AI underwriting picks up speed: Sixfold teams with Skyward Specialty and Mphasis
AI underwriting is moving from pilots to production. In 2025, Sixfold announced partnerships with Skyward Specialty to advance and scale AI underwriting, and with Mphasis to strengthen underwriting capabilities. Sixfold was also named to the CB Insights Insurtech 50, signaling growing traction and market confidence.
Mphasis is known for IT services with depth in cloud and cognitive solutions, which points to real integration work ahead-data, workflows, and change management. Skyward Specialty's engagement suggests targeted use cases with measurable underwriting outcomes, not hype.
Why this matters for carriers and MGAs
- Faster turnaround: prefill, entity resolution, and automated submission triage reduce cycle time.
- Cleaner intake: AI can normalize broker submissions, extract key terms, and flag missing data early.
- Sharper appetite fit: routing by class, hazard signals, and commercial rules limits wasted reviews.
- Consistent decisions: policy rules and checklists enforced by AI with audit trails for compliance.
- Underwriter leverage: more time on edge risks and portfolio strategy instead of admin work.
What these partnerships signal
- Enterprise integration: Expect connectors into policy admin, document stores, and data vendors. This is about workflow, not a demo.
- Human-in-the-loop: Underwriters stay in control, with AI handling low-value steps and providing explainable prompts.
- Market validation: Recognition in the Insurtech 50 hints at customer adoption and a maturing product. See the list from CB Insights here.
Implementation checklist for underwriting leaders
- Start with one line and 2-3 narrow use cases: submission triage, data prefill, and referral rules.
- Map data access: broker emails, ACORD forms, loss runs, third-party data. Decide what the model can read and store.
- Set governance: approval thresholds, override rules, reason codes, and audit logs.
- Define success metrics before kickoff (see below). Lock a baseline.
- Plan change management: job aids, 30-minute playbooks, and "what good looks like" examples.
- Security review: PII handling, encryption, retention, vendor pen-tests, and SOC reports.
Metrics to track in the first 90 days
- Submission-to-quote time (median and 90th percentile).
- Bind ratio by segment and broker.
- Triage accuracy: percent routed to correct team or appetite bucket.
- Underwriter time on admin per file (self-reported and sampled).
- Loss pick variance vs. selected; referral rate and rework rate.
Questions to ask any AI underwriting vendor
- What data do you use, and where is it stored? Can we keep data in our tenant?
- Explainability: can the system show sources, confidence, and reason codes?
- Controls: how do overrides work? Are there hard stops for authority limits?
- Retraining: how often do models update, and who approves changes?
- Integration: email intake, document ingestion, SSO, and PAS/CRM connectors.
- Compliance: PII/PHI handling, audit exports, and retention options per jurisdiction.
Practical next steps
- Pick one unit with clean broker flow and a cooperative leader. Run a 6-8 week pilot.
- Limit scope to two KPIs. If they move, expand. If they stall, fix data or process first.
- Keep a "no heroics" rule: whatever works in pilot must scale without manual patches.
The signal is clear: AI is becoming part of the underwriting desk. Start small, measure hard, and scale what proves value.
If you're building skills for your team, you can scan practical AI programs by role here.
Your membership also unlocks:
