Operational Admissibility, at Pharmaceutical Precision.
A capabilities briefing for NATCO Pharma.
SkyEdgeAI AI Platform Private Limited
Prepared for internal discussion. Confidential. Not for external distribution or reference use without prior written consent.
The Financial Context
Revenue held. Margin didn't.
What's the difference between the two — and who controls it?
NATCO's FY26 consolidated revenue was approximately ₹4,376 crore, with net profit of about ₹1,418 crore — down roughly 25% year-on-year as the Revlimid contribution wound down. That's substantially a product-mix story, not a plant-performance one. But conversion cost and evidentiary efficiency are levers that stay inside NATCO's own control — and unlike a product event, an improvement here compounds every quarter.
These are NATCO's own publicly disclosed FY26 results, used only to frame this conversation and open to correction — not financial advice.
EBITDA margin by quarter, FY26
45.5%
Q1
46.4%
Q2
30.7%
Q3
25.1%
Q4
Recognition
What happens when your record is only as strong as who's willing to vouch for it?
Regulatory findings industry-wide are shifting in character — from procedural gaps to evidentiary ones. It's a well-documented pattern across global pharmaceutical manufacturing, not any single company's story.
The common root cause is architectural, not human: when a record's authenticity depends on trust rather than proof, even strong teams doing careful work can find their evidence questioned.
Platform Positioning
OAL sits above what you already run — and turns every decision into proof.
The Operational Admissibility Layer integrates read-only across your existing CDS, LIMS, MES, and QMS. It does not change how your teams work today.
It changes what happens the moment a decision is made — batch release, deviation closure, CAPA sign-off, environmental monitoring override — by capturing it as an immutable, timestamped record the instant it occurs.
The Question in the Room
Your systems already log activity. Is that the same as proof?
YOUR NATIVE AUDIT TRAIL
Lives inside the system that generated it
Can be reconfigured by an administrator
Trusted because the source is trusted
GUARDIANLEDGER™
Sits independently outside every source system
Cannot be unilaterally rewritten by any single system
Trusted because it's cryptographically provable
Not a replacement for Part 11 or Annex 11 controls — the layer that makes those controls independently verifiable.
Capability — Data Fabric
Seven plants. Four kinds of systems. One evidentiary record.
CDS
LIMS
MES
QMS
Data Fabric
Quality-critical data spans separate systems — chromatography, environmental monitoring, batch records, LIMS — reviewed independently and reconciled manually. Data Fabric unifies decision-relevant data into a single governed layer without displacing or duplicating any of them.
Outcome: One source of truth for every inspection, every time, across every site.
Capability — AI Analytics
Every CAPA deserves proof its root cause was found, not assumed.
97%
decision accuracy in live manufacturing deployments
Investigation quality varies under time pressure across the industry — a "most probable cause" conclusion is faster to write than an evidence-backed one, but only one holds up under later scrutiny. AI Analytics surfaces the underlying evidence trail behind every anomaly, correlating process data across systems before a conclusion is signed off.
Capability — CVS / EdgeVision™
Aseptic integrity: caught on review, or caught in the moment?
TODAY, INDUSTRY-WIDE
Environmental monitoring and aseptic process validation in Grade A/B areas typically rely on procedural and visual confirmation reviewed after the fact — a point of vulnerability common across sterile manufacturing.
WITH CVS / EDGEVISION™
Real-time, confidence-gated computer vision watches aseptic-critical zones continuously, flagging deviations from acceptance criteria the moment they occur — not on next week's video review.
Process optimization — yield improvement, cycle-time reduction, API batch parameter changes — carries real validation and compliance risk when tested directly on the line. TwinCore™ builds a live digital twin of the process, allowing scenario testing and optimization before physical implementation.
Outcome: Faster iteration without incremental validation risk.
Capability — Command & Control Layer
Five functions, one clear record — and everyone in it protected.
QA
IPQA
Micro
Production
Engineering
Recorded approval
Batch release and CAPA sign-off often depend on individual memory and manual coordination across QA, IPQA, Microbiology, Production, and Engineering. The Command & Control Layer structures decision orchestration across functions — AI is always advisory, never autonomous, on any quality-critical decision.
Outcome: Faster, better-coordinated sign-off — with a record that protects the people who made each call.
Capability — ESG & Compliance Layer
Audit-ready every day — not just inspection week.
