Tatvam™
tatvam - essence, the underlying principle of a thing · formerly Pragna AI
The whole failure-mode workflow in one control room, auditable end to end.
At a glance
- Category
- Design Risk & DFMEA
- Theme
- Design Risk & FMEA
- Industries
- Healthcare, Manufacturing, Automotive
- Capabilities
- 6 composed
- Pairs with
- Parivart
Who it is for
- Medical device risk and quality teams under ISO 14971
- Design assurance groups running DFMEA across programmes
- Regulatory teams preparing submission evidence
Systems it sits on
PLM systems · ALM and requirements tools · Jira · Test management platforms
Also used in
Pairs with
Parivart
Explainer
Tatvam in two minutes
A narrated walkthrough of the problem, how Tatvam works, and what changes. Captions are available in the player.
Tatvam is an AI-assisted DFMEA platform built for medical device and regulated product risk analysis. It brings the full failure-mode workflow into one control room: a product hierarchy tree, a working table for scoring risk, AI-generated failure-mode suggestions an engineer accepts or refines, a risk register comparing initial against residual RPN, and end-to-end traceability from requirements through functions and failure modes to tests. A traditionally spreadsheet-heavy compliance process becomes fast, structured, and auditable.
The problem
DFMEA lives in spreadsheets, so risk scoring is inconsistent, failure modes are missed, and requirements-to-test traceability has to be reconstructed before every audit.
The outcome
A structured, auditable digital risk workflow from product hierarchy through failure modes, scoring, and verification test coverage.
How it works
How Tatvam works
- 01
Model the product
The hierarchy of systems, subsystems, and components is captured as the spine of the analysis.
- 02
Discover failure modes
Candidate modes are suggested from design context; the engineer stays the decision-maker.
- 03
Score the risk
Severity, Occurrence, and Detection are applied consistently, producing comparable RPN across the programme.
- 04
Trace to verification
Every requirement resolves to the failure modes it guards and the tests that verify them.
Module 01
Product Hierarchy
The system, subsystem, and component tree that risk analysis hangs from.
- 02
DFMEA Working Table
Severity, Occurrence, Detection, and RPN scored consistently across the team.
- 03
AI Failure-Mode Discovery
Candidate failure modes suggested from design context for an engineer to accept, refine, or reject.
- 04
Requirements Traceability
Requirements to functions to failure modes to verification tests, held as one chain.
- RPN compared in a sortable risk register
- Initial vs residualRPN compared in a sortable risk register
- traceability from requirement to test
- Audit-readytraceability from requirement to test
Integrations
Ideal for
- Medical device risk and quality teams under ISO 14971
- Design assurance groups running DFMEA across programmes
- Regulatory teams preparing submission evidence
- Engineering leads managing risk impact on design changes
Capabilities
Tatvam composes these capabilities from the shared Agent Platform layer. The same atoms power the rest of the portfolio, so adopting a second accelerator reuses the integration work rather than repeating it.
Composition
Tatvam is assembled from shared platform atoms. Hover a partner to see what you reuse on the next deployment.
- Document Intelligence
- Structure from unstructured sources: regulatory bulletins, specifications, contracts, claims, and reports.
- Risk Scoring
- Severity, likelihood, and exposure quantified consistently so prioritization survives an audit.
- Generative Design
- Designs, tests, and code generated from stated intent and checked against constraints before a human reviews.
- Evidence & Audit Trail
- Every input, decision, and action captured as tamper-evident evidence a regulator or auditor can follow.
- Policy & Compliance
- Obligations from standards and regulations mapped onto live workflows and continuously monitored.
- Human Approval
- Explicit approval boundaries: anything consequential waits for a person, and the wait is recorded.
Ingest
Reason
Reason
Govern
Govern
Govern
FAQ
Does the AI decide the risk score?
No. It proposes candidate failure modes and supporting context; scoring and acceptance remain with the engineer, and every decision is recorded.
Is it only for medical devices?
It was built for medical device risk analysis under ISO 14971, and the same workflow applies to automotive and industrial DFMEA.
How does it help at audit?
The requirements-to-test chain is maintained continuously, so evidence is queried rather than reconstructed.