Govern
From Shadow AI to governed assistants
Bring scattered ChatGPT use into one platform: the same models teams already like, now with permissions, audit and data that never trains third-party models.
Insurance
All your insurer's AI, governed and compliant
Centralize how every insurance team uses AI on one platform: any model, no lock-in, corporate data that never trains third-party models, cost under control, and Solvency II, DORA, GDPR and IDD compliance by design.
See how it worksShadow AI under control
Replace scattered, ungoverned chatbot use with one platform — permissions and audit included
Your data stays yours
Corporate data never trains third-party models; you control residency and access
Insurance-grade compliance
Solvency II, DORA, GDPR and the IDD — traceability on every AI interaction
What your teams build, governed
One governed platform for every insurance team
Underwriting, actuarial, claims, pricing, compliance — each team builds what it needs with any AI model, while you keep one place to set permissions, control cost and prove compliance.
Conversational AI
Conversational AI on your data, under your control
Give teams a governed way to chat with any AI model over approved data and tools (MCP) — instead of pasting claims and policy data into public chatbots. Every conversation runs with permissions and a full audit trail, and corporate data stays private.
- Any model — switch without re-platforming
- Acts through approved tools (MCP), with guardrails
- Permissions and audit on every conversation
- Corporate data never leaves your governance
Threads · Claims analysis claims_dbpolicies.csv
Understanding intent…
Querying claims_db
Auto-generated SQL
Calling servicenow-mcp…
MCP tool call
{ "tool": "servicenow-mcp", "action": "create_task", "summary": "Flag the top 10 suspicious claims for fraud review" }
via servicenow-mcp · just now
Corporate assistants
Corporate assistants for every line and department, not Shadow AI
Give each department its own assistant — claims, underwriting, policy wording — shared across the insurer, on any model, with permissions, team memory and skills. They work like agents: they search claim files, read policy documents and open tasks in your tools, always within the access of the person asking.
- Claims triage: summarizes a claim file and suggests next steps
- Underwriting support: pulls key facts from a risk file and checks them against your guidelines
- Policy-wording Q&A: answers agents and brokers from your approved policy docs, citing the clause
- Team memory: assistants read and update a knowledge base of underwriting guidelines, claim precedents and coverage interpretations
- Acts as an agent through approved tools (MCP), with permissions, audit and cost limits
Set the assistant's name, model and base instructions.
🤖
* Name
Data Analyst
Description
An AI assistant that helps you analyze and explore your data.
Model
Claude Fable 5.1 (EU) 🇪🇺
Instructions
You are a data analyst assistant. Help users explore, analyze, and understand their data. Write clear SQL queries, explain results, and suggest insights.
Optimize prompt
Conversation starters (2 / 4) Add
Explore tables What tables are available? Show me their schemas.
Quick summary Give me a summary of the most important table.
Skills
Your insurer's know-how, as reusable skills
Turn the way your teams work into skills: the underwriting guidelines, the claims reserving criteria, the format for internal and supervisory reports. Anyone picks them with / while writing, and the assistant only loads them when they're needed.
- Company-wide skills shared from one project and edited in one place
- Line skills for auto, home or health, visible only to that team
- Pick them while typing with / — only what you pick travels with the message
- The assistant can write a new skill from an answer that worked
Search skills…
AllOrganizationProjectPersonal
- Organization 2.4k
/underwriting-guidelines
Appetite, limits and referral rules by line
- Project 1.3k
/property-risk-review
Checklist for commercial property submissions
- Project 1.1k
/claims-reserving-criteria
How the claims team sets initial reserves
- Organization 950
/regulatory-report-format
Structure for internal and supervisory reports
- Personal 380
/my-broker-emails
How I write to brokers
underwriting-guidelines property-risk-review
Review this commercial property submission against our underwriting guidelines
Only the skills you pick travel with this message.
Artifacts
Apps built from a conversation, organized and governed
Ask for a claims backlog tracker or a premium quote simulator and the assistant builds it as an artifact — a live app. Each one stays with its conversation, keeps every version and is published under its own URL with permissions, ready to become a team tool.
- Lives with the conversation that created it — no loose files
- A version per iteration; restore or republish any of them
- Published under its own URL, with permissions
- Export a copy to the project when it becomes a team tool
- Reads claims and policy data through MINEO, server-side
Build a tracker of open claims by line that the claims managers can open.
Querying claims_db · 3 lines of business
index.html · app.js · styles.css
Done — version 3 adds the SLA breaches you asked for. It reads live from claims_db through MINEO, so it stays up to date.
Highlight claims open for more than 30 days…
v1v2v3 Restore
Publish mineo.app/a/open-claims
Claims · Open
Open claims by line
Open claims 1,460
Past SLA 37
Closed within SLA 88%
- Auto 730
- Home 438
- Health 292
Notebooks
Governed notebooks for actuaries and pricing teams
More than Jupyter. For the teams that need code: Python, AI cells on any model and interactive widgets in a collaborative, governed environment — with data connected securely and one-click deployment to live apps.
