Help Center: Digital Twins

Digital Twins is Native AI's generative AI feature that lets you run focused conversations with digital representations of your customers and your competitors' customers. Unlike a general-purpose chatbot, Digital Twins are built from customers' real words, so they respond as if you were speaking to a live customer panel.

The Digital Twins home page organizes your work into Folders, Threads (each thread is a conversation with an audience), and Audiences (saved audience definitions). Suggested questions on the right offer one-click starting points. Jump to section:

• Asking Questions (Simulation)
• Choosing Your Audience
• Synthetic Output Slider
• Attaching Images
• Threads & Follow-ups
• Question Tips

Asking Questions (Simulation)

Simulation is where you ask questions. Type into the prompt bar (or start from an example prompt: Feedback, Hypothetical, Motivation, or Comparison), pick an audience, set the Synthetic Output slider, and hit Ask Questions. You can stack multiple questions in one thread with + Add another question. Each is asked separately, which works far better than one compound question.

When you ask a group of Digital Twins a question, the system asks each one individually and returns a summary of all answers. Shortly after you ask, the platform polls a random sample of your audience to generate a preview response. Note: while the sample is random, it is not large enough to predict the final result with high accuracy.

Choosing Your Audience

Pick a saved audience from the dropdown next to the prompt bar. Audiences are created from filters: build one in the Conversation tab's filter panel and hit Save Audience, or manage them from the Audiences tab on the Digital Twins home page.

There is a tradeoff between large and small audiences: a larger audience produces a more robust answer, but takes longer to complete. Use filters to define exactly who you want to poll before asking.

Synthetic Output Slider

The Synthetic Output slider controls how strictly answers stay grounded in your source data:

• High Fidelity - Digital Twins answer only when the source data contains a precise match for your question. You'll see exactly what's in your data, but answers will be sparser; use a large audience with this setting.
• Balanced - Digital Twins may fill in some plausible details that don't appear verbatim in the source data, but only when the response is predicted with high accuracy.
• High Creativity - Digital Twins may invent details beyond the source data. This is most valuable for creative use cases like scenario building. Fewer twins are needed to get responses, but treat the output as inspiration to validate rather than fact.

Attaching Images

The paperclip on the prompt bar lets you attach up to 5 images to a question, so you can get Digital Twin reactions to visual concepts: packaging comparisons, label variants, ad creative, and more. The twins consider the images alongside your text. If an attached image can't be processed, the question won't run with a partial set, so you never receive an answer based on incomplete visuals.

Threads & Follow-ups

Every question lives in a thread. The Threads tab lists them all with status, dates, and audience size; open one to see the full conversation. Inside a thread, the Follow up Question bar lets you keep the conversation going with the same audience: drill into a surprising answer, ask "why," or explore a related angle. Threads can be organized into folders and shared with your team using the per-thread Share toggle.

Answers come back as a structured summary of what the audience said, with the audience size, output setting, and timestamps right in the thread header.

Question Tips

Whether your twins are built from robust data (e.g. interview transcripts) or sparse data (e.g. individual product reviews), how you phrase the question matters: the more specific it is, the fewer twins will be able to participate.

• Ask questions the data set can realistically answer.
  • Correct: What do you think about the flavor of this product?
  • Incorrect: Do you use your mobile phone while shopping?
• Use proper grammar.
  • Correct: What do you think about the flavor of this product?
  • Incorrect: What do you think about flavor of this product?
• Keep questions short.
  • Correct: (with an Amazon filter applied) What do you think about the taste of this product?
  • Incorrect: What do you think about the taste of the pasta you bought while shopping on Amazon?
• One question per query, with no "loaded" questions. Use + Add another question instead.
  • Correct: Q1) What do you think about the flavor of this product? Q2) What do you think about the smell of it?
  • Incorrect: What do you think about the flavor and smell of it?
• Don't over-specify. Use filters instead.
  • Correct: What health factor do you pay attention to when buying it?
  • Incorrect: What health factor do you pay attention to when buying our gluten-free macaroni with mozzarella?