How TextifAI solves it
How TextifAI approaches the problem
Arc starts with the manuscript. In the current preview workflow, the safest inputs are Markdown and plain text files. The ingestion flow stages the source material, detects structure, and builds a project workspace.
From there, TextifAI works toward several connected surfaces.
Canon
Canon is the structured representation of what the story currently says or implies. It is not just a summary. It is a working memory layer for the project.
Canon can include entities, aliases, relationships, claims, source references, and review state. It should help the author understand what is established, what is uncertain, and what needs attention.
VaERL
VaERL means Vault-aware Entity Resolution Layer.
Its job is to help stabilize story identity: characters, aliases, pronouns, places, factions, objects, concepts, and other narrative entities. In a novel, identity is rarely clean. Characters have titles. Places have old names. Narrators omit names. Pronouns depend on context. Drafts rename things.
VaERL exists because a useful AI assistant must know whether it is talking about the same thing, a different thing, or something unresolved.
Evidence
Evidence is what keeps the system honest.
A canon claim should not be accepted only because a model sounds plausible. The author should be able to inspect the source material behind it. Evidence turns “the AI says this” into “the system found this here, and you can decide whether it is valid.”
That is essential for trust.
Review Queue
The Review Queue is where uncertainty becomes visible.
Instead of silently converting every extracted claim into truth, TextifAI can surface items for human review: ambiguous references, possible continuity risks, uncertain entity matches, unsupported claims, or places where a model needs author confirmation.
This is one of the most important product principles: uncertainty should become workflow, not hallucination.
Graph
The graph makes narrative structure visible.
Characters connect to places, events, objects, factions, and each other. A graph view cannot replace reading the manuscript, but it can reveal patterns that are hard to see in linear text: dense relationship clusters, isolated entities, repeated references, or unexpected connections.
Entity Fiches and editor views
Entity fiches and editor views give the author a place to inspect and revise structured knowledge. The goal is not to trap the writer inside a database. The goal is to give the writer better handles for the story they are already building.
What TextifAI is not
TextifAI is not an AI that writes the novel for you.
It is not a magic continuity oracle.
It is not a guarantee that every extracted fact is correct.
It is not a replacement for the author’s judgment, taste, voice, or intent.
TextifAI is a workspace for making AI assistance more grounded, inspectable, and useful during revision. It helps organize the memory of the story so the author can work with stronger context and clearer evidence.
How this changes AI assistance
Without structured memory, an AI assistant often receives either too little context or too much undifferentiated text.
With TextifAI, the intended direction is different.
The model should receive focused, evidence-backed context:
- the relevant entities;
- their aliases and relationships;
- the source passages that support a claim;
- the review state of uncertain items;
- the chapter or scene boundaries that matter for the task;
- the author’s corrected canon where available.
That does not make the model infallible. But it makes the assistant easier to challenge, easier to correct, and less likely to invent unsupported continuity.
The point is not to replace creative work. The point is to make revision less blind.
The product promise
TextifAI’s promise is not “press a button and get a novel.”
The promise is:
- your manuscript can become a structured narrative workspace;
- your canon can be inspected instead of guessed;
- your AI context can be grounded in evidence;
- your continuity risks can become reviewable tasks;
- your creative control stays with you.
That is the problem TextifAI solves.
Where to go next
If this thesis makes sense, continue with: