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

Company

AI notice

Effective 11 October 2026. Decosa, Inc., a Delaware corporation.

Decosa makes tools that draft, check and explain work with AI models. This page says plainly what that means for you.

The short version

  • The tools, the API and the site guide are AI models, not people.
  • Their output is a draft to check, not advice. A qualified person decides.
  • Each tool says which model it runs. A signed record shows a run wasn't changed afterwards; it doesn't show the answer is right.
  • Studio media carries a content credential that marks it as made with AI.

1. You are using AI

When you run a tool, call the API or chat with the site guide, an AI model produces the answer. No person reviews it before you see it. When a person from our team replies to you, for example to a request or an email, they say so.

2. Not advice

Decosa's output can be wrong, incomplete or out of date, even when it sounds sure. It is not professional advice, and using Decosa doesn't create a professional relationship with us. Treat every result as a draft: check it against the sources it cites, and have a qualified person make the decision.

3. Health

  • The clinical tools draft and check paperwork for clinicians. They are not medical advice, they don't diagnose or treat, and they are not medical devices.
  • A licensed clinician must review anything that goes into a patient's record or care.
  • The hosted tools are demos for sample data. Don't enter real patient information: run the tool on your own hardware for that.
  • In an emergency, call 911 or your local emergency number.

4. Legal

  • The legal tools help lawyers and legal teams work faster. They are not legal advice, and Decosa is not a law firm.
  • Check every citation and quotation against the source before you rely on it or file it. A lawyer is responsible for the work product.

5. Money, tax and accounts

Tools that read filings, books, tariffs or tax figures help you check numbers. They are not financial, investment, tax or accounting advice. Have a qualified professional confirm anything you file or act on.

6. Hiring and HR

The HR tools organise records and draft documents. They must not decide on their own who is hired, promoted, disciplined or let go. A person makes those decisions and is responsible for following employment law. See the acceptable use policy.

7. Which models, and who serves them

  • Each tool's page names the model or models it runs. Our accuracy measurements, with their dates, are on metrics.
  • Hosted model calls are served by NEAR AI through OpenRouter, with Reka AI as the only fallback, under Decosa's account, with zero data retention required. The companies are listed in our privacy policy and on where your data goes. The signed record of a call names the model that answered and who served it.
  • If our own model is unavailable, the site guide may be answered by Anthropic's model instead.
  • We don't train models on your content.

8. What a signed record proves

Metered runs leave a record signed with Decosa's key (Ed25519): which model answered, when, how much was used, and hashes of what went in and came out. Anyone can check one; verify a run shows how.

  • It proves the record wasn't changed after it was signed, and which key signed it.
  • It doesn't prove the model actually ran as described, that the output is correct or complete, or that it suits your purpose.

9. Studio media

Each picture, track and clip made in Studio carries a C2PA content credential and a signed render record naming the model, its licence and the settings. Please keep them attached when you share the file.

10. Known limits

  • Models can invent facts, citations or quotes, miss things, and repeat biases in the data they learned from.
  • They can misread poor scans, handwriting, accents or unusual layouts.
  • Tools marked Labs or preview are early. Tell us when a tool gets something wrong: support@decosa.ai.