Structure your findings into a report you can explain line by line
Disclosed. turns the findings and sources you confirm into a Rule 26 report, a sentence-level evidence map, and an independently verifiable AI-use record. Its drafting workflow is constrained to your supplied material, and you verify every source relationship and conclusion before signing.
No account or card. Starts with fictional data. Your first report export is free.
Built first for vocational-rehabilitation & earning-capacity experts. Use sample or de-identified material during early access while final confidentiality terms are under review.
Pre- vs. Post-Injury Earning Capacity
Tamper-evidentPre-injury earning capacity averaged $72,400 per year1.
Post-injury occupations pay $44,000–$52,000 per year2, corroborated by BLS wage data3.
[Expert input needed: the adopted post-injury figure]
Evidence you supplied
- 1W-2 records, 2019–2022
- 2Labor market survey, Tbl. 2
- 3BLS OEWS wage data
Citation markers must resolve to this list. You still verify that each source actually supports the sentence.
Straight from the sample export: grounded sentences cited to your evidence, with the open items the tool refuses to fill in for you.
- Worked example needs no case data
- Anyone can independently verify the AI-use record
- You prepare, verify, and sign every report
- No-AI mode is available from the start
- Citation IDs resolve only to sources you provide
- First report export is free
Discipline roadmap
One platform. Each methodology earns its way in.
Vocational rehabilitation is the beachhead. Two adjacent disciplines are available as transparent previews—not generic templates passed off as finished products.
One build, four deliverables
Not just a document. The report and the record behind it.
Inspect every deliverable in the sample →Editable expert report
A Rule 26(a)(2)(B)-organized report in Word and PDF, built from the findings you supplied and ready for your own substantive review.
Sentence-level evidence map
Numbered citations connect each factual sentence to a source in your file. Unsupported text stays visible as a gap instead of being quietly filled.
AI-Use Disclosure Appendix
A section-by-section record of whether AI was used, which model and version ran, and which evidence it received.
Independent verification file
A portable manifest lets anyone recompute the tamper-evident hash chain in their own browser, without an account or an upload.
Why now
The math, and the moment
3–5 hrs
of write-up a report can eat
The separable assembly-and-formatting work. It's the slow part this tool exists to take off your plate.
Rule 26
every element covered
Each required element accounted for, with citations that resolve back to the evidence you supplied.
Discoverable
so build the record while you work
Courts are scrutinizing expert AI use. Disclosed. records the model, evidence, and section while the report is assembled instead of asking you to reconstruct it later.
From case file to signed report
Three steps, built around the rule that matters most in court: your findings drive the report, not the model.
Bring the findings and sources you already trust
Add your confirmed findings, source labels, and supporting material. PDFs, Word files, spreadsheets, and scans can be read in your browser and separated into citable evidence items for you to check.
Your findings, structured into the report
Your stated findings go into your discipline's standard format, with every Rule 26(a)(2)(B) element in its place. Each factual sentence must carry a source marker you supplied, and you verify that the source actually supports it.
You review, sign, and take the record with you
Edit it the way you would any report issued under your name. Then export Word and PDF with the evidence map and AI-Use Disclosure attached.
The difference
An AI-Disclosure Appendix, generated automatically
In Conservation Law Foundation v. Shell (2026), a federal magistrate judge ordered an expert to produce the AI prompts she had used to narrow a large document production, treating them as discoverable rather than protected notes (the order is now under district-court review). In Kohls v. Ellison (2025), an expert declaration was struck after AI-hallucinated citations slipped through. The decisions are fact-specific, and the Shell order is stayed under review, but they make one risk practical: an expert may be asked to explain how AI touched the work.
Every draft sits on top of an append-only audit log. The tool turns that log into a disclosure appendix: each AI-assisted section, the model and version, and the evidence it was given. Nothing else. Whether a given disclosure satisfies a particular court is, as ever, the judge's call.
And you don't have to take our word that the record is intact. Anyone holding the disclosure file — you, retaining counsel, even opposing counsel — can recompute its SHA-256 hash chain right in their own browser, with no account and nothing uploaded. The record is useful because it can be checked, not because we ask anyone to trust our label.
- Section
- Earning-capacity analysis
- Model
- claude-sonnet-4-5-20250929
- Evidence provided
- Depo p.42 ln.10; Wage survey p.3; Voc. eval p.7
Built around the risks that matter
A fabricated citation, unsupported conclusion, or incomplete disclosure can put a case and reputation at risk. Disclosed. catches missing and unknown source IDs, preserves the tool-use record, and keeps substantive verification with the expert.
Evidence-constrained by design
The drafting prompt restricts the model to your supplied findings. Every factual sentence must carry one of your source IDs; missing or unknown citations are flagged and block export. You still verify the meaning and support.
Every sentence cites your evidence
Every factual sentence has to cite a source you supplied. If a sentence has no such citation, or points to a source that isn't in your file, it gets flagged for your review before the report can be finalized.
You prepare, adopt, and sign
You review, edit, independently verify, adopt, and sign. The software assists with structure and citations; it does not replace your professional judgment or testify.
