/ Technology · Workspace

An intelligent workspace scientists actually use.

Natural-language sessions, 3D response surfaces, active-learning suggestions, and GxP compliance, unified in one environment. This is a live, interactive view of the GyaniMed Virtual Formulation System.

GyaniMedVFS
3
PS
Dr Priya Sharma
Formulation Scientist
Active session
Optimisation campaign
LNP‑mRNA Optimisation
Running · iteration 7/12
Recent formulations
View all
LNP‑247Current
TFR 8.1 · FRR 3.0 · N/P 4.5
Size 82 nm · PDI 0.08
LNP‑24618 min ago
TFR 7.3 · FRR 2.8 · N/P 4.0
Size 95 nm · PDI 0.09
LNP‑24545 min ago
TFR 9.0 · FRR 3.6 · N/P 5.0
Size 78 nm · PDI 0.07
LNP‑2441 hr ago
TFR 6.5 · FRR 2.5 · N/P 3.5
Size 112 nm · PDI 0.11
LNP‑2432 hr ago
TFR 10.2 · FRR 4.1 · N/P 5.5
Size 145 nm · PDI 0.13
Response surface · Particle size (nm) TFR × FRR
Optimal 80–100 nm
40170
Total flow rate (mL/min) →
Optimisation trajectory real‑time
Observed Predicted (GEM) Target band 80–100 nm
Active learning · next experiment (iteration 8) TFR 8.2   FRR 3.1   N/P 4.5 Predicted outcome: 88 ± 6 nm · PDI 0.08 · EE 92%
AI agent assistant
Online
Based on PINN analysis, reducing PEG‑lipid to 1.8 mol% should improve EE by ~12% while keeping size below 100 nm.
Confidence 94%
Key molecular insight
PEG‑lipid 1.8 mol% · predicted ↑ EE, ↔ size
Experiment queue
LNP‑248 · iteration 8queued
LNP‑247 · iteration 7complete
LNP‑246 · iteration 6complete
LNP‑245 · iteration 5complete
BIO‑112 · mAb viscositycomplete
Throughput · last 14 days
Designed214
Predicted214
Run in lab48
Confirmed44

Active learning replaced 166 wet‑lab runs with in‑silico prediction this cycle, a 78% reduction in experiments.

GyaniMed‑FDD · deployed ensemble
PCBE · physics‑chemistry‑biology44 rules
PEM · embedding model32‑dim
BNN · Bayesian network12→10→4
GEM · generative explanationcausal
Ensemble weightingPCBE 60 / BNN 40
Held‑out validation · R² by CQA
Particle size0.86
PDI0.74
Encapsulation0.81
Zeta potential0.77

469‑formulation held‑out benchmark · target R² ≥ 0.7 across all critical quality attributes.

Generated documents
CMC formulation summary · LNP‑247signed
Development report · campaignsigned
Risk assessment · ICH Q9signed
Stability protocol · draftreview
Compliance posture
21 CFR Part 11 · e‑recordsactive
Audit trailcomplete
ICH M15 alignmentready
Deploymentair‑gapped

Every parameter change, prediction, and sign‑off is version‑controlled and traceable by design.

All systems operational GxP compliant · 21 CFR Part 11 Air‑gapped · secure environment Audit trail active Session VFS‑2026‑06‑13‑1024 v2.4.1
Scroll horizontally to explore the full workspace
/ Workflow in action

From experiment to insight in minutes.

Version-controlled experiments, automated compliance review, and real-time degradation analysis, all within one workspace.

Experiment diff · 3 parameters changedLNP‑2026‑47
ParameterBatch 02Batch 03
Ionisable lipidSM‑102 50.0%SM‑102 50.0%
Cholesterol40.5%38.5% updated
PEG‑lipid (DMG)1.0%1.5% updated
N/P ratio4.06.0 updated
Microfluidic mixing12 mL/min12 mL/min
Particle size
87.3 nm
↓ 2.3%
PDI
0.14
↓ 0.03
Encapsulation
92.1%
↑ 2.1%
Cellular uptake
68%
↑ 7%
Commit message · auto‑drafted
Reduced cholesterol to 38.5% to relieve phase‑separation threshold; raised PEG‑lipid to 1.5% to stabilise corona and reduce aggregation; N/P to 6.0 to improve encapsulation.
Signed · Dr S. Khadke  +  Agent‑QC (1012 checks passed)
Agent‑QC · review of batch 0312 observations
1
critical
2
warnings
9
passing
CriticalParticle size > USP tolerance at T = 5°C after 72 h
DLS at 0 h: 87.3 nm. At 72 h: 116.8 nm (+33.8%). Exceeds USP <729> by 20%. Recommend buffer ionic-strength optimisation and a repeat at 5°C.
WarningReplicate 3 zeta-potential outlier (z = 0.91)
Replicate 3 reads −4.2 mV vs group mean +1.8 mV. Flagged for human sign‑off; suggest re‑measurement before inclusion.
WarningInstrument export missing SHA‑256 for run 04
Raw DLS export for run 04 lacks an integrity hash. Re‑export required to satisfy 21 CFR Part 11 before lock.
PassEncapsulation efficiency > 86% · 9 further checks
Size in target band, PDI < 0.2 (monodisperse), endotoxin < 0.5 EU/mL, mRNA integrity > 80%, pH 6.6–7.4, osmolality in range, appearance clear. All passing.
Mechanistic hypothesis

Observed growth in hydrodynamic diameter over 72 h at 5°C suggests ongoing colloidal aggregation, consistent with limited colloidal stability from reduced cholesterol. Elevated ionic strength may reduce neutral PEG‑corona repulsion.

Confidence: moderate · ICH Q9 risk score: high

See the workspace on your own formulations.

We run guided pilots on air-gapped infrastructure for LNP, biologics, ADCs, and cell therapies.

Get in touch