Integrius Optic
Ask a question in plain English. Get KPIs, the right chart, a forecast, and a written summary. Nothing leaves your network.
Optic resolves which governed data products answer your question, fetches live data through Core, and a local LLM composes the answer. On-prem inference via Ollama by default. No OpenAI. No data exfiltration. No per-query bill.
How an Answer Happens
Four steps between a question and a board-ready answer. Governance is not a checkbox at the end, it is step three.
You ask
"Which trial sites have the highest dropout rate?" Plain English. No SQL, no semantic model, no dashboard to find.
Optic resolves
Optic works out which governed data products answer the question, using the same ownership and schema metadata Core already maintains.
Core enforces
Data is fetched live through Core with your identity. RBAC is applied before the query runs. You can never see data you are not permitted to see.
The answer appears
A local LLM composes KPIs, the right chart, a forecast, and a written summary. The answer is the dashboard.
The AI runs where your data lives.
Every other AI analytics product sends your questions, your schemas, and often your data to someone else's cloud. Optic runs a local LLM inside your network, on-prem inference via Ollama by default. Your questions never leave. Your data never leaves. And there is no per-query bill quietly growing in the background.
Governance enforced upstream, not promised downstream.
Optic calls Core with the user's identity, and RBAC is applied before the query runs. A sales analyst asking about revenue sees revenue. The same analyst asking about patient data gets nothing, because Core never returns it. There is no prompt-injection path to data you were never permitted to see.
One Question In. A Whole Intelligence Layer Out.
Everything below runs on the same governed data layer. No separate tools, no exports, no copies.
Ask
One-shot Q&A. A question in, a complete answer out: KPIs, chart, narrative.
Explore
Multi-turn analysis with context. "Show me revenue" then "break it down by tier" then "just enterprise". Each question builds on the last.
Around 30 chart types
D3 sankey, force network, sunburst, stream, spiral, and geo maps alongside every standard chart. Optic picks the right one for the answer.
Forecasting
Exponential smoothing and linear regression with 95% confidence bands. Projections on any metric, no data science team required.
Watchers
Threshold, change, absence, and anomaly alerts, each with AI-generated context explaining what moved and why it matters.
Scheduled Briefs
Auto-emailed intelligence reports. The Monday morning state of the business, written and delivered without anyone asking.
Dashboards
Pin any answer as a tile. Multi-tile grids with persistent, shareable URLs that always show live data.
Report templates
5+ templates with time period selection, executive summary, and recommendations.
PDF and PPTX export
Board-ready decks from a single question. Ask, review, export, present.
No-login share links
Share an answer with anyone, no account needed. Every view lands in the audit trail.
Entity resolution
Optic knows "Acme" and "Acme Corp" are the same company, across every data product.
Real-time schema updates
Server-sent events keep Optic in sync with Core. New fields and products appear as soon as they are governed.
Optic vs Tableau, Power BI, and Looker
They are dashboard tools with AI features attached. Optic is answer-first, and it knows the governance graph.
| Concern | Tableau / Power BI / Looker | Integrius Optic |
|---|---|---|
| How you get an answer | Find or build the right dashboard first | Ask the question. The answer is the dashboard. |
| Where the AI runs | Vendor cloud (Tableau Cloud, Azure, Google) | Inside your network. Ollama on-prem by default. |
| Does data leave your network? | Yes, for cloud AI features | Never |
| Governance | Bolted on per workbook or workspace | Enforced upstream by Core. RBAC applies before the query runs. |
| Per-query AI cost | Metered cloud AI consumption | None. Local inference, flat platform pricing. |
| Pricing model | Per seat | Per governed data product. Everyone in the org can ask. |
| Forecasting | Separate feature or add-on | Built in, with 95% confidence bands |
And versus Monte Carlo: data observability tools monitor data quality, whether the pipeline ran and the rows arrived. Optic's watchers monitor data meaning: whether churn is accelerating, whether a metric is behaving abnormally, whether a number a regulator cares about just moved.
Runs entirely inside your network.
Optic deploys next to Core, on your infrastructure. Inference runs on Ollama, on-prem, by default. The same deployment posture that makes Core viable for pharma, banking, and defense applies to the AI layer too: no SaaS dependency, no phone-home, no data leaving the building.
Stop building dashboards. Start asking questions.
Governed answers, on-prem AI, zero data leakage.