Read existing blocks, schemas, samples, options, results and field lineage.
Use AI across the analytics lifecycle, without losing visibility or control.
Use AI inside Omniscope with Workflow Ninja and Insight Explorer, or connect external AI agents through the MCP server. Whichever route you choose, the work becomes a visible, editable and verifiable Omniscope project.
AI can plan and build. Omniscope keeps the work inspectable and reusable.
01 · Prepare & buildWorkflow Ninja connects sources, builds transformations and authors pipelines and reports.
02 · ExploreInsight Explorer turns questions into explainable visual analysis across multiple datasets.
03 · VerifyInspect the queries, calculations, transformations and lineage behind each result.
04 · ReuseKeep useful work as editable and auditable workflows, reports and repeatable analytics.
New · Workflow Ninja
From intent to working, inspectable analytics.
Describe the outcome you need. Workflow Ninja operates Omniscope for you: it finds the data, chooses the right blocks, configures and connects them, executes the workflow, reads what came back and continues until the job is done.
It leaves a real Omniscope project behind.
Real blocks, options, connections, outputs and reports, all open to inspection and editing. The model plans; Omniscope’s engine performs the work exactly as it does when you build by hand.
A plausible answer is not yet trusted analysis.
AI can move quickly, but important decisions need a route back to the data, the logic and the person responsible for the result.
Fast to receive. Hard to take over.
The result can look convincing while the joins, filters, assumptions and calculations remain buried in a conversation.
- Logic may be difficult to inspect
- Corrections can mean starting again
- Useful work disappears into chat history
Fast to build. Possible to verify and own.
The workflow, report and generated analysis remain available in the same visual workspace for a person to examine and continue.
- Open the workflow and inspect every block
- Challenge queries, calculations and assumptions
- Rerun, edit, share and automate the result
Explore · Insight Explorer
Turn connected data into explainable visual insights.
Insight Explorer lets people ask questions across one or multiple datasets and receive structured visual answers. It is the established exploration experience alongside Workflow Ninja’s new authoring capability.
- See the interpretation, activity and explanation behind an answer
- Inspect generated queries, transformations and visualisations
- Check calculations, caveats and explicit assumptions
- Download results or promote useful queries and charts into the report
Ask across your data.
Use plain language to explore one or several connected data sources.
Follow the logic.
Review the query lineage, transformations, chart configuration and caveats.
Challenge the answer.
Challenge assumptions, inspect calculations and compare the result with the source data.
Make it durable.
Save the exploration or promote useful queries, data and views into the report.
Visibility and control at every layer.
Verification is a property of the working environment: what was connected, what ran, what changed, what was shared and what can be reused.
Data and lineage
Keep sources, joins, field changes and transformations visible in the workflow.
Queries and calculations
Inspect the logic behind answers, including formulas, unit conversions and assumptions.
Charts and reports
Open generated views, edit their configuration and reuse them in production reports.
Models and providers
Choose the provider or local endpoint appropriate for each integration and task.
Data sharing
Set provider-level controls for what a configured AI integration is allowed to receive.
Access and reuse
Use project permissions, governed sharing, scheduling and APIs to operationalise approved work.
Verification in practice
Trace every field, value and answer.
Follow fields and values through sources, joins, formulas and transformations, then inspect the query and chart logic behind every answer.
That traceability gives you the evidence to verify and defend every answer, and the confidence to make a decision you own.
Two ways to put model intelligence to work.
Use the architecture that fits the job. Both routes lead back to visible Omniscope projects and reusable analytical work.
Author inside Omniscope with the model you choose.
Use Workflow Ninja and other native AI integrations with a configured frontier provider or an OpenAI-compatible local or private endpoint.
- Author and diagnose complete workflows with Workflow Ninja
- Explore report data with Insight Explorer
- Build and refine reports with Report Ninja
- Add model reasoning to repeatable data pipelines
The outputFocused AI assistance grounded in Omniscope execution.
Give an external AI agent Omniscope as a tool.
Connect Codex, Claude Code, OpenCode or another capable agent to an Omniscope project through MCP and authoring APIs.
