Skip to main content
Workspace memory is TalkCut’s system for remembering your preferences across projects. Instead of re-explaining your brand guidelines, style preferences, and creative patterns every time you start a new project, workspace memory carries that context forward.

​
How memory works

TalkCut maintains two levels of memory:
  • User memory — Your personal creative preferences that follow you across all projects
  • Project memory — Context specific to a team workspace, shared across all members
When you interact with the AI agent, both memory levels are automatically included in the context. The agent uses this information to make generation decisions that align with your established preferences.

​
What gets remembered

Memory builds up naturally as you work. The AI agent automatically extracts and saves relevant patterns from your conversations:
  • Brand assets — Color palettes, typography preferences, visual identity elements
  • Style preferences — Preferred rendering styles, lighting approaches, compositional patterns
  • Terminology — Project-specific language, character names, location names
  • Creative patterns — Pacing preferences, aspect ratios, common compositions
  • Negative preferences — Things you’ve asked the AI to avoid

​
Examples of useful memory entries

Here are the kinds of entries that make generation more consistent:
Brand color palette: Primary navy #1B2A4A, accent coral #FF6B5A, warm neutrals. Never use pure black — use dark navy instead.
Video style: 24fps cinematic feel. Prefer slow, deliberate camera movements. Avoid handheld or shaky-cam.
Character “Kai”: Male, mid-30s, short dark hair, olive skin, always wears a grey wool coat. Expressions are subtle — avoid exaggerated emotion.
Avoid: Do not use lens flare. No Dutch angles. No purple or magenta in the color palette.

​
Managing memory

You can view and edit your workspace memory at any time. Memory is organized as files — you can:
  • Browse all memory entries
  • Edit individual entries to correct or refine preferences
  • Delete entries that are no longer relevant
  • Organize entries into folders for different aspects of your brand
This gives you direct control over what the AI “knows” about your creative preferences.

​
Compounding value

The more you use TalkCut, the better your workspace memory becomes. After several projects:
  • The AI agent requires less explanation per request
  • Generated content more consistently matches your brand
  • New team members benefit from the accumulated context immediately
  • Cross-project consistency improves without manual effort
Workspace memory is stored securely within your workspace. Each workspace has its own independent memory — switching workspaces means switching to a different set of preferences.

​
Tips for building good memory

  • Be specific in early conversations — The more detail you provide in your first few projects, the faster memory builds useful context
  • Correct the AI when it’s wrong — Corrections are captured as strong preferences
  • Upload brand guidelines — Reference material analyzed by the vision engine contributes to memory
  • Review periodically — Check your memory entries occasionally to remove outdated preferences