Optiq emerged from a recurring annoyance encountered while dealing with actual codebases. Writing the code wasn’t the problem. Understanding it was. The functions we want aren’t always on hand. When we need to find out how something is connected and reference something across files, that often takes forever. And figuring out why something works the way it does takes up much more time than it should. A majority of software assists in code generation, while the less fortunate aid in automating existing systems. Initially, we took the same approach as all but one, trying to create a smarter AI layer. It didn’t work well. In some cases, it made issues more problematic by adding noise, latency, and inconsistency. We redirected our attention. We began to focus on context itself rather than more AIs. When programming, you want to structure the code to make it clear how different parts of a program relate to one another. It also needs to be easy to pull the right parts of a program together without too much effort. This resulted in developing a context engine that treats relationships and structure as first class, rather than just surface similarity. The process was messy. We refined the system repeatedly, eliminated pleasing concepts that proved impractical, and streamlined numerous overengineered features to be simpler and easier to use. Most of the advancement arose from being aware of what not to do. The current system shows greater levels of focus than before. The only idea is that if we can get the right context, then everything else is easy. That’s what Optiq is focused on.

Optiq Code

Context engine for accurate code retrieval

Context engine for accurate code retrieval