Platform
AI Assistant
Your AI teammate on every screen. The AI Assistant builds flows on the Canvas, triages the review queue, and pulls answers from any document — through plain conversation..
Plain English. Real actions.
A sample of what your team can ask
Try asking
Conversational actioning
Build the flow by describing it.
Tell the AI Assistant what you need — "let's create a new FNOL auto loan claim extraction flow" — and watch the Workflow Canvas build itself. Fields, classification logic, routing rules, the lot. The AI Assistant proposes the configuration, you confirm field-by-field, it commits the changes.
- Flows from plain English. The AI Assistant proposes nodes, fields, and logic. You confirm each step before anything is committed.
- Configuration without code. Adjust extraction prompts, classification rules, confidence thresholds, and reviewer assignments through chat.
- Drafts stay drafts. Every change lands in a draft version. Nothing reaches production until your team publishes.
- For business builders, not developers. Operations teams configure their own flows. IT keeps governance, not gatekeeping.
Document-aware assistance
Ask the document. Get a grounded answer.
"Where is the date of loss?" "Summarize the latest work item." "Why did this field get flagged?" The AI Assistant reads the same document your reviewer is reading, points to the exact source on the page, and explains its reasoning. The AI Assistant is grounded in your work item, your flow, and your documents.
- Answers tied to the source. Every response shows the page, paragraph, and confidence. Highlight Mode is one click away.
- Cuts review time in half. Reviewers stop hunting through 40-page submission packages. Ask the question, get the location.
- Grounded in your data. The AI Assistant works from the work item in front of the reviewer — not the public web.
- It can explain its own decisions. Ask why a field has 0.87 confidence. The AI Assistant tells you what it saw and why it wasn't sure.
Context persistence
One teammate. Every screen. Full memory.
The AI Assistant doesn't reset when you change pages. Start a conversation on the Canvas, finish it on Analytics , pick it back up in a work item review. It knows what you were looking at, what flow you were building, and what question you asked five clicks ago.
- Continuous context. The AI Assistant carries your conversation across every screen - Canvas, queues, work items, dashboards.
- Embedded guidance. Contextual onboarding for new users. Next-best-action prompts for experienced ones. The right help at the right moment.
- Every interaction trains the platform. Corrections, confirmations, and rejections flow into Bevaya's governed learning loop, the same patented architecture behind HITL.
- Governed and auditable. Every AI Assistant action is logged with reviewer ID, timestamp, and outcome. Same audit standard as everything else on the platform.
Who it's for
An expert beside everyone on the platform.
Builders ship faster. Reviewers verify faster. Leaders see what's happening, in plain English, in real time.
Business builders | Ask the Assistant
Operations leaders, claims managers, and underwriting ops — the people who know the work — change extraction logic, routing rules, and thresholds themselves. No JIRA ticket, no waiting on a sprint.
- Adjust extraction logic from the canvas — same-day shipping
- Edit routing rules and confidence thresholds without code
- Conversational edits with diff preview before publish
- Versioned workflows with one-click rollback
- Side-by-side testing on sample submissions
- Sandbox-to-production promotion with approvals
- "Add a duplicate check before the Guidewire write step."
Reviewers | Ask the Assistant
Claims adjusters and underwriters ask the Assistant where data came from, what a field means, or why something was flagged. Then move on.
- Field-level citations to page, paragraph, and coordinates
- X-Ray Mode source highlighting on every extracted value
- Calibrated confidence scores with model rationale
- Inline definitions for fields and coverage codes
- Carrier-specific terminology resolved in context
- One-click navigation from value to source artifact
- "Why is this field below the threshold?"
Leaders & analysts | Ask the Assistant
VPs, COOs, and ops leaders ask for STP rates, queue depth, reviewer SLA, or override patterns — without writing a query or waiting for a BI ticket.
- Live operational metrics in plain English
- Queue depth, throughput, and SLA at a glance
- Override pattern detection across reviewers
- Exception spike alerts with explanations
- Weekly executive summaries generated on demand
- No SQL, no BI ticket, no dashboard refresh wait
- "What's our STP rate on Claims Indexing this week?"
Feature index
Inside the AI Assistant
Configure flows, verify documents, manage queues, and audit decisions - all through conversation.
Conversational actioning | Plain-English actions
Take action through plain conversation.
Tell the Assistant what you need. It proposes the configuration step by step and confirms each change before it commits — no menus, no tickets, no code.
- Natural-language flow building on the Workflow Canvas
- Field-by-field confirmation before changes commit
- Configure extraction, classification, and routing rules
- Adjust confidence thresholds without code
- Reviewer reassignment and queue management
- Quick-prompt suggestions in every panel
- Multi-turn conversation with full edit history
Document-aware assistance | Grounded Q&A
Ask the document. Get a grounded answer.
Every response is tied to the source. The Assistant reads the same document your reviewer is reading, points to the exact location on the page, and explains its reasoning.
