Great question! The answer depends on the AI model.
Some AI models utilize live web search (also known as search augmentation or RAG), while others rely entirely on static training data from months prior.
While in the future, all AI Models will include live search, for now, customers are still using some models that reference their own foundational data.
Why It Matters
Gumshoe helps you test how both types of models see your brand:
- If youâre focused on fast impact, pay attention to reports from models with citations.
- If youâre considering long-term influence and influencing "static" models, youâll need a sustained content presence to appear when theyâre next retrained.
In short, Gumshoe provides a comprehensive view of what all the models are saying, no matter how they collect their data.
Models That Use Live Web Data
These models issue real-time search queries and pull current information from the web. Youâll know theyâre doing this if your Gumshoe report includes citations. Citations are direct clues that the model searched the internet for its answer.
- These models reflect recent content updates
- Theyâre more responsive to changes you make on your website
- Theyâre useful when you want to test visibility right now
Examples include:
- Perplexity/Sonar
- Some Google Gemini Models
- Some OpenAI/ChatGPT Models
Models That Use Static Training Data
Other models donât use the internet when answering. They rely on the data they were trained on, which usually has a cutoff date (e.g., October 2023 or March 2025). There is, however, likely some training that the models undergo through their interactions with customers. The AI Models do not share when their "static" models will be updated; it could happen at any time. Quickly optimizing your brand for AI visibility and maintaining it regularly will ensure readiness for any future updates.
- These models donât cite sources
- They wonât reflect recent content changes until the next big update for the model
- Many users are interacting with these static models