Practice profile

Veteran-led consulting; federal VA disability ratings, appeals, and veteran representation

Core pain

Extreme page counts, spreadsheet-era workflows, need for defensible QA as headcount grew

Technology

Superinsight for structured medical and evidence review; manual validation on closed matters before trust

Evidence grade

Self-reported metrics and hypotheses; suitable for operational learning, not financial due diligence

Executive summary

This organization grew from after-hours, founder-led work into a multi-advocate team with a sharply higher matter load. Along the way, leadership kept a disciplined rule: do not harm a veteran’s existing rating while pursuing improvement. That conservative posture made record discipline non-negotiable: every recommendation had to trace to evidence.

They described early workflows as nights in Excel, correlating conditions, injuries, and exhibits across federal and private records. At the extreme, one matter involved on the order of 6,000 pages reviewed by hand. Fatigue at that scale does not only slow people down; it creates missed connections between symptoms, service history, and treatment gaps.

Superinsight entered as a force multiplier on the medical-record layer: faster first passes, structured outputs, and a way to back-test the tool against work the team already trusted. The honest financial story in the source material is not a guaranteed ROI slide. It is a hypothesis under measurement: whether subscription cost is recovered through more matters, deeper review, and surfacing issues such as clear and unmistakable error-type patterns.

Operating context: why “facts first” mattered

Leadership framed advocacy as playing with the team on the court: you work with the evidence you have, not the hypothetical file you wish existed. Veterans often lack contemporaneous service treatment records; strategy therefore leans on longitudinal civilian treatment, lay statements where appropriate, and careful sequencing of filings.

That philosophy placed enormous weight on reading the chart correctly the first time. When the same leader described Superinsight, the through-line was not automation for its own sake. It was robustness: fewer overlooked facts compared to manual review alone.

Challenge in depth

Volume and cognitive limits

At four-digit and five-digit page counts, even strong reviewers hit diminishing returns. The advocate described manually reviewing on the order of 6,000 pages of records in at least one stretch and giving each page about 60 seconds when working manually, paired with an Excel-based condition list. That is the scale Superinsight was meant to compress, not a median for every file.

Scaling without losing the thread

Growth also showed up in how the team used the vendor: from a single primary contact to a double-digit number of team members reaching out within about a year or two. Scaling that fast introduces variance: different reviewers, different shortcuts, different comfort with dense PDFs.

Succession and dependency risk

Leadership also discussed a worry familiar to founder-led shops: what happens to clients if something happens to me? Tools that encode repeatable review steps are one small part of succession planning, but they matter when knowledge has historically lived in spreadsheets and individual memory.

Solution design

The practice did not treat AI as a black box that replaces judgment. They described a trust-but-verify pattern: running completed matters through the pipeline to compare outputs against work they already stood behind, then widening use as confidence accumulated.

Superinsight’s role in that architecture is intentionally narrow and high leverage: ingest voluminous medical material, produce chronologies and structured views, and highlight gaps or patterns worth advocate attention. Final strategy, client communication, and filing decisions stayed with humans.

Implementation notes (what others can copy)

1. Anchor on QA, not novelty. Back-test on closed files before you bet a live rating on new output.
2. Keep the “no harm” rule visible. When incentives push speed, explicit guardrails on existing benefits reduce regret.
3. Instrument the business question you actually care about. Here, leadership said they were tracking whether the subscription pays for itself via throughput and issue discovery, not vanity usage stats.
4. Train the team on what “good” output looks like. Scaling headcount only helps if everyone agrees which details matter in a rating narrative.

Numbers and quantities in this profile

The table lists figures or durations attributed to this account. Treat them as self-reported context, not audited benchmarks.

Topic	As described	Role / note
Manual file size	About 6,000 pages of records reviewed manually (“not fun”)	Advocate
Manual pacing	About 60 seconds per page when working the old way; skipped a page if nothing recognizable in that window	Advocate
Team growth	Double-digit people on the team now contacting the vendor vs. one primary contact a year or two earlier	Practice growth (as reported)
Quarterly review	Once a quarter, deeper internal check on whether the tool is doing what they expect	Advocate
Economics	Hypothesis that subscription pays for itself via more matters, depth, and more clear and unmistakable error (CUE)-type work	Advocate; not audited
Military career (context)	24 years in service; enlisted at 17, retired at 43	Advocate (biographical)

Takeaways for disability and veterans practices

• Insight extraction: The credible benefit in this account is finding lines in the record humans skim past, not replacing advocates.
• Time: Expect first-draft compression on chronology and structure; do not expect zero attorney minutes on verification.
• Cost: Treat the platform as a tracked line item tied to hypotheses you can defend to yourself and partners.
• Risk: If your shop cannot describe its QA step in one sentence, fix that before you add software.