Trust & Security FAQ

Inqura is an investigation platform: evidence organization, AI-assisted analysis, and defensible reporting for compliance-critical work — workplace investigations, healthcare quality reviews, audits, and oversight inquiries. It is not a general-purpose AI assistant or chatbot, and it is not a case-intake/hotline system (it complements case management tools rather than replacing them). You bring the evidence and the questions; Inqura structures the inquiry, drafts analysis in which every claim cites specific evidence, and requires your documented judgment before anything becomes a finding.

In US-based AWS data centers, encrypted at rest (AES-256 via AWS KMS) and in transit (TLS 1.2+). Inqura is designed for sensitive but unclassified information — appropriate for most government and healthcare compliance work; classified work requires specialized systems.

No. Your data is processed to generate your analysis but is never used to train or improve AI models. Inqura uses Claude by Anthropic through AWS Bedrock, which does not use customer inputs for model training — your evidence is processed within the AWS environment, not sent to a model provider's training pipeline.

Beyond passwords: passkey/biometric login (WebAuthn), hardware security keys, and TOTP authenticator-app MFA are all supported, with secure session management. If MFA is enabled, password reset also requires MFA verification to prevent unauthorized account recovery.

Yes. You can export the full case — evidence, analyses, findings, and your recorded review decisions — at any time. Your work product is yours; we don't create lock-in by withholding your own file.

Yes — complete audit logging of user actions, and, most importantly, of professional judgment: every sign-off decision (agree, disagree, unsure, and your reasoning) is timestamped and recorded as a workpaper artifact, including the fact that you opened the underlying evidence before signing off. When someone later asks "did a person actually verify this?", the answer is in the file.

Only the investigator who opened it. There is no ambient, org-wide access to case content: a case is visible to its opening investigator, and on the Pro plan that investigator can invite specific colleagues into the investigation — visibility is per-case and invitation-only. Inqura staff do not access customer case content in the normal course of operations.

Yes — this is the core of the product. Every claim in an analysis cites the specific evidence item supporting it, and citations link to the source. Each finding carries a confidence level — Established, Probable, Possible, or Insufficient — computed from the evidence, and "Insufficient" is an allowed answer. Where evidence doesn't support a conclusion, the analysis says so instead of papering over the gap.

Yes, it can. Inqura is built on the same class of language models as every other AI tool, and we won't claim otherwise — treat any vendor claiming to be "hallucination-free" with skepticism. Our safeguards are structural:

• Validated methodology, not an improvised prompt. The analysis instructions behind Inqura were developed and validated using ConKurrence, our statistical validation tool — multiple independent AI model instances rate the same evidence and their agreement is measured with the same inter-rater reliability statistics used in research (Fleiss' kappa, Kendall's W) — with outputs confirmed against expected results before deployment.
• The analysis grades itself before you see it. Every analysis is independently scored for faithfulness (are claims actually supported by your evidence?) and coverage (was the relevant evidence considered?).
• Every statement is referenced. Claims cite evidence; unsupported claims get flagged, not smoothed over.
• Sign-off is gated. You cannot agree with a finding until you've opened the cited evidence. Your judgment — not the model's — is what goes on the record.
• PHI/PII guardrails. AI content filters and PHI/PII detection run via AWS Bedrock Guardrails.

(Longer version: Why Our AI Won't Let You Agree With It.)

AI-assisted analysis is the core of the product, so there's no "AI-free mode" — but no AI output becomes a finding without your documented review, and much of the workflow isn't AI generation at all (evidence organization and metadata, the evidence-readiness assessment, your notes and determinations). If your counsel's concern is "AI making findings," it's structurally addressed: the AI does not make determinations — the system cannot record a finding as accepted without a human's recorded decision.

We word this carefully, because the difference matters:

• Infrastructure: Inqura is built on FedRAMP-authorized AWS services. That is a statement about our infrastructure provider — Inqura itself does not hold independent FedRAMP authorization. (For government buyers who need to stay inside an authorized boundary: the Enterprise licensed edition deploys inside your own AWS environment, including GovCloud.)
• SOC 2: Architecture designed with SOC 2 readiness in mind; formal compliance documentation is in progress. We do not currently hold a SOC 2 report.
• HIPAA: HIPAA-ready architecture with PHI detection guardrails. A Business Associate Agreement (BAA) is available — and Inqura itself operates under a BAA with AWS, so the commitment runs through the full stack.

If a vendor's security page won't tell you the difference between "runs on authorized infrastructure" and "is certified," ask them why.

Inqura is built by JE Vectors LLC, founded by a 25+ year veteran of federal oversight and healthcare — military officer, oversight lead, and product owner for software deployed across government agencies. It exists because he needed it and it didn't exist.

• Trial — free for 14 days: full Pro feature access, 15 AI analyses (total), 2 GB storage. No credit card required.
• Core — $99/month: solo plan, 25 GB storage, generous monthly analysis allowance.
• Pro — $199/month per user: 100 GB storage, team collaboration, generous monthly analysis allowance.
• Enterprise — custom pricing based on organization size: unlimited storage and analyses, SSO/SAML, and a licensed single-tenant edition that deploys inside your own AWS boundary.

Paid plans include a generous monthly analysis allowance. Your billing page always shows where you stand, and you’ll see a warning as you approach it. An analysis already running always finishes; if you reach the allowance, new analyses wait until your next monthly cycle. Public-interest oversight organizations may qualify for discounted first-year pricing.

Yes — arguably especially. Two things carry a first-time or "additional duty" investigator:

1. The methodology is built in. Evidence evaluation criteria drawn from professional standards (CIGIE, GAO Yellow Book, IIA, ACFE, and others), gap flagging, confidence levels, and a review gate that makes the diligence step unskippable. You still make every determination; the system makes it hard to skip the steps an experienced investigator wouldn't skip.
2. The AI Guide walks you through the entire process. An in-app guide provides contextual help at every step — what to do next, how the methodology works, and why — so you're never staring at a blank screen wondering what a professional would do here.

Check with your legal/IT team before uploading sensitive employee data — that's true of any third-party platform, and we'd rather you ask.