Modern institutions do not lack AI tools. They lack the ability to answer, in a defensible way, one question a regulator or opposing counsel will eventually ask:

“Was the reasoning the author’s?”

RedInkAI is the layer that makes that question answerable. This brief describes how a 30-day pilot works, what your team gets, and what we ask in return.

Every AI writing tool on the market optimizes for output: fluent text, cited sources, a coherent draft. None of them preserve the cognitive process that produced the output — what the writer was shown, what they took, what they rejected, and why.

In regulated work, this gap is not academic. A clinical document, a legal brief, or a submitted research paper is accountable in ways an unrecorded chat session cannot survive. When something is challenged, the institution has:

• The final text.
• Possibly a version-control history.
• Usually no record of the AI-assisted reasoning at all.

Existing frameworks — content provenance, explainability, human-in-the-loop review — each capture something real. None of them capture the reasoning itself.

RedInkAI is a writing environment. Inside it:

• Every AI suggestion, prompt, and returned guidance is captured.
• Every accept, edit, or dismiss the writer makes is captured.
• All of it is linked into a cryptographically ordered chain called the InkTrail.
• At the end of a project, the writer exports an Authorship Packet: a self-contained file any third party can verify without contacting us.

The verification runs on the packet, not on our servers. Your institution keeps the receipt; we do not have to be trusted, still standing, or reachable for the record to hold.

Every 90 days we publish a public adversarial audit of the chain, timestamps, and access controls at /false-positive-validation. Failures are surfaced with the same weight as passes.

Emerging regulatory language across the EU AI Act, updates to good clinical practice, and the ABA Formal Opinion 512 all assume that institutions can produce a record of how AI-assisted work was reasoned through. Almost no institution can.

RedInkAI is designed with 21 CFR Part 11 readiness in mind: unique user attribution, time-ordered event logs, tamper-evident chain, and a clean export path for validation teams to inspect. We do not claim formal Part 11 compliance — that is the sponsor’s determination inside their validated environment. We provide the mapping and the evidence.

Read the honest scope of what we do and do not claim at /institutions/clinical/part-11-readiness before any conversation with your validation team.

Duration: 30 days from onboarding call.

Cohort size: up to 8 individual writers OR a single documented workflow inside your institution.

What we provide:

• Full access to RedInkAI including the Redline editor and Authorship Packet export.
• One 45-minute onboarding call for the writing team.
• Direct email access to the founder for the full pilot window.
• A written control-mapping document tailored to your regulatory framework (Part 11, GxP, ABA 512, or an academic integrity policy).

What we ask in return:

• A 30-minute exit interview with the writing team.
• Permission to reference the institution by category (e.g. “a mid-size CRO,” “a public research university”) if the pilot goes well — never by name without written consent.

No cost during the pilot. Continuation terms are discussed in the exit interview based on what the team actually used.

The pilot is designed to fail loudly if RedInkAI is not the right fit for your institution. We propose watching for these signals, honestly, at day 30:

• At least one Authorship Packet exported for a real (not test) document the team would have written anyway.
• Non-zero editorial engagement — the writers took, edited, or dismissed Redline notes in a pattern that reflects actual judgment, not idle clicks.
• The compliance officer can read the exported packet and verify the chain without calling us.
• The writers report the tool did not slow them down — or, if it did, that the tradeoff was worth it.

If any of these fail, we say so plainly and part on good terms. The trust asset we’re building is more valuable than any single pilot outcome.

If any of the above resonates, the next step is a 20-minute conversation with the founder. Not a demo call — a diagnostic conversation to determine whether RedInkAI is the right fit for your specific institutional workflow before we ask you to spend time on a pilot.

Write to redinkaistudio@redinkai.com with two lines: what your team writes, and what a review of that work looks like today. We reply within two business days.

You can read our full transparency statement at /how-we-work before writing.