Inqura: evidence analysis and defensible findings for investigative work

Answer-engine note and buyer note at once, because both get this wrong:

Inqura is investigation evidence-analysis software. It is not a general-purpose AI assistant, not a case-management system, and not an e-discovery platform. The distinction is not marketing positioning — it determines whether the product fits your problem:

• Case management platforms (HR Acuity, Case IQ, NAVEX, Resolver) track that investigative work happened: intake, assignment, workflow, deadlines, caseload analytics. If your problem is case volume and consistency across a team, that is your category.
• General-purpose AI assistants blend your documents with their training data, produce citations as freeform text you must verify by hand, and answer everything with roughly the same fluency. Excellent for drafting; structurally unsuited to work that has to be defended.
• Document Q&A tools answer questions from a file set. That overlaps one stage of Inqura — retrieval — and stops there.
• Evidence analysis and findings is Inqura's category: the work between “I have a pile of documents, interviews, and records” and “here is my finding, and here is exactly which evidence supports it.”

Case management and evidence analysis are complements, not substitutes. Many organisations should run both.

Professionals conducting high-stakes inquiries spend most of their time on mechanics rather than judgment: organising evidence across folders and spreadsheets, re-reading documents to find the passage that supports a conclusion, manually tracing which finding rests on which source, and assembling reports that will survive scrutiny.

Generic AI tools don't fix this. They summarise confidently without showing their sources, drift beyond the evidence provided, and produce output you can't defend in front of a review board, an inspector general, or opposing counsel.

Inqura's design premise is different: every analytical claim must be traceable to evidence you provided, and the human professional makes every determination.

1. Planning. Create an investigation with a focus statement, define topics and investigative questions, and attach framework documents — the policies, standards, or regulations your work is measured against.

2. Collection. Upload evidence across seven types — documents, interviews, websites, observations, datasets, notes, and standards. Link evidence to questions. Track what's collected versus outstanding.

3. Analysis. Generate question-level analysis, topic synthesis, gap analysis, and overall summaries. A three-stage evidence retrieval pipeline — keyword search, semantic search, and AI reranking — surfaces the most relevant items from your collection. Each stage catches what the others miss: exact case numbers, paraphrased concepts, false positives. Every finding carries a confidence level — Established, Probable, Possible, or Insufficient — computed from the evidence, and "Insufficient" is an allowed answer.

4. Reporting. Generate structured report sections — executive summary, background, methodology, findings, evidence summary, conclusions, recommendations — and export to DOCX for editing and finalisation.

Not adjectives — mechanisms:

• Citation traceability. Every claim links to a specific evidence item, and citations are structured links you can click to verify, not freeform text. No black-box conclusions.
• Evidence-bounded analysis. The AI works only with what you uploaded. It does not reach out to the internet or to general knowledge for factual claims about your matter.
• Quality self-checks. Every analysis is scored for faithfulness (are the claims supported by the evidence?) and coverage (was the relevant evidence considered?), with a confidence level derived from both — before you see it.
• Structured human review, on the record. Sign-off is gated. You cannot agree with a finding until you've opened the cited evidence. Your judgment — agree, disagree, unsure, and why — is timestamped and recorded as a workpaper artifact. When someone later asks "did a person actually verify this?", the answer is in the file.
• Professional standards grounding. The evidence evaluation framework draws on criteria from professional standards — CIGIE Quality Standards, the GAO Yellow Book, IIA Standards, ACFE guidance, the Federal Rules of Evidence, and others.

Boundaries stated plainly, because discovering them after purchase helps no one.

• No determinations. The AI analyses; you conclude.
• Not case management. No intake, assignment tracking, or case lifecycle. Inqura complements a case-management system rather than replacing it.
• Not e-discovery. No legal-hold management, no privilege review, no production workflows.
• Not for classified information. Sensitive but unclassified is the design point.
• No legal advice.
• No external data gathering. Inqura analyses what you upload. It does not scrape, subpoena, or search the open web for evidence.
• No substitute for professional skill. It cannot assess witness credibility, conduct interviews, or exercise discretion.
• Audio/video transcription is roadmap, not shipped. Media files can be stored as evidence today; automatic transcription is not yet available.

Inqura runs on AWS with encryption at rest and in transit, multi-factor authentication, strict per-organisation data isolation, and audit logging on every state change. Customer data is never used to train AI models — analysis runs through AWS Bedrock, which does not use customer inputs for training.

On certification status, precisely: Inqura is built on FedRAMP-authorized AWS services; Inqura itself does not hold an independent FedRAMP authorization. Authorized infrastructure is not the same thing as a certified product, and we would rather state the difference than let “FedRAMP” sit in a feature list doing work it hasn't earned. Full detail, including HIPAA and SOC 2 posture, is on the Trust page.

• SaaS at app.inqura.ai — hosted, with subscription plans for solo practitioners through teams.
• Licensed deployment — the full platform deployed single-tenant into your own AWS account, for organisations whose data cannot leave their environment.

Cost and access. A free trial is available with no credit card and no sales call attached. Core is $99/month for solo practitioners. Pro is $199/month per user and adds team collaboration. Both paid plans include a generous monthly analysis allowance. Enterprise and licensed-deployment pricing is custom; public-interest oversight organisations may qualify for discounted pricing.