Chainlocker is local AI eDiscovery: document indexing and review for legal discovery and CPA transaction services. At launch, it reads every file in a matter or engagement folder, tags each one against criteria you define with written reasoning, and builds a reviewable index, then exports it as a court-ready package that opens offline. It runs on an appliance in your office, behind your firewall: your documents never leave your network. A chain locker is where an anchor chain is flaked so a heap pays out clean; that's the job.
Your tags and instructions, per matter or engagement. Nothing is hardcoded: a contract review playbook and a diligence checklist are just different criteria lists.
Point it at the shared folder. It lists every file it understands. PDF, Office, email, images, and a dozen more.
Text extraction locally, OCR for scans. At launch, AI assigns each document a tag, written reasoning, and a confidence score. Async, and you're notified when it's done.
A person confirms every verdict. Accept by default; reject with a correction. Cheap to review at volume, and every decision is recorded as structured data.
A self-contained package: the documents, the Excel index, and an HTML viewer that works with no server and no network.
Chat-with-your-files tools are built for questions: you ask, and they answer from the documents. Chainlocker is built for review: it works through every document against your criteria and gives you an index that a person has confirmed.
| Question | Chat-with-your-files tools | Chainlocker |
|---|---|---|
| You start with | A question | Your criteria: tags and instructions |
| What gets read | Usually the passages that answer the question | Every file it can read in the folder, each judged on its own |
| What you get | An answer in a conversation | An Excel index: a tag, written reasoning, and a confidence score for every document |
| The person's role | Reads the answer and decides what to trust | Confirms or corrects every verdict, and each decision is recorded |
| What you keep | A chat history | A self-contained package: the documents, the index, and an offline viewer |
| Best for | A quick question about a few documents | A matter or engagement folder that has to be accounted for, file by file |
Plenty of firms will use both. Ask a chat tool a question. Use Chainlocker when the whole folder has to be accounted for.
Every matter gets its own tag list and instructions. The tool adapts to the work: a document production, a deal folder, or anything shaped like "sort these documents against what matters."
PDF, Word, Excel, PowerPoint, email (.msg/.eml), text, CSV, XML, Markdown, and images including TIFF scans. Digital files are extracted directly; scans go through OCR.
At launch, every judgment carries written reasoning and a confidence score, with a minimum-confidence threshold you control. Below the floor, a document goes to review instead of being filed. No black-box tags.
The index is an Excel file your team already knows how to use, one row per document, with a clickable link straight to the file.
Corrections aren't thrown away. Every human verdict is captured as structured data, so each decision is on the record.
The export opens straight off a laptop or USB drive, no app, no internet, no absolute paths. Built for the courtroom, where wifi isn't a given.
Run your review playbook against a production the same way every time: tag documents against the matter's criteria, flag what was found and why, and hand the court an index you can defend, fully offline. Legal judgment stays with the lawyer.
Chainlocker for law firms →Transaction services and due diligence: point it at the deal folder and get every contract, statement, and side letter tagged against your diligence checklist, with reasoning attached.
Chainlocker for CPA firms →Forensic accounting, internal investigations, insurance claims. The criteria are yours per engagement, so anything shaped like "sort these documents against what matters" fits.
Tell us about the pile →As delivered to firms at launch, the appliance is installed at your office, on your network, behind your firewall, and is reached from a browser on your LAN or VPN. AI runs on a Chainlocker device delivered to your office alongside the appliance. All processing happens on site: text extraction, OCR, and the AI that tags every document. No third-party AI APIs means no metered usage: the cost is fixed, however big the folder. You rent the appliance rather than buy it, and one monthly fee covers the hardware and the software.
Ask us about the deployment model →| What | Kept on | Leaves your network |
|---|---|---|
| Your original files | Wherever your firm keeps them | No. Chainlocker never changes, moves, or deletes them. |
| Digital files: Word, Excel, email, text-based PDFs | The appliance, as uploaded copies | No. Text is extracted on the appliance. |
| Scanned PDFs and images | The appliance, as uploaded copies | No. OCR runs on site. |
| Each document's text and file name | The appliance | No. Tagging runs on site. |
| Your criteria: tags and instructions | The appliance | No. |
| Tag, written reasoning, confidence | The appliance | No. Produced by the AI on site. |
| Review decisions and corrections | The appliance | No. They are not used to train models anywhere else. |
| Excel index and export package | The appliance, then wherever you take it | Only when you take it. |
| Updates, support and health checks | Not applicable | Yes, over an encrypted connection, with no document content. |
Radtke Fournier CPAs is an example firm built on example data. The live product reads documents on your own premises.
Pick a process to see example documents beside the result, or open the Review tab and press J and K to move between documents.
We're targeting availability by late 2026. Chainlocker ships once it has survived real matters, not before. If your firm drowns in document folders, tell us what the pile looks like and we'll tell you honestly whether this tool will help, and when.