Visokio · Product history
More than twenty years of building and rebuilding Omniscope.
Omniscope looks very different from the first release in 2005. We moved from a desktop visualisation product to a web-based platform for workflows, reports, applications, automation and AI-assisted data work, while keeping the work connected inside one environment.
Visokio is founded
Visokio is founded in London; Omniscope 1.0 follows in 2005. Its name combines “vis”, from visualisation, with “okio”, echoing “occhio”, the Italian word for eye. The aim is still to help people see and understand their data. See the current company history ↗
Omniscope 1.0
The first release brings multiple coordinated visualisations into one desktop application. Selections in one view flow through the others, turning data exploration into an interactive process. See the archived official history ↗
Classic grows beyond visualisation
Omniscope Classic develops into an integrated environment for data import, transformation, analysis, scripting, interactive reporting and browser publishing. The product establishes the principle that data preparation and presentation should remain connected.
The platform is rebuilt for the web
After three years of engineering work, the new generation of Omniscope is publicly released on 12 November 2018. At the time I wrote about the former and current colleagues who had built it together, and admitted that I would never stop being a developer because creating software can make you feel a little invincible. The release’s web reporting, workflow and extension architecture became the foundation for server deployment, collaboration and a much wider class of data applications.
Workflows and reports become one system
Parameters, localisation, JSON and XML handling, improved workflow and report design, and plans for custom Python and R blocks make the new architecture increasingly extensible.
Projects become operational services
Working Copies separate editing from live production. The Workflow Execution REST API and scheduler let other systems run parameterised workflows. Multi-tenancy, data editing and external storage support move Omniscope beyond interactive analysis into repeatable operations.
Authentication, templates and isolated Python/R
OIDC authentication, publishing controls, project templates and performance work make deployment more systematic. Custom blocks can run Python and R in isolated Docker environments, combining no-code assembly with specialist code where it adds value.
The data-application platform becomes explicit
Mobile reports, streaming and capture, richer connectors, profiling and reusable blocks support a clear proposition: build complete data web applications by combining ETL, analytics, visualisation, automation and optional Python, R or JavaScript in one product.
From deployed project to managed product
Improved data editing, templates, automated diagnostics, publishing, revision history, release management, staging-to-production synchronisation and multi-project scheduling support longer-lived operational solutions. The first AI integrations arrive inside that established platform.
Models start operating real platform tools
Report Ninja, Instant Dashboard and Data Q&A let models construct reports and analyses using visible Omniscope operations. Bring-your-own-model support, private models, white-label AI applications, schema resilience and DataOps work broaden where the platform can be used.
Verifiable, model-neutral AI
Insight Explorer, Workflow Ninja, AI Request and AI Insights expand the model’s reach across exploration, preparation and reporting. Support for multiple providers and local models keeps deployment flexible, while the generated transformations and outputs remain visible for review.
How the releases connect
The first coordinated views made it easier to explore a dataset directly. Workflows connected preparation to analysis; web publishing and APIs connected projects to day-to-day operations; application features gave the same logic a focused interface. AI is the latest interface into those accumulated capabilities.
An agent is useful in Omniscope because it can operate tools we have spent years building. Its work is verifiable for the same reason: joins, calculations, workflows and reports already have visible representations in the platform.
Continue