first, workspace knowledge next.
  • The product is real: dashboard, API, SDK, MCP, docs, billing, memories, workspaces, and Context Vault.
  • The trust layer is differentiated: versioning, feedback, citations, scopes, and source grounding.
  • The expansion path is natural: individual users, teams, AI apps, internal copilots, support bots, and workspace agents.

Avoid Saying

  • Do not claim benchmark wins unless MemContext has verified public benchmarks.
  • Do not claim SOC 2, HIPAA, GDPR, zero-trust, or enterprise compliance unless finalized.
  • Do not claim revenue, customer counts, usage numbers, or retention unless confirmed.
  • Do not say “best,” “most accurate,” or “fastest” unless proven.
  • Do not say MemContext replaces all RAG systems. Say it gives apps memory, document context, extracted facts, and citations.
  • Do not make the video sound like it is for investors.

Short Cutdown Version

If we need a 30-second cut:

Your AI should not start from zero every time.

MemContext gives AI persistent user context and a Context Vault for team knowledge.

Save preferences, facts, decisions, and project context. Ingest docs, URLs, files, and curated facts. Search everything with hybrid retrieval, feedback, version history, scopes, and citations.

Connect it through MCP, REST API, or the TypeScript SDK.

MemContext is memory that evolves. Not just stores.

Give your AI memory.