21 CFR Part 11
EU Annex 11
CGMP
21 CFR Part 11, EU Annex 11, and CGMP audit readiness is typically assembled reactively ahead of an inspection, across multiple systems and stakeholders. The ESG & Compliance Layer maintains continuous, standing audit readiness — governance evidence generated as a byproduct of normal operations, not a special exercise.
Outcome: Inspection preparation becomes inspection confirmation — directly supporting product quality and compliance as a material issue in your own sustainability reporting.
Capability — InfraOps
Uptime, protected — across every site, not just one.
~60%
ROI within 90 days, in live deployments
A multi-site manufacturing footprint carries distributed asset-reliability risk — unplanned downtime on any one line affects the whole supply commitment. InfraOps applies predictive reliability monitoring across the operational estate, read-only and non-intrusive to safety-critical control systems.
GuardianLedger™
The record that proves itself.
An append-only, blockchain-anchored, cryptographically hash-chained decision record. Every entry is permanent — but permanence isn't rigidity. If context changes, GuardianLedger™ records a new, linked entry rather than altering the original; the history stays whole and legible, never overwritten. Commission begins before go-live.
GUARDIANLEDGER™ ENTRY
#GL-2026-081402931
EVENT
Deviation closure — Batch #B24-1187
TIMESTAMP
2026-08-14 14:32:07 IST
ACTOR
Named QA reviewer (role-verified)
HASH
a1f9…e02c → chained to prior entry
STATUS
Immutable · independently verifiable
Illustrative example — not an actual record.
Validation & Data Residency
Read-only by design. Validated on your terms.
OAL's read-only, non-modifying integration means it does not alter source-system records or their existing validated state — it observes and independently attests.
Deployment follows a right-sized IQ/OQ/PQ approach scoped to what OAL actually does, agreed with your IT and Quality teams before commission. Data residency and anchoring location are configured to your governance requirements, including applicable Indian data protection law.
Proof
97% and ~60% aren't projections. They're already running.
Two live deployments — a manufacturing campus and a public-access facility, both under NDA — produced the accuracy and ROI figures shown on the two highlighted capability slides. These are directional reference points from those specific environments, not pharma-validated benchmarks; the proposed engagement is designed to establish NATCO's own baseline and measure results against it.
Manufacturing Campus
Public-Access Facility
Both under NDA
Assurance
Three questions your CFO will ask that no other slide answers.
Who owns what gets built?
NATCO-specific configuration, data, and models built for NATCO belong to NATCO. Platform IP remains with SkyEdgeAI, licensed to NATCO through the engagement.
What if NATCO wants to stop?
Data export in open formats at any point. Engagement gates are contractual and separately scoped — nothing is open-ended.
Why not build this in-house?
A fair question, and NATCO can. Our argument is time-to-evidence, not capability scarcity — the baseline diagnostic gives your team the facts either way, on your own data.
Boundaries
The honest edge of this platform — stated before you sign, not in month five.
Not a validated GxP system.
OAL is not a replacement for your MES, LIMS, CDS, or eQMS. It reads from them; it does not become one, and does not carry their validated status.
Not autonomous control.
Nothing on this platform takes an action on a plant, quality, or safety system without a named human decision. Any proposed write-back is a separate change control.
Not a guaranteed outcome.
We publish a measurement method, not a promised percentage. Any figure quoted to NATCO's finance team comes from NATCO's own data.
Not a substitute for qualified authorship.
GuardianLedger™ preserves the evidence trail; it does not author, sign, or file a CAPA or submission — that remains with your qualified personnel.
The Path Forward
One clear ask today. Two more only if it earns them.
One facility walked; evidence architecture assessed against your CDS/LIMS/MES/QMS estate. Output belongs to NATCO regardless of whether you proceed further.
GATE 2 — Lighthouse Site
4–6 months · capex + annual platform fee
GuardianLedger™ commissioned at one facility; benefit measured against the Gate 1 baseline.
GATE 3 — Network Rollout
rolling · per-facility licence + managed service
Template applied site by site, on NATCO's own schedule.
Each gate is separately contracted. This engagement, and any results from it, remain confidential and are not referenced externally without NATCO's prior written consent.
"Where other platforms stop at capability, SkyEdgeAI adds continuous assurance."
Lives inside the system that generated it
Sits independently outside every source system