- 5 cell types: Code, Markdown, AI cell, Widgets, Snippets
- AI cells on any model for code and analysis
- Connected data with permissions, not loose copies
- Real-time collaboration with access control
- One-click deployment to governed live apps
Loss Ratio Review 
100%
Code Markdown Assistant Widget Snippet
Monthly loss ratio analysis
Compare loss ratios across lines of business and regions.
Region: All Regions
Results: 8
Apply df = mineo.query("SELECT line_of_business, SUM(incurred_losses) FROM claims GROUP BY line_of_business")
px.bar(df, x="line_of_business", y="incurred_losses", title="Loss Ratio by LoB")
NA
EU
APAC
LATAM
Assistant
Key insight: North America drives 43% of total revenue. APAC showed the strongest growth at +23% QoQ, driven by expansion in Japan and Australia.
[2]
df.sort_values("revenue", ascending=False).head()
| Region | Revenue | Growth |
|---|---|---|
| North America | $1.8M | +12% |
| Europe | $1.1M | +8% |
| APAC | $820K | +23% |
| LATAM | $480K | -3% |
v12
ProWorker 2 CPU · 6GB
Deploy as App Live Apps
Internal apps your teams use, with permissions
Turn the work into always-on dashboards and applications for the whole insurance org. Deploy with Streamlit, Gradio, Dash and 6+ more frameworks — with permissions and access control built in, ready for your stakeholders.
- 9+ Python frameworks: Streamlit, Gradio, Dash, Panel, and more
- Always-on apps with custom domains and 99.9% SLA
- Permissions and access control on every app
- Real-time data from governed data sources
MINEO · Claims dashboard Live
Claims overview
Last updated: 2 minutes ago
Filters Line: Auto Export
Claims filed
4,218
+14% vs Q3
Active policies
82,340
+6% vs Q3
Loss ratio
62%
-2% vs Q3
Claims by month FY 2026
$850K Q1
$920K Q2
$1.1M Q3
$1.3M Q4
| Customer | Region | Revenue | Status |
|---|---|---|---|
| Acme Corp | NA | $245K | Active |
| TechFlow GmbH | EU | $182K | Active |
| Sakura Ltd | APAC | $156K | Pending |
| DataBr SA | LATAM | $98K | Active |
claims.mineo.app Powered by Streamlit
Dev Environments
Full VS Code in the cloud, governed
A complete Linux environment in the browser for technical teams. Install any tool — Claude Code, Codex, Gemini CLI — connect governed data sources and collaborate, without code or data leaving your control.
- Full VS Code with extensions and terminal
- Any tool: Claude Code, Codex, Gemini CLI
- Custom Docker images and GPU acceleration
- Connected to governed data sources and Git integration
VS Code — Dev Environment Explorer
src
app.py
pipeline.py
config.yaml
requirements.txt
Dockerfile
claims_dashboard.py
pipeline.py
import streamlit as st
from mineo import DataSource
import plotly.express as px
st.set_page_config(layout="wide")
st.title("Claims Dashboard")
ds = DataSource("claims_db")
df = ds.query("SELECT * FROM claims")
# Layout
col1, col2 = st.columns(2)
col1.metric("Incurred losses", f"${df.incurred_losses.sum():,.0f}")
col2.metric("Claims", f"{df.claim_id.nunique():,}")
# Charts
fig = px.bar(df, x="month", y="incurred_losses", color="line_of_business")
st.plotly_chart(fig, use_container_width=True)
TERMINAL | zsh
mineo-dev@workspace:~/project$
Successfully installed plotly-5.22 scikit-learn-1.5
Your app is live at https://claims-dashboard.mineo.app
✓ Deployed successfully
Pipelines
Automate AI workflows with control and audit
Chain notebooks and AI steps into production workflows. Schedule with cron, trigger via REST API, and govern every run from one platform — with full history and audit trail.
- Chain notebooks as sequential pipeline steps
- Schedule with cron or trigger via REST API
- Configurable compute resources per pipeline
- Execution history with logs and audit trail
- Assistants as scheduled steps — every morning, a summary of new claims flagged for review
/ Projects / insurance / Daily claims refresh General Elements 4 Executions API
Elements
1 extract_claims.ipynb
2 validate_claims.ipynb
3 publish_claims_report.ipynb
4 Sales assistant: summarize the day Assistant step
Resource Config
Worker Environment
ClaimsWorker 2 cores · 6 GB
Scheduling
Crontab Expression 0 6 * * *
Every day at 06:00
Active
API
Enabled
POST /v1/pipelines/{id}/run
Last Execution Running • -- • Waiting
Why insurance leaders choose MINEO
Govern AI across the insurer
For a CTO or CISO the value isn't one more tool — it's one governed place where every team's AI is visible, controlled and compliant.
Control cost
AI spend you can actually see
Track AI cost per team and use case, set budgets and route to the right model — no more surprise invoices across departments.
Prove compliance
Audit-ready for insurance regulation
Every AI interaction is logged and traceable, helping you evidence Solvency II, DORA, GDPR and IDD requirements without extra tooling.
Govern all your insurer's AI on one platform
Every model, no lock-in, data that never trains third-party models, cost under control and insurance-grade compliance — Solvency II, DORA, GDPR and the IDD.