- Query extracted fields in plain English
- “Where is the date of loss?” returns the source location
- One-click X-Ray Highlight Mode from any answer
- Work item summarization on demand
- Duplicate and exception explanations
- Confidence-score rationale (“why is this flagged?”)
- Cross-document Q&A within a work item
Context persistence | Cross-screen memory
One conversation. Every screen.
The Assistant doesn’t reset when you change pages. It carries your conversation across the Canvas, queues, work items, and dashboards — and remembers what you were working on.
- Memory across dashboards, flows, work items, and documents
- Knows the active work item, flow version, and reviewer
- Continuous conversation across sessions
- Embedded onboarding for new users
- Next-best-action prompts in every screen
- Shortcut suggestions for power users
Governed & explainable | Audit & compliance
Built on the same trust architecture.
The Assistant operates within your audit trail. Every action is logged, every response carries citations, and every interaction respects your role-based permissions.
- Citations and confidence on every response
- Reasoning shown alongside the answer
- RBAC-aware — only acts within the user’s permissions
- Immutable audit log for every Assistant action
- Reviewer feedback feeds the governed learning loop
- No external data leakage — grounded in your tenant
- SOC 2 Type 2, HIPAA, GDPR, NYCRR 500
Resources & insights
More on conversational AI for insurance
Research
Page stream segmentation with LLMs
How Bevaya Labs approaches a foundational problem in insurance document AI.
Case Study
Workers' comp carrier processes claims 100x faster
How indexing automation delivered 432% ROI in 12 months.
Architecture
Inside the Bevaya platform architecture
How specialized models, HITL controls, and integrations come together in production.
More Capabilities
Explore the rest of the platform.
Designed, deployed, and governed together. Powered by InsurGPT™ and accessed through the AI Assistant.
Workflow Canvas
Visual builder and production runtime for every automation.
Current page ReviewHuman-in-the-Loop
Configurable review queues with X-Ray verification and a patented feedback loop.
Current page DocumentsDocument Intelligence
Read any insurance document — hundreds of carrier formats, scanned or digital.
Current page GroundingGrounded Explainability
Every value traceable to its source. X-Ray Highlight Mode brings citations to reviewers.
Current page AnalyticsAnalytics Dashboard
Live accuracy, STP rates, reviewer SLA, and agent performance across every workflow.
Current page GovernanceGoverned Automation
Immutable audit trails, role-based access, flow versioning. Compliance is the architecture.
Current pageInsurGPT™
A mosaic of specialized insurance AI models, trained on real insurance data.
Current pageAn AI agent across the platform
Builds your workflows. Executes work alongside you. Explains every step.
Current pageTrust & Security
Trust by design
Built for an industry where data security isn't optional.
Your data stays yours.
Never shared with other customers or vendors. Bevaya doesn't train shared models on your data.
Visit the Trust CenterYour data · only your team sees it
Encrypted end-to-end.
256-bit AES encryption, in transit and at rest. Independent third-party audits conducted annually.
Visit the Trust CenterAudited annually · independent third party
Runs in Azure.
Enterprise-grade infrastructure, hosted where insurance organizations already trust their data.
Visit the Trust CenterDeploy where your stack already lives
Every decision audited.
Immutable audit logs. Confidence scoring. Human-in-the-Loop review on low-confidence items.
Visit the Trust CenterImmutable trail · every decision, every reviewer
FAQ
Common questions
No. A chatbot answers questions. The AI Assistant takes action — building Flows on the Workflow Canvas, filtering queues, summarizing work items, reassigning reviewers. It's an AI teammate inside the platform, not a help widget bolted onto it.
Every change the AI Assistant proposes lands in a draft version of a Flow. Nothing reaches production until a human reviews and publishes it. The Workflow Canvas keeps strict separation between draft, staging, and production environments.
Generic LLMs guess at insurance. The Bevaya AI Assistant is grounded in your work item, your flow, your documents, and your platform actions — and it's powered by InsurGPT™, our mosaic of specialized insurance AI models. It doesn't hallucinate document fields. It cites its sources. And it operates within your role-based permissions and audit trail.
Yes — that's ultimately our design goal for our platform. The AI Assistant proposes the configuration step by step and confirms each field before committing. Bevaya's team is alongside you for production deployments and ongoing tuning, but day-to-day adjustments — new fields, new routing rules, new thresholds — will soon no longer require an engineering ticket.
Every AI Assistant action is captured in the immutable audit log: the user, the prompt, the proposed change, the confirmation, the resulting flow or work-item update, and the timestamp. Logs are exportable for SOC 2 and HIPAA reporting and respect the same RBAC partitioning as the rest of the platform.
Yes — through Bevaya's governed learning loop. Every confirmation, correction, and rejection feeds back into the models, scoped to your tenant. The AI Assistant gets better at recognizing your flows, your fields, and your team's preferences over time. No data leaves your environment.