- Describe an objective, not a sequence of clicks
- Let the agent create, connect and configure workflow blocks
- Run the workflow, inspect outputs and correct mistakes
- Build reports and operational analytical applications
The outputAn Omniscope project a person can open and take over.
ModelInterprets the request, plans the work and proposes analytical steps.
OmniscopeExecutes queries, transformations and workflows through its deterministic engine.
HumanInspects the data, logic, assumptions, permissions and result.
ArtefactPreserves the useful work so it can be edited, rerun, shared and automated.
Choose the right model.
Use a frontier model when maximum capability matters, or a local or private LLM when data and inference need to remain under your control.
Frontier model
Configure a leading provider for focused exploration, report authoring, workflow assistance and data processing inside Omniscope.
Best for: highest-capability managed inferenceLocal or private LLM
Keep data and inference under your control. Connect a supported OpenAI-compatible endpoint for full privacy and control over infrastructure, processing and capacity.
Best for: sensitive data and controlled deployment
Private stack in practice
OpenCode, Qwen and Omniscope in one controlled analytics stack.
One implementation combines OpenCode as the agent, a privately hosted Qwen model, and Omniscope for data preparation, execution and reporting. The agent works through MCP; the result is an ordinary Omniscope workflow you can inspect, edit and rerun.
- Choose where the agent, model and data are hosted
- Keep sensitive data within a controlled environment
- Build real blocks, transformations and reports through MCP
- Open the result in Omniscope and continue the work yourself
Start with the capability you need.
Build and diagnose workflows with Workflow Ninja, connect an external agent through MCP, or explore connected data with Insight Explorer.
Build with Workflow Ninja.
Create, configure, connect, execute, diagnose and document workflows and reports directly inside the project.
Explore Workflow NinjaConnect through the MCP server.
Point a compatible agent at an Omniscope project, grant the right permissions and let it discover the available authoring tools.
Open the connection guideAsk questions with Insight Explorer.
Explore connected report data, inspect the answer and preserve useful analysis as reusable artefacts.
Read the Insight Explorer guideSee how it works in practice.
Read the product thinking, implementation experiments and guidance behind AI-assisted, verifiable analytics in Omniscope.
Omniscope as a tool for AI agents
How external agents and native AI integrations create inspectable work in the same Omniscope environment.
Read the articleOpenCode, Qwen and Omniscope
A documented end-to-end private agentic analytics stack using an open agent, a local model and Omniscope.
Read the articleWhy trust matters more than plausibility
Why Omniscope treats the LLM as the planner and the product as the deterministic execution and verification layer.
Read the articleOne complete analytics platform
How Omniscope brings workflows, analytics, automation and visualisation together for data and decisions you own.
Read the articleQuestions teams ask.
AI can accelerate the work. Omniscope keeps a person’s route back into it.
Is Workflow Ninja only an assistant that explains workflows?
No. Workflow Ninja can inspect an existing project or author one: find sources, create and configure blocks, connect them, execute the workflow, inspect outputs, diagnose problems, apply fixes, document the canvas and begin a report. Its output is an ordinary editable Omniscope project.
Does Omniscope guarantee that an AI-generated answer is correct?
No. A generated workflow can still contain a bad assumption, an incorrect join or an unsuitable calculation. Omniscope makes the work inspectable and rerunnable so people can verify it; it does not turn model output into truth by default.
Can I use an AI model that my organisation already controls?
Omniscope can work with supported frontier providers and OpenAI-compatible local, on-premises or privately hosted endpoints. Availability depends on the integration, model capability, licence and deployment configuration.
What does an external agent actually create?
Through the Omniscope MCP server and authoring APIs, a compatible agent can create and configure workflow blocks, connect and execute them, inspect outputs and build reports. The result is an Omniscope project that a person can open and edit.
How is this different from “chat with your data”?
The useful output does not have to remain in a conversation. Queries, transformations, charts, reports and workflows can be inspected and promoted into reusable Omniscope artefacts that teams can govern, rerun and maintain.
Move from raw data to decisions you own, faster with AI and without stitching tools together.
Build, explore and automate in one workspace, with the data and logic kept visible.
AI, MCP and local-model capabilities vary by Omniscope version, licence, deployment and integration configuration.
