We use essential cookies to keep the site working. We’d also like to use analytics cookies to understand which pages are useful — only if you’re comfortable. Your choice, revocable at any time from the footer.
What it is
Audit trails record everything. InkTrail records what mattered. The saves you kept. The AI you consulted. The suggestions you accepted or declined. The drafts you finalized. Autosaves, cursor movement, and routine formatting stay noise — because a record of everything is a record of nothing.
Every entry is cryptographically linked to the one before it. Change any past event and the chain breaks — visibly, to anyone verifying. There is no silent rewrite. Anyone with the link can check the record at a public endpoint; no RedInkAI account required, no trust in us required.
Because “trust me” is not a governance answer.
What InkTrail records
Meaningful moves only. The decisions that shaped the piece. Not the keystrokes.
Every draft you keep is on the record with its content hash, timestamp, and author. Autosaves and cursor movement stay noise.
The prompt you sent, the guidance it returned, and which model answered. Not for surveillance — so a reader can see the questions that shaped the work.
Adopt, dismiss, or edit-before-adopt on any Redline note. The choice is on the record with the exact suggestion it acted on.
When another editor or collaborator returns your draft, the round trip is recorded — who touched what, when, and what changed.
The moment you mark a version as final, the whole chain up to that point becomes independently verifiable. Anyone with the link can check it.
If a suggestion violated a constraint you established — no new characters in the final act, no profanity, no touching Chapter 12 — the breach is on the record alongside the rule.
Why the two together
An audit trail on a weak draft is a receipt for a piece that didn’t deserve the ink. A stronger draft nobody believes you wrote is a defense you can’t give. You need both.
Redline is the reason a writer stays. InkTrail is the reason a reviewer trusts. Together, they are the moat.
The contract
Autosaves, cursor movement, routine formatting, and panel opens are noise. InkTrail is a record of the decisions that shaped the work, not a keystroke log.
When you accept a suggestion, the AI stays credited as the origin of the words. Your authorization is the accountable event. The two are structurally distinct.
Every entry is cryptographically linked to the one before it. Alter any past event and the chain breaks — visibly, to anyone verifying. There is no silent rewrite.
Anyone with the link can verify the record at a public endpoint. No RedInkAI account required. No trust in us required. The math checks itself.
For the technical reader: how the hash-chain is constructed, how deterministic classification decides what qualifies as an accountable event, and how independent verification works end-to-end.
Read the architecture deep-dive