Symbolic Engineering

About Symbolic Engineering

When company data loses its meaning.

When I was still just a child, I encountered this section in a history book presenting Linear B. It was the name given to the language of the ancient Mycenaean. It went for a long time undeciphered, but to me what made it fascinating wasn't just the mystery, but how its meaning got eventually recovered. Now that I work with companies from different domains, we have modern Linear B everywhere.

A clay tablet whose signs become legible when connected to ordinary places and records

The tablets of this day and age are old databases, data lakes and sometimes even brand-new data warehouses. The problem is not necessarily with the tools, but the series of accidents that lead to an information loss: the semantics is not encoded anywhere.

I faced this loss of information many times. The answer is hardly ever to record more, although that often helps helps. The real challenge lies on recording with greater precision what the business actually mean. Facts without their meaning and context have become normal, but in a world where there is an ever increasing expectation to give more context and leverage the large information processing capabilities from AI agents, this is unnaceptable.

For this reason that I started building Symbolic Engineering. My goal is to leverage logic and symbolic manipulation to unlock value from stale, ambiguous and underused data. Contour, our adaptable software, is this discipline productized, not a generic tool expecting your organization to adapt around it, but a non-intrusive tool that helps you make your demo, that at first might have worked, but now struggles to be deployed in production, finally show results.

Your hardest questions shouldn't share the fate of Linear B, waiting centuries to be answered. You should be able to ask them now, ad get answers that still cary their meaning.

Company records without semantics and with Contour FACTS WITHOUT SEMANTICS Databases, lakes and warehouses Facts recorded Meaning not encoded Question waits or agent guesses WITH CONTOUR Existing records Business semantics encoded Logic + symbolic operations Answer with meaning + context
Figure 1Records survive in databases, lakes and warehouses, but their meaning is often absent. Contour adds the semantic layer that lets questions return answers with their context still attached.