What Eunice does
Every capability below is in the current release. Where a limit matters to you, it is stated.
Ask and analyze
Ask Your Document Library
Upload permits, guidance, orders, and reports, then ask questions in plain language and get an answer with the exact source passages attached.
Connected Public Data Sources
Query federal and state data from agencies such as EPA, NOAA, FEMA, USGS, Census, and HUD by asking for it, with health monitoring, caching, and an automatic pivot when a source is down.
Resolve Records Across Systems
Decide whether a facility or asset that appears in two systems is the same thing, with every attribute carrying its source and time validity. Nothing merges automatically: a person decides each match, and every merge can be reversed.
One Catalog For Every Dataset
Every upload, derived table, document, and export is registered once with lineage, duplicate detection, and grouping, so an analyst finds what exists instead of pulling it again.
Documents, spreadsheets, and reports
Classify And Extract From Documents
Sort incoming documents against your own taxonomies and pull structured fields out of them, including batch labeling that turns a set of narrative documents into a queryable table. Extraction confidence is reported by the model itself, so review remains part of the process.
Generate Reports And Spreadsheets
Turn an analysis into a finished Word document, Excel workbook, or geospatial export, then keep editing it in an in-app document or spreadsheet workspace with version history.
Maps and screening
Map And Analyze Locations
Join, buffer, filter, and geocode facility and boundary data, then see it on an interactive map with public data overlays, without leaving the conversation.
Screen A Chemical Release
See where a buoyant or dense-gas release would travel, contoured against published exposure thresholds for the specific chemical. This is a screening tool, not a regulatory-grade model, and the output is a top-down concentration view.
Models and charts
Build Explainable Risk Models
Train a scoring or anomaly model on your own data and get per-record risk scores plus a feature-by-feature explanation of why any record scored the way it did.
Explore Data In Interactive Charts
Click into a chart and have the selection feed back into the conversation. The same chart renders identically in the conversation, in the side panel, and full screen.
Governance and integration
Replay Any Conversation
Every message, tool call, input, output, and cost is recorded without truncation and can be replayed, filtered, or exported as a report.
Human Approval For Record Merges
Merging or splitting records across systems is proposed to an approvals ledger and released only by an authenticated person, with two approvers for higher tiers. Sending email or changing a ticket asks the person first.
See AI Spend
Every model call is metered and attributed by user, module, and conversation, with monthly budget checks and a dashboard. Budgets alert rather than block.
Save A Procedure And Reuse It
Capture a repeatable review or reporting procedure once, in readable prose plus optional scripts, then invoke it by name or let the assistant surface it. A saved procedure is private to its owner; there is no sharing interface yet.
Connect Your Existing Systems
Attach outside tools and ticketing systems through a managed connection catalog with an encrypted credential vault, and pull work items into one triage view. Work items are snapshots at the time of the pull.
Fine Grained Access Controls
Access is granted per capability rather than per job title, conversations are private to their owner by default, and each customer's data is isolated at the tenant boundary.