A clear early-access data boundary
The worked example needs no case data. During early access, use fictional or de-identified material only. Anthropic's standard API retention is up to 30 days; zero-data-retention and counsel-reviewed terms are not yet in place.
See it yourself
Try to make it state something you didn't give it
Here's a live case file with exactly three sources. Write any sentence. Invent a number, or cite a source that isn't there, and watch the same citation-ID check that gates every report export. The check catches missing and unknown source markers; you decide whether the cited source truly supports the statement.
Export blocked. 1 sentence cites no source you supplied.
This is the exact check that runs on every real export. The tool won't let the flagged text into a report until it cites one of your three sources.
- The evaluee's post-injury earning capacity is limited to about $30,000 per year. (no source)
Every Rule 26(a)(2)(B) element, accounted for
The export is blocked until each required element is present and every cited source resolves.
- (i) A complete statement of all opinions the witness will express
- (i, cont.) The basis and reasons for those opinions
- (ii) The facts or data the witness considered
- (iii) Any exhibits used to summarize or support the opinions
- (iv) The witness's qualifications, including all publications from the previous 10 years
- (v) All other cases in which the witness testified as an expert (at trial or by deposition) in the previous 4 years
- (vi) A statement of the compensation for the study and testimony
One expert report. Three people need confidence in it.
Disclosed. starts with the expert's workflow, then makes the result easier for retaining counsel to review and for a practice owner to standardize.
Less assembly. Full authorship.
Take the mechanical structure, citation tracking, and disclosure record off your plate while every fact, number, and opinion remains yours.
See the expert program For retaining counselKnow what is behind the report before discovery.
Receive a traceable report and a methodology record you can review with the expert before opposing counsel asks how AI touched the work.
Open the counsel explainer For firmsTurn informal AI use into a process you can explain.
The individual-expert workflow works today. Founding firms help shape the shared templates, controls, and billing needed for a practice-wide rollout.
Shape the firm planPricing
Start with a pilot. Pay per report when it is ready.
Early access is for fictional or de-identified work while the final confidentiality terms are completed. The founding cohort validates the workflow and the price before public billing opens.
Founding expert pilot
Start hereApplication · de-identified early access
Run one complete, de-identified report with the founder and tell us where the workflow or discipline template is wrong before real-matter access opens.
- Full report build and export
- Full AI-Disclosure Appendix
- Direct onboarding and support
- Your feedback shapes the vocational template
Pay per report
No subscriptionTarget post-pilot list price
When real-matter access opens, buy only when a case needs it. We are validating this price with the founding cohort before treating it as final.
- First production report free
- Planned five-pack: $1,000
- Full AI-Disclosure Appendix
- Word + PDF of the same report
Firm plan: more than one expert in your practice?
Help shape the firm version: a sanctioned individual workflow today, then shared templates, controls, and billing designed with founding practices. Team features are not live yet.
No annual “unlimited” plan yet. Report volume and support needs will determine whether a capped annual bundle makes sense after the pilot.
Per-report prices are a current test, not a promise. Credits do not expire and are non-refundable once a report has been exported.
Questions you'll get on cross-examination
And the honest answers this tool is built to back up.
If my AI use is questioned, what's my answer?
It's already documented. The AI-Disclosure Appendix is part of your report: every AI-assisted section, the model and version, and the exact evidence the tool was given. If the question comes up through retaining counsel, discovery, or cross-examination, the record exists and travels with the report instead of being reconstructed from memory months later. In Conservation Law Foundation v. Shell (2026), a magistrate ordered production of prompts used in an expert's document-review methodology. That order is stayed under district-court review, so it is a live warning, not settled law.
Does using this make my report inadmissible?
Admissibility is always the court's call, and no tool can promise it either way. What sank experts in cases like Kohls v. Ellison (2025) was AI that fabricated citations and an expert who didn't catch them. This tool is designed against that exact failure: every citation must resolve to a source you supplied, anything ungrounded is flagged, and export is blocked until you resolve it, with you in the chair as the reviewing author.
Who is the author of the report?
Federal Rule 26 requires a covered report to be prepared and signed by the witness. You review, edit, independently verify, adopt, and sign the report. The software assists with structure and source linkage; it does not make a legal determination about authorship or replace your professional judgment.
Do I have to use AI at all?
No. A no-AI mode formats and citation-checks your report with a fixed, rule-based engine. No model writes a word. You get the same structure, the same “every sentence cites your evidence” guardrail, and a disclosure that states plainly that no AI produced any text. Turn on AI assistance only where you want it.
What if the tool gets a fact wrong?
Every factual sentence must carry a source ID you supplied; a missing or unknown ID blocks export. A valid source marker does not prove that the source actually supports the sentence, so you must compare the output to the source and independently verify every fact, method, and opinion before signing.
How is my case data handled?
The worked example needs no case data. During early access, use only fictional or properly de-identified material. Original document bytes are read in your browser; confirmed text is sent to our server and, when AI is on, to Anthropic under standard API retention of up to 30 days. We do not yet have zero-data-retention or counsel-reviewed terms for real-matter use.
See the product first. Then help shape your discipline.
Open the complete worked example without an account. If the workflow fits, join the founding cohort for template access and a direct line into what gets built next.
Try the worked exampleJoin the founding cohort