COPEscore AI is live — AI submission assessment in 90 seconds Try free →

AI Engineering
for Insurance.
Built by underwriters
and actuaries.

We design, build, and deploy AI systems for the decisions that matter most in commercial P&C — submission, pricing, coverage, claims, and portfolio. Eighteen years of underwriting and actuarial depth, shipped as code.

18+
Years underwriting & actuarial
10
AI systems scoped
8
Decision Stack layers
L1L8
End-to-end coverage

AXA Gulf
Regional Head P&E · Gulf
RSA Insurance
Pricing Actuary
Flagstone Re
CAT Modelling Analyst
IIT Bombay
B.Tech
ARe
Associate in Reinsurance

What we do

AI engineering and deep consulting for the full insurance stack.

On-demand, packaged, or retained — across every layer of the underwriting decision, from raw submission to portfolio learning.

AI Engineering

Bespoke AI systems designed from first principles for your underwriting operation. Multi-agent pipelines, extraction engines, scoring models, and decision-support tools. Python + Claude. Human-in-the-loop by design.

Submission AI Pricing Engines Coverage AI Portfolio Intelligence

Consulting & Advisory

Strategic advisory on AI transformation, underwriting operations, pricing governance, reinsurance analytics, and data science. Grounded in 18 years of practitioner authority across the full P&C stack.

AI Transformation Pricing Governance Reinsurance Analytics Data Science

SaaS Products

Standalone AI tools for underwriters, coverage counsel, and pricing actuaries. No integration, no enterprise sales — log in and use. COPEscore AI live now; CoverageAI in build. More to follow the same architecture.

COPEscore AI CoverageAI No subscription Pay per use
Engagement models
🔭

Technical Discovery Sprint

Fixed-fee scoping engagement. Architecture, data audit, build plan. Done in two weeks.

📦

Packaged Projects

Pre-scoped AI builds from the product roadmap. Fixed scope, fixed price, fast delivery.

🔄

Fractional AI Engineering

Ongoing embedded capacity — fractional AI engineering support on retainer. Monthly or quarterly.

On-Demand Builds

Scope a build, agree a price, ship. Works across all sizes — from a single model to a full pipeline. Also available on Upwork.


Products

AI tools built for the two moments where underwriting error costs most.

Submission intelligence at intake. Wording intelligence at placement. Both live and in active build.

COPEscore AI
"Your first underwriter on every submission."
Live

Upload a commercial property submission, risk survey, or broker cover note and get a COPE completeness score, risk signals, gap analysis, and ready-to-send broker questions — in under 90 seconds. Built to Lloyd's MRC and ISO standards. Data processed and discarded; never stored.

L1 · Data L2 · Signal
3 free checks · $5/commercial · $3/residential · No subscription
CoverageAI
"Your first coverage counsel on every policy."
In Build

Reads policy wordings, MRC slips, and endorsements the way a senior underwriter or coverage counsel would — surfacing ambiguous clauses, missing sub-limits, CBI trigger gaps, and submission-vs-wording inconsistencies. Designed for underwriters, coverage counsel, claims handlers, and pricing actuaries stress-testing T&C assumptions.

L3 · Coverage L2 · Signal L4 · Pricing
Notify me when live
🗺️
Eight more AI systems in the roadmap
SubmissionIQ · PriceDesk · LossLens · RiskReader · TreatyDesk · PortfolioMind · AuditStack · ParametricDesk — each targeting a specific layer of the Underwriter's Decision Stack.
View roadmap →

How we build

The Underwriter's Decision Stack.

An 8-layer framework mapping the commercial underwriting decision — from evidence assembly to portfolio learning. Every system we build targets one or more layers. AI that works together, not in silos.

L8
Learning
Portfolio Calibration
What did this decision teach the portfolio? Metrics update, models recalibrate, performance tracked. Closed loop.
AuditStackGovernance & drift detection
L7
Decision
Bind, Decline, Refer
All upstream layers converge. Final terms confirmed. Bind, decline, refer, or counter.
SubmissionIQRouting recommendation
L6
Reinsurance
Treaty & Net Retention
Fac need, cession logic, event limit consumption, net vs gross profitability, RI cost per risk.
TreatyDeskNet pricing at point of decision
L5
Judgement
Unstructured Experience
Market context, broker relationship, moral hazard read. Soft information no dataset captures.
PortfolioMindAppetite calibration
L4
Pricing
Technical Adequate Rate
Burning cost, exposure rating, GLM/GBM technical premium, CAT load, capital loading. Pricing follows coverage.
PriceDeskTechnical rate engine
L3
Coverage
Terms, Deductibles, Clauses
Perils, limits, sub-limits, deductibles, warranties, conditions. Coverage is defined before price.
CoverageAIWording intelligence
L2
Signal
Risk Grades & Interpretation
Occupancy risk grade, NatCat exposure score, COPE quality rating, moral hazard markers, loss trend. Evidence becomes interpretation.
COPEscore AISubmission intelligence
L1
Data
Evidence & Raw Elements
Submission docs, SOV, COPE data, loss history, external hazard signals. Evidence assembly.
COPEscore AIData extraction & enrichment
↑ Evidence flows upward
Portfolio learning flows downward ↓

The team

Practitioner depth meets research rigour.

DVM
Dev Vaibhav Markandey
Principal & Founder
18 years across the full commercial P&C stack — underwriting authority, actuarial pricing, CAT modelling, and reinsurance. Regional Head P&E at AXA Gulf, Pricing Actuary at RSA Insurance, CAT modelling analyst at Flagstone Reinsurance. Author of The Underwriter's Decision Stack.
AXA Gulf RSA Insurance Flagstone Re IIT Bombay ARe
KVR
K. Vibhooti Rajkumar
AI Research Advisor
AI Research Advisor at UnderwriteAI. Assistant Professor at CSVTU. Research scholar in object detection and deep learning. Expert in data analytics, text mining, and structured data modelling for insurance applications.
CSVTU Deep Learning Data Analytics Text Mining

Insights

From the Underwriter's Decision Stack.

All articles →
Article Jun 2026

The Commercial Underwriting Audit Trail Exists. It Just Cannot Teach.

Every bound policy is a decision record. The problem is no one reads it as one.

Read on LinkedIn →
Article Jun 2026

The Underwriter Is the Integration Layer — And That's the Problem

When the human is the middleware, every system upgrade touches the human first.

Read on LinkedIn →
Whitepaper Coming soon

From Inbox to Insight: SubmissionIQ Architecture

A technical deep-dive into the five-agent pipeline that reads commercial property submissions end-to-end.

Coming soon

Work with us

No pitch decks.
No sales calls.

A direct conversation about your situation — what you're trying to build, what's blocking you, and whether we're the right fit. Response within 48 hours. Remote-first, global.

📧 support@copescoreai.io
💬 WhatsApp available