Gaps & Inconsistencies

RunLLM can automatically detect gaps and inconsistencies in your data sources and flag them for you. These inconsistencies are surfaced for you automatically on the RunLLM UI, and you can tell RunLLM which data source is correct or why the data sources are not contradictory. RunLLM learns based on this information, and if your documentation is ever incorrect, it will automatically suggest an update to your documentation.

Here's how it works:

1. When a question is answered, RunLLM will look at the set of documents that were used to answer that question, and it will check if any of those documents have contradictory or inconsistent information.
2. Those inconsistencies are surfaced on the RunLLM admin dashboard for you. You can select which one of the sources is correct or explain why these data sources are not contradictory.
3. If one data source is marked as correct, RunLLM will forget the other data source; it will not be used to answer questions in the future. If you select "Keep Both", then the explanation for why they are not contradictory will be used to disambiguate the sources in the future.

Suggesting Documentation Improvements

If you ever select a correct source that is not a part of the documentation, RunLLM will automatically determine that your documentation can be improved. Within a few hours, RunLLM will summarize the information that you told it was correct and will suggest that you improve your documentation by updating it with the generated text snippet.

You can choose to copy the generated text directly into your documentation or edit it for further clarity before using it.

Example coming soon!