Adham's Literary Analysis
100% Private. Your novel never leaves your device.
Powered by WebLLM
ℹ️ How it works, tips & privacy
Privacy & Offline: Everything runs locally in your browser using WebLLM. Your novel is never uploaded to any server. Once the model is downloaded you can safely disconnect from the internet. Do not close this tab or let your computer sleep during analysis — the browser will pause the process.
- Upload your novel (.txt or .docx). Optionally add a glossary of world-specific terms to reduce hallucinations.
- The AI reads your novel in chunks (~6,000 characters) and writes a short summary of each.
- The summaries are condensed into a single master plot document.
- A literary agent persona analyses the master plot and produces each chosen output in its own dedicated AI call — so nothing gets cut off.
- An author map is generated, anchored to your comparable authors and expanded to a wider literary landscape.
💡 Suggested use
- Start with a synopsis. If you have a 1–3 page synopsis or query-letter summary, run that instead of the full novel — it's much faster, and the analysis is often just as useful for marketing purposes. Use "Pre-summarized Text" mode and paste it in directly.
- Results are a starting point, not a verdict. A 3B local model is impressive for what it is, but it will sometimes suggest authors that aren't quite right, write a blurb that needs polish, or place authors oddly on the map. Use the output as raw material to react to, not as finished copy.
- Re-run sections you want to explore. Because each section is a separate AI call, you can re-run the whole analysis with a different temperature or slightly tweaked "Additional Context" to get a fresh take. The Author Map in particular benefits from a few runs at different temperatures to see which comps keep appearing.
- Save your intermediate files. Download the Condensed Plot after the first run. On future runs you can paste it straight in as "Condensed Plot" mode and skip the long summarisation step entirely.
Skipping steps: If you have already run the pipeline, use "Pre-summarized Text" or "Condensed Plot" below to jump straight to the analysis without re-reading the novel. You can also select "Chat Only" to talk directly to the LLM without analyzing any text.
Upload your full novel file below. The AI will read and summarize it in chunks.
Drag & Drop your novel here
Supported: .txt, .docx
Selected File:
Drop .txt/.docx file or click to browse
Loaded:
Controls how focused vs. creative the chat replies are. 0.6 is balanced.
Drop Glossary (.txt, .docx) or click to browse
Format one term per line, e.g. Radiant: A winged alien species
Glossary Loaded:
Leave blank to use 6,000-character chunks with overlap.
Initializing AI Engine...
Downloading weights (This only happens the first time)
The Condensed Plot is the final compact version the AI used for analysis. The Pre-Summary is the longer joined chunk-summaries before condensation — equivalent to the abridged novel.
Discuss your book, blurbs, and genres with the agent.