Skip to content
decosa
decosa
Get an API keyTry a tool

Industries

LegalBriefs, depositions, discovery, records, client calls.HealthcareThe visit, then notes, audits and denials.Finance & insuranceAlerts, SARs, disputes, claims, recorded calls.HRInvestigations, leave and I-9 audits.Compliance & trustSecurity questionnaires, control evidence, AI audits.OperationsForms, shop pages, imports, listings: any industry.Life sciencesTrial numbers, safety cases, labels, papers.

More industries

Public sector 2 toolsSales and marketing 6 toolsScience and research 6 toolsMusic 5 toolsFilm, TV and games 6 toolsAll industries

Create and tools

CreateTools

Developers

Decosa APIQuickstartAPI referenceAPI pricingBuilding blocksDecosa agentAssembly promptsFor AI agentsSelf-hosting docsVerify a runAccuracyStatus

Trust

Where your data goesHosted APISelf-hostOn a MacConfidential optionDecosa for ChromeAccuracyVerify a runStatusSecurityRegulation watchCompare ways to run it

Company

PricingAboutApps by DecosaContact

98 · Legal · live

Privileged call notes

Open the toolJSON

Eval results

Scored on a held-out or test splitRun 28 Sep 2026

  • Memo fact recall (blind grader)89.3%test splitn = 20Dev: 92.2%. Grader: Claude Code Opus 5.5, blind, saw only script, ground truth and memo.
  • Memo lines invented / with a wrong detail0 / 12 of 796test splitn = 7967 of the 12 in one call (a year the call never said); the no-model year check now flags them
  • Conflicts-name recall, strict / phonetic89.6% / 94.0%test splitn = 20Strict misses are mostly speech-recognition spellings; no invented names
  • Dated deadlines found100%test splitn = 20Dates overall: 92.0%
  • Recording notice heard, right26/26syntheticn = 26All 26 calls (dev and test). 6 calls never mention recording and are flagged
  • Time entry = call length rounded up to 0.1 h26/26syntheticn = 26All 26 calls (dev and test). Narrative acceptable 20/20, client email 14/20 (blind grader)

Dataset

26 synthetic attorney-client calls (6 dev, 20 test) in 9 practice areas, 3 in Spanish, voiced with Decosa house voices (Kokoro-82M) with crosstalk and a telephone band on half; run through the diarizer and the full pipeline.

Caveats

  • Synthetic calls voiced by TTS (cleaner than real phone audio); the same model family wrote the scripts and runs the pipeline.
  • Fixes after cold-user tests and after the first test runs were mechanisms (spelled names, the client on the list, a weekday and year check), not tuned to test values; the year check was applied post-hoc to the final outputs.
  • The claim check blocks invented lines but marked only 7 of 12 lines the grader called wrong (with the year check).
  • Time savings are estimates from two blind persona tests, not measured with real lawyers.

Nightly smoke check

Loading the nightly status…

Result
pass
Run
28 Sep 2026
Latency, this run
n/a
p50 over passed runs
148 s
Receipts
44
Model calls
n/a
Tokens
n/a
Cost per run
$0.017

Self-host verification

Verified on 28 Sep 2026: fresh clone of decosa-api into a clean directory on our server, api image built from docker/api/Dockerfile, compose api service with a named data volume on the host network, direct route, local signing, against the already-running Voxtral, MOSS diarizer and Qwen3.8; torn down after

Assembly prompt §6: TX/CA gives all-party; 400 without consent; custody call upload 39 s, 11 names incl. opposing counsel, 0.1 h, audio dropped, record verified (135 entries), 41 receipts attested. Live replay at 2x: 155 s, 52 captions, record verified (238 entries).

Rehearsal bundle: privileged-call-notes.zip (3 KB, 10 checks). Mock inputs plus the expected results, so you can prove your own setup works before any real data touches it.

Known limits

  • Not a certified transcript: speech recognition mishears names and numbers. Check anything that matters against the call.
  • The conflicts list is the names heard on the call; it does not search your conflicts database.
  • Deadlines are the ones said on the call. It recomputes the arithmetic, but it does not know court rules or limitation periods.
  • The time entry is the call audio's length rounded up to 0.1 hour; your firm's and the client's billing rules decide the entry.
  • The hosted demo is for synthetic calls. On the hosted API the server sees the audio and text in memory while it works; real client calls belong on your own box or the Confidential tier.

Models and licences

Standard tier. Licence posture: permissive (Apache, MIT or BSD).

All quality evidence

Every sourced number on the tool’s Stack tab, by tier. Some are proxies from another task; their labels say so.

Lite · one 48 GB card, uploaded calls (1)
  • call memo quality: not measured yet
Standard · the hosted demo (3)
  • memo fact recall, 20 synthetic test calls: 89.3%blind grader, 20 synthetic test calls; decosa-api docs/evals/privileged-call-notes.md (28 Sep 2026)
  • conflicts-name recall, 20 synthetic test calls: 89.6% strict / 94.0% phonetic20 synthetic test calls; decosa-api docs/evals/privileged-call-notes.md (28 Sep 2026)
  • invented memo items left after the claim check: 0 of 796blind grader on the full memo; 12 items (1.5%) had a wrong detail; decosa-api docs/evals/privileged-call-notes.md (28 Sep 2026)
Best · DeepSeek-V4-Flash writes and checks (1)
  • call memo quality: not measured yet

How we measure · All tools