Visokio · Omniscope
Over twenty years of real-world data experience, built into one product.
I’m COO at Visokio, the independent company behind Omniscope. The product has grown through several generations, but the job has stayed recognisable: help people get from raw data to a useful result without losing control of the logic along the way.
From raw data to decisions you own.THE CONTINUING IDEA BEHIND OMNISCOPE
What we have actually been building
Omniscope began as an integrated visual data tool and gradually became an operational analytics and data-application platform. We added data preparation, workflows, web reports, APIs, scheduling, application features, private deployment and now AI agents because customers kept needing the next part of the same job.
A project can begin as an exploration, then acquire validation, parameters, automation and a focused interface until it becomes an internal tool, a branded report or a customer-facing data application. The data preparation and calculations do not need to be rebuilt somewhere else just because the audience or delivery method changed.
This is also why the current AI work is useful. A model can plan an analysis or operate tools, while Omniscope performs the joins, transformations, calculations and report changes as normal platform operations that a person can open and examine.
The platform
The work runs from source data to an operated result.
These are the six parts we repeatedly see in real projects. Omniscope keeps them in the same environment so a change in the source or logic can be followed through to the report, application or automated output.
Make the logic visible
Bring together files, databases, APIs and applications, then clean, combine and validate the data in a workflow people can inspect.
Move from questions to evidence
Profile data, investigate relationships and create interactive reports without losing the path back to the underlying work.
Turn analysis into a tool
Create focused internal tools, branded portals and embedded customer experiences on top of the same data workflows.
Run the complete process
Schedule, parameterise, call and monitor workflows so a prototype can become a repeatable production process.
Keep control of the answer
Retain visible logic, permissions, validation and deployment choices instead of accepting an opaque result.
Let models operate real tools
Use local or frontier models to plan and accelerate work while Omniscope executes checkable analytical operations.
Product evolution
What changed across the product generations.
The industry moved through visual analytics, big data, cloud, no-code and now generative AI. We rebuilt major parts of Omniscope along the way, while the capabilities accumulated: data preparation, publishing, APIs, applications, automation and model-assisted work all sit on the product today.
Visokio is founded in London
An independent software company begins with the aim of making complex data easier to work with visually. 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.
Omniscope 1.0 is released
Multiple coordinated visualisations let people explore large datasets interactively in one application.
The rebuilt Omniscope is released
After a multi-year engineering effort, the new web-based generation is publicly released on 12 November, establishing the foundation for workflows, reports, extensibility and deployment at a much wider scale.
Analytics becomes an operational platform
Working copies, workflow APIs, scheduling, multi-tenancy, OIDC, isolated Python and R extensions, mobile reports and reusable templates turn projects into governed data applications.
Automation, publishing and governance deepen
Release management, project automation, data discovery and the first AI-assisted capabilities extend the complete path from source to deployed result.
Models start operating real Omniscope tools
Report Ninja, Data Q&A, local and frontier model support, white-label AI data applications and stronger DataOps capabilities make AI useful through real platform actions.
The generated work stays visible
Models can plan, build and investigate. Omniscope keeps the resulting logic and evidence visible enough for a person to check, change and trust.
Independence gives customers real deployment choices
Visokio has been bootstrapped from the beginning. Omniscope can run in the cloud, on-premises, in private infrastructure or in controlled offline environments. Some customers want a managed setup; others cannot send their data or inference requests to an external service. We have kept those choices open because the real constraint varies from one organisation to another. When Azure and AWS both had problems on the same day, I wrote about the reassuring alternative: open the laptop and keep working with no internet, perhaps on a plane surrounded by real clouds.
My work
My role crosses operations, product and engineering.
I run operations at Visokio, stay close to product and engineering, and support customers directly when real workflows need expert help. Depending on the week, that can mean investigating a problem, building a prototype, preparing a release, making a commercial decision or testing where AI is genuinely useful.
A personal constant
My background is in computer science. More than twenty years after I started writing software, I still care about the practical consequences of design decisions once a product is in production and relied on by real people. Read the 2018 release reflection ↗
Go deeper
If you want the detail, start here.
The Omniscope timeline
How an integrated visual analytics product became a complete operational platform.
The whole data journey
Why connected data preparation, analysis, reporting and automation matter.
From workflow to data app
How the same foundation supports internal tools, branded portals and embedded analytics.
Verifiable AI analytics
What changes when models can operate a mature analytics platform without hiding the work.
New articles
Ten articles from the product work.
The two Visokio series cover the experiments behind verifiable AI, local models and agents using controlled Omniscope tools, then follow a prototype as it becomes a data application that other people can depend on.
Verifiable and private AI analytics
LLMs as planners, governed tool use, local deployment and the route from a question to an inspectable artefact.
Building complete data products
Integrated preparation, analytics and reporting; operational maturity; internal products; and the boundary between blocks and code.
Follow the work
Visokio carries the official product record. On LinkedIn I share experiments, progress and observations while the work is still moving.