Is it Secret, Is it Safe?

Strategic data handling in a complex world

In the past year we have worked with many different AI related topics and use cases. Both internally and for different clients. In all the hype, one thing stands out in terms of what you need to succeed with AI, machine learning and data driven projects. That is having a thought through, strategic approach to storing, structuring and protecting your data. 

It’s about being able to answer questions like: 

• What is the core, strategic data that you need to run your business? 

• Where is that data stored and who owns those systems? 

• Does everyone who needs that data to get the job done understand it, and have access to it? 

• Are there protections and access control in place to make sure the data stays with the people who need it and does not leak to malicious actors? 

In a world where everyone tries to follow the tech companies lead in data, why do we still see a lot of organizations treating data strategy as a technical problem instead of a business problem? 

Do you, as a business leader, know where the data that is core to running your business exists, right now? 

When the strategic vision is limited to lofty words about being data driven or being an early adopter of AI everywhere execution becomes a purely technical problem. IT and engineering selected a technical solution and started storing data somewhere. In digital companies that is likely a big US cloud provider, something that suddenly carries political risk. In more traditional organizations it resides in siloed, on-prem solutions with varying security. 

Did you have a strategic discussion to keep core data in one place or is it spread out across multiple providers and Software as a Service (SaaS) solutions? And is your way of interacting with those SaaS solutions primarily through web-UIs or through data heavy APIs? This touches on the SaaS is dead theme from last year that even included Satya Nadella for a brief stint. Beyond the fact that AI agents will interact with data differently, SaaS may also block you from combining your data in new and innovative ways. 

In the days of big data, access controls were an afterthought. In a future of value creating AI agents, it needs to be built in at the core. 

Existing big data platforms evolved from the tech and consumer application world. They are built to ingest billions of click stream events, logs and other data with the primary goal of building ad profiles or recommending content. In those organizations, the data is usually available to a lot of users across the org, in order to move fast and innovate. Only after more traditional industries starting to onboard to big data and regulations like GDPR appeared did this start to change. Add AI tools on top of these platforms and we have concerns about leaking proprietary data.

Data platforms today have taken leaps forward in terms of properly cataloguing and classifying data, combined with access controls. To let your users and AI agents thrive on top of these evolving systems, mapping your business value in data terms needs to be an integral part of your strategy. 

Whether to implement data meshes, data fabrics or a standard database will depend on your organization and business needs. But strategic decisions about your core data must drive the technical choices. 

 Fitting the technical pieces together; Some assembly required.

From recent implementation projects we know that all technical pieces that enable secure innovation with data exists, but they do not always fit well together. It is easy to mess up the puzzle by adding blanket permissions or throwing in unvetted AI tools. At the other end of the spectrum, many projects never leave the idea stage due to fear of regulations. Even when customer knowledge is fundamental to the business and technical tools that handle regulatory concerns do exist. 

That said, we are seeing an ecosystem evolving that is able to combine good metadata, access rules and ease-of-use. Regardless of your maturity you can get started today, some assembly required. Based on exploratory projects we are involved in, we are also getting very close to easy-to-use data platforms that can be geographically localized and have strong security built in from the start.

So: get started thinking about that crucial piece of data that drives your business and what you want to do with it.