Co-presence for regulated evidence work.

VedAGI's Interaction Layer is a production multimodal overlay for high-stakes regulated workflows, starting with Pramana (working name) life-sciences evidence packs. It gives HEOR teams and named signers continuous, low-latency co-presence (audio, screen, and text in ~200 ms micro-turns) while the deterministic sandbox still computes every locked figure and a named human still signs. The layer never writes locked values; it only surfaces typed Findings or recommended checks that route through the existing gates. Default deployment is fully local or air-gapped (or on-prem Canadian), with hybrid mode requiring explicit customer acceptance and full markings propagation. Interaction provenance is hashed into the checkable receipt so a reviewer can see that live human review occurred without receiving raw media. This is "own your intelligence" for regulated work: the customer keeps the data and the accountability, runs the continuous presence where they choose, and still gets an auditable, fail-closed record. Data never leaves Canadian jurisdiction unless the customer explicitly accepts a hybrid boundary.

Built for market-access and HEOR directors, payer pricing leads, and procurement and security reviewers.

Pramana (working name) supports trust through continuous co-presence while a sandbox recalculates figures from source, labels estimates, and waits for a named signer. A browser-verifiable receipt records the review. A person stays in charge.

Market-access budget-impact work still starts with source figures, clearly labeled estimates and a named signer. The model never mints a locked figure. The receipt makes the signed result reviewable, including that live review took place.

Pramana (working name) is pending trademark counsel review. The name may change.

Independent research found technical errors in 39 of 41 company economic models in NICE’s 2017 appraisals (95 percent). Source: doi:10.1017/s0266462320000422.

Field audits summarized by Panko found errors in at least 86 percent of spreadsheets. Source: Panko, arXiv 0802.3457.

Target: Pramana (working name) is built to catch model errors before submission, where a peer-reviewed study found errors in 95% of company models submitted to NICE in 2017.

We plan to let your team create, edit and approve wording templates and reuse verified records with links back to their sources. The goal is faster drafting and more consistent wording.

Templates and reused answers would not create figures or turn cached text into evidence. Estimates would remain labeled, missing evidence would still trigger refusal, and a named person would still sign off before release.

This extension is planned and is not available today. It means reusing approved business information, not training an AI model. It would not involve sharing information between companies.

Pramana (working name) is designed to support 21 CFR Part 11-style controls for electronic records and signatures by locking recomputed figures only after a named signer acts, with a receipt that links the signature to the released values. Source: 21 CFR Part 11.

Pramana (working name) is designed to support FDA ALCOA data-integrity expectations by making each released figure attributable to a signer, contemporaneous with the lock, and accurate to a recomputation from the stated source workbook. Source: FDA data integrity Q&A.

Pramana (working name) is designed to support AMCP Format 5.0 economic-evidence needs by locking budget-impact figures to their source model before a dossier version leaves the company. Source: AMCP Format 5.0 (PDF).

Pramana (working name) is designed to support CMS Medicare Drug Price Negotiation information submissions by keeping evidence-pack figures locked, estimate-labeled, and receipt-backed so manufacturer data packages can be reconstructed without silent edits. Source: CMS Medicare Drug Price Negotiation guidance.

Pramana (working name) is designed to support EU HTA Regulation (EU) 2021/2282 joint-assessment evidence discipline, where 13 JCAs were already initiated in 2025, by freezing figure versions and receipts so clinical-economic numbers do not drift between national reuse cycles. Source: EIPG HTA Regulation report.