1
Run ingestion
Start a run by uploading a zip (or connecting signals): tickets, logs, regression output, diffs, release notes, environment context.
Inputs are preserved and linked to every work item.
Governed AI-powered operations for insurance software
Artifacts in. Reviewable work items out. Cadence protected.
Kargha is an AI-powered workbench that reduces friction in repeatable BAU workflows and enables faster delivery with built-in governance through evidence-backed work, structured review, and traceable decisions.
Human-in-the-loop approvals
Inspectable evidence
Audit trail + KPIs
Built by Yovant - full-suite Duck Creek subject matter experts.
Operating loop
Runs -> Work Items
Control points
Review-first + traceable decisions
How it works
Yovant standardizes production support (BAU) operations as repeatable workflow runs. Kargha supports the model by producing evidence-backed work items, routed through human approval queues before anything is executed.
2
Agent analysis
Generate proposed work items and drafts, with an inspectable evidence bundle for every recommendation.
AI is the multiplier, not the decision-maker.
3
Workbench governance
Humans review, refine, approve, and export payloads: bug payloads, runbooks, reports, and draft manuscripts.
Cadence protection through control points.
Sample Actions
Systems
Regression test outputs
Database discrepancy scan results
Release notes
Support artifacts (zips, exports)
Kargha
Triage test results
Draft data fixes
Produce draft manuscripts
Propose payloads + KPIs
Humans
Review results + evidence
Approve / reject / refine
Provide feedback for improvement
Execute approved payloads
Two ways to start
Prove value quickly in the BAU workflows that block cadence the most: regression and upgrades.
Regression Triage Agent
Reduce overnight failure noise and shorten triage loops with evidence-backed classifications, AI-generated defect reports, and automated re-baselining.
Run ingests results + logs + diffs
Agent proposes work items w/ evidence
Humans approve and export payloads
Time-to-triage
Down
Signal quality
Up
Release delays
Down
Upgrade Agent
Make upgrade work more predictable with run-based impact review, conflict surfacing, and governance queues.
Run ingests manuscript sets
Agent drafts impact review work
Humans approve changes and track risk
Upgrade cycle time
Down
Rework
Down
Stabilization
Faster
Trust, control, and cadence protection
Kargha is designed for enterprise delivery: review-first, evidence-backed, and traceable decisions end-to-end.
What Kargha is
Automates high-friction BAU workflows (testing, triage, upgrade readiness)
Produces work items with inspectable evidence bundles
Routes proposals through human approval queues in the Workbench
Exports to your existing systems (issue tracking, pipelines, runbooks, reports)
What Kargha is not
No autonomous production changes
Not a push-button Duck Creek implementation engine
No speculative AI product bets without validated demand
Not a replacement for domain SMEs; it scales their output
Governance doesn't slow delivery. Governance makes delivery repeatable.
Kargha design principle for carrier-grade BAU
See it first-hand
Get a short walkthrough of how runs, evidence, and approval queues work together in the Workbench.
Kargha Workbench
Intake
Runs
Queues
KPIs
What you'll see
Runs view with linked artifacts
Work item queues and statuses
Evidence panel for every recommendation
Approval checkpoints and export
Book a live demo
Assessments
Start with a scoped assessment to baseline noise, cycle time, and readiness - then define a pilot with success metrics.
Regression Triage Assessment
Baseline triage time, failure noise, and release delay drivers. Define a pilot scope and success metrics.
Request regression assessmentUpgrade Readiness Assessment
Baseline the carrier environment, manuscript customizations, and target upgrade cadence. Define a pilot with reviewable outcomes.
Request upgrade assessmentAbout Kargha
Name meaning
Kargha means loom - reflects weaving AI, delivery expertise, and insurance domain knowledge.
Purpose
Reduce time to value and improve BAU consistency with scalable AI-powered capabilities.
Book a demo / start an assessment
Tell us what you're trying to improve and we'll route you to the Yovant contact page.