How We Use AI

A plain-language guide to the AI behind Analyze My Narrative

What AI Does

When you upload a manuscript and run an analysis, your text is sent to a large language model (LLM) that reads it and extracts structured information. This is the same kind of technology behind ChatGPT and Claude. Specifically, AI powers:

  • Knowledge graph generation — Identifying characters, relationships, plot beats, themes, conflicts, emotional arcs, and more from your text
  • Entity extraction — Recognizing and profiling characters, locations, items, and concepts
  • RAG chat — Answering your questions about your manuscript by finding relevant passages and reasoning over them
  • Image and video generation — Creating visual representations of scenes, characters, and settings
  • Chunking — Some advanced methods use AI to find the best places to split your text for analysis

What AI Doesn't Do

  • Write or rewrite your manuscript — We analyze your text; we never generate new creative content in your voice or modify your words
  • Share your work — Your manuscript is never shown to other users, used to train AI models, or made publicly accessible
  • Make permanent judgments — Every analysis can be re-run, and you can edit, override, or delete any result
  • Replace human expertise — AI analysis is a tool for exploration and discovery, not a substitute for an editor, agent, or beta reader

Limitations You Should Know

AI is powerful but imperfect. Here are the limitations we want you to be aware of:

Hallucination

LLMs can occasionally generate information that sounds plausible but isn't actually in your text. This is most likely in RAG chat responses. If a chat answer seems off, check the source passages it references — they're provided so you can verify.

Missed or misidentified elements

Character extraction may miss minor characters or identify non-characters (e.g., chapter titles) as people. Theme analysis may surface connections that are a stretch. Re-running analysis or adjusting your chunking settings often improves results.

Cultural and genre bias

LLMs are trained on broad datasets that may not equally represent all literary traditions, genres, or cultural contexts. Analysis of non-Western narratives, experimental fiction, or niche genres may be less nuanced than analysis of mainstream Western literary fiction.

Long document limitations

Very long manuscripts (500K+ words) are analyzed in chunks. While we preserve cross-chunk context where possible, the AI processes sections independently and may miss connections that span large distances in the text.

Visual generation accuracy

AI-generated images and videos are creative interpretations, not precise depictions. Character appearances, settings, and scenes may not match your exact vision. Use entity profiles and reference images to guide consistency.

Which AI Providers We Use

We use established AI providers with commercial-grade privacy protections:

  • OpenAI — Text analysis and embeddings. API data is not used for model training.
  • Anthropic — Text analysis (user-selectable). API data is not used for model training under commercial terms.
  • Google (Vertex AI) — Image and video generation. Zero data retention by default.

You can choose your preferred analysis provider in Settings.

You're Always in Control

  • You choose when to run analyses — nothing happens automatically after upload
  • You can review, edit, or delete any AI-generated result
  • You can purge your manuscript text after analysis while keeping your results
  • You can delete your entire project and all associated data at any time
  • You can export your analysis results as PDF before deleting anything

Our Approach to Responsible AI

Our AI governance practices are informed by the principles of ISO/IEC 42001, the international standard for AI management systems. This includes data minimization, human oversight of AI outputs, transparency about how AI is used, and zero-retention agreements with our AI providers. For details on how we protect your data, see our Privacy Policy.

Questions?

If you have questions about how AI is used in Analyze My Narrative, reach out at:

david@analyzemynarrative.com