Omniscope includes several AI-enabled blocks that help you generate text, analyse data, and automate more advanced analytical tasks directly inside your workflow.

The three core AI workflow blocks are:

AI Request — for building precise AI prompts and sending one request per row or one standalone request.

AI Insights — for asking analytical questions about connected data and receiving written summaries, explanations, and answers.

AI Python — for using AI to generate and run Python-based data analysis, transformations, calculations, and custom logic.

Together, these blocks support a wide range of use cases, from simple text generation and dashboard commentary to row-level classification, data summarisation, automated analysis, and custom Python-powered processing.

AI Request block

The AI Request block gives you direct control over how AI requests are constructed and sent from your Omniscope workflow.

It can send a single AI request, or generate one request for every row in an input dataset. This makes it useful when you want to process many records individually, such as customer complaints, support tickets, survey responses, product descriptions, or internal notes.

The block supports:

• static request text
• row-level fields from the connected Requests dataset
• optional system-level guidance for tone, behaviour, and context
• optional shared context datasets
• model and output configuration
• one-off requests when no Requests dataset is connected
• row-by-row AI processing when a Requests dataset is connected

For example, you can ask the AI to summarise each row, classify each support ticket, rewrite product descriptions, extract entities, generate labels, or answer a question for each record using additional shared context.

Main uses for AI Request

Use AI Request when you need explicit control over prompt construction.

Typical use cases include:

• summarising each complaint, ticket, review, or survey response separately
• enriching records with AI-generated categories, labels, sentiment, or explanations
• answering row-level questions using a shared knowledge base
• generating content without needing an input dataset
• analysing a context dataset by asking a single question about the connected data
• combining fixed instructions with row-specific fields

A simple example would be:

“For each customer complaint, summarise the issue in one sentence and classify it as Billing, Product, Delivery, Support, or Other.”

In this case, each row becomes a separate AI request, and the output can be used downstream in reports, dashboards, exports, or further workflow processing.

AI Insights block

The AI Insights block is designed for asking questions about your data and receiving clear written answers inside your Omniscope workflow.

Where AI Request focuses on precise prompt construction and row-by-row AI calls, AI Insights is more focused on analytical interpretation. It helps you generate summaries, explanations, commentary, and answers based on connected datasets.

It can:

• generate plain-English summaries of a dataset
• answer one or more analytical questions
• produce written commentary for reports and dashboards
• support exploratory analysis without formulas or code
• compare values, trends, and patterns across inputs
• create narrative explanations based on connected data

If a Requests input is connected, the block generates one output row per prompt. If no Requests input is connected, it produces a single response.

You can also require the AI to use the connected data inputs. This is useful when you want the answer to be grounded in the actual project data rather than relying on general knowledge.

Main uses for AI Insights

Use AI Insights when you want written analysis of your data.

Typical use cases include:

• executive summaries
• dashboard commentary
• data-driven question answering
• trend analysis
• comparison across datasets
• data quality assessments
• narrative explanations alongside charts and tables
• written interpretation for reports, exports, or downstream workflows

Example questions include:

“Summarise the main trends in this sales dataset.”

“Which regions are underperforming, and what values support that conclusion?”

“Are there any data quality issues in this dataset?”

“Compare this month’s performance with the previous month.”

AI Insights is especially useful when you want Omniscope to help explain what the data shows, not just calculate values.

AI Python block

The AI Python block brings AI-assisted Python generation into Omniscope workflows.

It allows users to ask for data analysis, transformations, calculations, diagnostics, or custom processing in natural language. The block can then help generate Python-based logic to perform the task on the connected data.

This makes Omniscope more than a visual analytics and data preparation platform. It becomes an intelligent assistant that can help analysts explore, transform, and model data using Python without having to write all the code manually.

The AI Python block can help you:

• diagnose datasets and explain their structure
• identify patterns, anomalies, and possible insights
• recommend useful analyses
• perform calculations on connected data
• clean and transform datasets
• join or reshape data
• run statistical analysis
• create clusters or predictions
• generate custom Python-powered processing blocks

Example requests for AI Python

You can start with requests such as:

“Give me an overview of my datasets.”

“What useful analyses can you propose to give me more insight into my data?”

“Cluster my data into five groups and choose an appropriate clustering method.”

“Help me join these datasets using a common field.”

“Remove empty rows and columns.”

“Calculate average salary by department.”

“Detect outliers in this dataset.”

“Create a forecast based on the date and revenue fields.”

The AI Python block is especially useful when you need more advanced or customised logic than standard workflow blocks provide.

For example, you could ask it to join datasets, perform clustering, run statistical analysis, engineer new fields, or prepare data for modelling. This can save time while making advanced analytical techniques more accessible.

Important note about AI Python outputs

AI Python outputs should always be reviewed and validated.

Because the block can generate code and analytical logic, you should check that the results are based on the project data, correct calculations, and appropriate assumptions. This is especially important for statistical analysis, clustering, prediction, or business-critical outputs.