Why a Protection Company Built a DSPM

For most of Seclore’s life, we were not in the business of finding data. We protected it. A customer told us which documents mattered, and we made sure those documents stayed protected wherever they traveled, inside the company or far outside it. Finding the sensitive data in the first place was someone else’s job, handled by some other tool, before the file ever reached us.

So when we decided to build ARMOR DSPM, the first question came from inside our own walls, and it was a fair one. Why would a protection company build a discovery product?

The honest answer is that we had already solved the hard part. Protecting data, persistently, wherever it travels, is the problem most of the industry is still trying to crack. We cracked it years ago. What changed is that AI raised the speed and the volume of how data gets used, and protection now must reach data we were never handed in the first place. Building a DSPM was how we extend the protection we already had into every stage of the data’s life, not just the stages a customer happened to point us at.

The assumption that broke

For a long time, data protection rested on a comfortable assumption: that an enterprise already knew what its sensitive data was and where it was. You knew your crown jewels. You knew where the contracts lived and which folders held intellectual property. The hard part was keeping that known, catalogued data protected as it moved. That is the part we were very good at.

AI broke the assumption. An agent connected to a repository does not work from a narrow catalogue. It reads everything it can reach, at machine speed, including the documents nobody has opened in years and nobody ever classified. The data that no one was tracking, or knew existed, is suddenly the data being used the most.

This changed the shape of the problem for our existing customers and the market. The protection we are good at depended on an upstream step that was failing in plain sight: knowing what to protect, and why it matters. We could enforce a policy beautifully on a file. But, who could be not be sure the right files were ever being identified.

We had already solved the hard part

Here is what made this different for Seclore than for a company starting from discovery. The genuinely hard problem in data security is not finding sensitive data. It is protecting it persistently, so the controls travel with the file wherever it goes and proving that protection held. That is the problem we spent fifteen years solving. ARMOR EDRM protects a file no matter where it lands. ARMOR DSI Framework produces the evidence a regulator will ask for.

Most discovery tools stop short of all of that. They find sensitive data and hand you a report. The protection, the hard part, is left to you.

So, we did not need another tool that finds data and prints a list. We needed discovery good enough to feed the protection we already had, and in context, and to do it at the speed and volume AI now demands. Not a longer queue of findings.

Discovery that understands, not discovery that matches

That requirement set a high bar for what the discovery layer had to do. A tool that recognizes patterns was never going to clear it. A string of nine digits looks the same to a pattern engine whether it is a phone number or something that would end a deal if it leaked. Feed that into automated enforcement and you protect the wrong things and annoy everyone.

So, ARMOR DSPM was built to read what data means, not what it resembles. We use the Semantic Triad: Content, Context, and Intent. It is the difference between knowing a document contains a number and understanding that the number is intellectual property, a trade secret, or a regulated record, which changes everything about how the document must be handled.

It reads three layers at once. What the data is. What it means to the business. And what the law requires given its type and jurisdiction. Those three layers are what let a finding arrive with a business priority and a remediation path already attached, instead of arriving as one more row in a spreadsheet.

And because so many of our customers operate under data sovereignty rules, the analysis runs on self-hosted models inside a controlled environment. No external API calls. No data handed to a third-party model. Nothing retained. There is no training period either. Connect a repository and classification begins, with the reasoning behind each decision shown rather than hidden.

3+ Generations of DSPM

Why now

Two things had to be true at once for this to make sense, and only recently were they both true.

The first is that discovery itself grew up. Earlier generations of these tools were not wrong. They were built for a world where finding data was the hard problem, and they matched patterns because that was the best available method. AI changed what a discovery tool can understand. For the first time, it can read meaning well enough that the output is worth wiring straight into enforcement. A DSPM built five years ago would have produced patterns. Built now, it can produce understanding.

The second is that the front of the journey became urgent. If enterprises worked from a known catalogue, slow and approximate discovery was tolerable. The moment AI agents started reaching into everything, the gap between what an organization has and what it knows it has turned into real exposure, and it widened by the week.

We had already solved the hard part. AI made it necessary to apply that protection faster, across far more data, and at every stage of the data’s life rather than only the files a customer thought to flag. A DSPM was how we got there. So, we built it to connect to the protection we already had, not to sit beside it as one more report.

What it adds up to

In the ARMOR platform, a finding moves through a single sequence: Discover, Contextualize, Enforce, and Prove. Discovery hands its context to classification enforcement through our native ARMOR DAC or other classification products like Microsoft Sensitivity Labels or Google Classification Labels. Classification drives persistent protection through ARMOR EDRM. Data heading into AI pipelines is governed by ARMOR AI-DLP before it reaches a model. ARMOR DSI Framework holds the proof of compliance, ready before a regulator asks. Every step teaches the platform more about the data, and that understanding compounds the more of it an organization runs.

That same loop is what lets a company use AI broadly without losing control of what powers it. When data is classified and protected at the source, before a model ever sees it, the data carries its own rules wherever it goes. It governs how it can be used no matter who picks it up, a person or an agent.

So, the real question was never why a protection company built discovery. It was why anyone would build discovery that stops at a report, when the protection must happen anyway. We built ours to finish what it starts.