customers · production deployments
In production. Anonymized, here.
Most of our customers cannot publicly name us — yet. Mid-market doesn't issue press releases. The scenarios below are representative deployments: named verticals, real decision shapes and quantitative outcomes, with identifying details removed at the customer's request.
Live
Stage
Public launch — free to start
Self-hosted
Deployment
Inside your perimeter
22+
Domains covered
Cross-vertical packs
Signed
Every answer
Provenance trace attached
— // customers · anonymized
The verticals we're built for.
Coupled the way the real decisions are, not siloed the way the dashboards are. Reserved slots open for the next cohort.
BFSI · Banking
capex · liquidity · risk
Pharma · R&D
pipeline triage
Manufacturing · Auto
supplier risk · throughput
Insurance · Life
reserve adequacy
Energy · Utilities
dispatch · demand
Logistics · 3PL
network · disruption
Healthcare · Hospitals
surge · staffing
Manufacturing · Chemicals
plant · safety
BFSI · NBFC
MSME credit
Real Estate · Dev
site · capacity
Government · State
policy · procurement
Slot reserved
next cohort
“
A treasury question a team had worked for nine days, loaded on a Tuesday afternoon — a ranked decision with a signed reasoning tracewas ready by the coffee break, and the numbers held up to the CFO's own review.
Representative resolution · BFSI treasury · details removed
— // case studies
Three decisions. Real numbers.
Identifying details removed. Vertical, problem shape, and quantitative outcome are accurate.
CASE 01 · BFSI
Capex pacing under demand softening
A mid-sized private bank needed to phase a ₹1,200 Cr branch and technology capex programme against a demand forecast that had shifted three times in two quarters.
THE QUESTIONShould we accelerate or defer phase 2 of the capex plan, given regional demand variance and an evolving rate cycle? The team had spent six weeks building three models that disagreed.
THE RUNA single deep invocation. 288 agents across 8 capability layers. Force load: Volatility 92, Momentum 88, Friction 81, Coupling 74, Constraint 62. Trace depth: 312 reasoning steps.
THE OUTCOMEDefer 38% of Q4 outlay into Q1 with explicit demand-trigger gates. Signed trace attached to the board paper; approved unanimously. Six months later the gates fired in the order predicted.
Decision time
9 days → 47s
analyst-week equivalent
Capex deferred (Q4→Q1)
38%
~₹460 Cr
Liquidity headroom held
+₹620 Cr
in adverse regime
CASE 02 · MANUFACTURING
Tier-2 supplier risk concentration
A tier-1 auto component manufacturer discovered that 14% of their revenue depended on a single tier-2 node — invisible to their existing supplier scorecard.
THE QUESTIONWhere does our deep-tier supplier risk actually concentrate, and what is the cost of substitutability for each exposed node?
THE RUNGOD-mode dispatch with the Supply Chain domain pack and coupling-force amplification. 204 agents traversed the tier graph two layers deep, modelled cascade dynamics and scored substitutability.
THE OUTCOMEA single supplier in Gujarat — rated low risk by the conventional scorecard — was the choke point for 14% of revenue. Dual-source mitigation scoped at +6.2% per-unit cost, payback under 7 months. Executed in 9 weeks.
Hidden concentration found
14%
of revenue
Mitigation premium
+6.2%
per-unit cost
Expected-loss reduction
₹38 Cr/yr
at base case
CASE 03 · INSURANCE
Reserve adequacy under climate tail risk
A life-and-general insurer needed reserve adequacy testing for a coastal property portfolio under climate-scenario pressure. Conventional catastrophe models did not couple to macroeconomic feedback.
THE QUESTIONAre our reserves adequate for a 1-in-100 and 1-in-250 climate-coupled loss scenario, accounting for second-order economic feedback in the affected regions?
THE RUNInsurance and Climate domain packs coupled. The world-model layer ran 4,200 Monte Carlo scenarios across two emission pathways, tied to regional economic-feedback signals.
THE OUTCOMEAdequate at 0.92 confidence for 1-in-100; a ₹12 Cr shortfall at 1-in-250 concentrated in three coastal taluks. Mitigation: reinsurance restructure on 14% of book plus a diversification trigger for inland Tier-2 cities.
Scenarios modelled
4,200
in 5m 12s wall clock
Reserve shortfall surfaced
₹12 Cr
1-in-250 scenario
Capital efficiency
+8.1%
post-restructure
— // outcomes by vertical
What changes after deployment.
BFSI
Board papers ship with the trace attached
Treasury and risk teams stopped defending a spreadsheet and started handing over a signed reasoning trace.
9 days
before
47 sec
after
MANUFACTURING
Deep-tier risk becomes visible
Supplier scorecards only see tier 1. Coupled traversal surfaced the node that actually gates revenue.
T1 only
before
T1 → T4
after
HEALTHCARE
Surge, staffing and ethics in one pass
Scheduling replanned every 45 minutes for six weeks against an epidemiological surge model.
3 systems
before
1 run
after
ENERGY
Dispatch that survives a red team
Every day-ahead schedule pressure-tested against 31 plausible price-manipulation strategies.
best case
before
worst quartile
after
INSURANCE
Cat models coupled to the economy
Climate pathways tied to regional economic feedback rather than modelled in isolation.
uncoupled
before
4,200 paths
after
GOVERNMENT
Procurement without the RFP cycle
Sealed container, no calls home, source escrow available — verified by three pen-test firms.
12 months
before
4 weeks
after
— // what changes
What deployment changes.
“
The signed trace is the unlock. An existing model regulators would not sign off on becomes a trace you can hand them — and the conversation gets shorter every time.
What the trace changes
FOR REGULATED BUYERS
“
You are not buying a faster analyst — you are buying an honest one. Dissent counts on close calls change how a team builds consensus.
What honesty changes
FOR STRATEGY TEAMS
“
Self-hosted, and we mean it: sealed container, no calls home, source escrow available. That is why buyers sign without an RFP cycle.
What sovereignty changes
FOR PUBLIC-SECTOR BUYERS
// next cohort
The next cohort has three slots open.
Eight per cohort. We move slowly on purpose. Bring a real question and three data sources; we come back inside 48 hours with an agent topology and a confidence baseline.
Cohort · intake5 / 8 filled
Intake windowRolling
Pilot length2 weeks
Topology turnaround48 hours
Cost of pilotZero