Sandra Serkes

The unglamorous superpower of knowing what you actually have.

Every large organization is quietly drowning. Decades of contracts, emails, scanned paper, shared drives that nobody has opened since a forgotten reorganization. Most leaders treat it as wallpaper. Sandra Serkes treats it as the whole job. As co-founder, president and CEO of Valora Technologies, she runs a company whose software reads that mess, recognizes patterns inside it, and tags each file with what it is, what it concerns, and whether anyone is still allowed to keep it.

The category has a dry name: information governance. Valora calls its specialty autoclassification. In practice it means answering a deceptively hard question at industrial scale - what is in here, and who knew what, when? Lawyers ask it during litigation. Compliance officers ask it when a privacy regulation lands. IT teams ask it the week before a cloud migration. And lately, everyone asks it again before they feed their archives to an AI model and discover, too late, what was hiding inside.

Serkes has been answering that question since before "AI readiness" was a phrase anyone put on a slide. Valora's flagship engines, PowerHouse and BlackCat, scan files, apply customized classification tags, and turn an opaque pile into something searchable, defensible and safe to act on. The work is invisible when it goes well, which is most of the time, and that suits her. She would rather the data behave than be admired.

It started with a problem nobody could link.

The founding insight was almost embarrassingly simple, which is usually the sign of a good one. Serkes and co-founder Aaron Goodisman, Valora's CTO and chief architect, kept running into the same wall: information that obviously belonged together sat in separate systems with no way to connect it. The relationship was clear to any human who looked. The machines had no idea.

So they built one. "We built a software engine to identify patterns in files and associate related information," she explains. The first customers who felt the pain most acutely were lawyers, buried under document populations so large that the only path through was a small army of reviewers or a smarter machine. Valora bet on the machine, and on the idea that pattern recognition could surface the story inside the stack faster than people flipping pages.

Erratic revenue. Cheaper rivals. Copycats. She productized her way out.

The early years were not a clean ascent. Valora lived on project work, which meant feast-and-famine cash flow and a sales cycle that started over with every engagement. Then competition arrived from two directions at once: offshore labor offering manual review at a price Valora could not match, and newer technologies promising to do what Valora's engine already did.

Her answer was structural. Rather than chase lower prices, Valora turned its capabilities into products - SaaS, cloud and on-premise installations - and kept reinvesting in the core engine so it stayed harder to copy than to buy. The result reshaped the business: roughly 70% of Valora's work is now recurring rather than one-off, the kind of revenue that lets a small company plan instead of scramble. The reputation built over more than two decades became its own moat, the thing a six-month-old competitor could not manufacture.

It is a founder's lesson learned the slow way, and she states it plainly: the obstacles were "erratic, project-based revenues, and the rise of competition," and the way through was to keep "investing in the technology to make it stronger." No magic. Just a refusal to compete on the one axis where she would always lose.

CEO by day, professor by semester.

Since the fall of 2015, Serkes has taught at Columbia University's School of Professional Studies in its Knowledge Management program. Her course title is a mouthful that doubles as her entire worldview: solving information governance and knowledge management problems with pattern-recognition technologies and predictive analytics. She is teaching the next generation of governance professionals the exact discipline she has spent her career commercializing.

It tracks with how she shows up in the wider field - as a frequent speaker and panelist at the conferences where this work gets argued over. KMWorld. ARMA. The Info Gov Hot Seat podcast, where in 2025 she turned up to talk about AI adoption and why senior practitioners still need to get serious about it. She is fluent in three dialects at once: the law, the IT department, and the engineers who actually build the models. Translating between them is, quietly, half of what she does.

From speech recognition to GDPR readiness.

A long game, played on purpose.

There is a temptation, with founders who have run the same company for 25 years, to read it as stubbornness. With Serkes it reads more like conviction with a spreadsheet attached. She picked a problem that only gets bigger - more data, more regulation, more reasons to know exactly what you hold - and she stayed put while the world slowly agreed with her. Privacy laws made governance mandatory. Cloud migrations made it urgent. AI made it existential. Each wave washed more customers toward the shore she was already standing on.

Her aspiration is not subtle, and it does not need to be: make defensible, automated data management the default, not the heroic last-minute scramble. Help teams know precisely what to keep, what to retire, and what to protect, so that the next migration or lawsuit or model-training run does not turn up an unpleasant surprise. The "dark data" that companies generate and forget is, to her, not a nuisance. It is the work.