My friends over at Corewood are cooking up interesting solutions to some of the biggest API challenges we face. But, I am going to need your help to verify not only what they are onto with their approach to using machine learning, but also help them validate that it works for your enterprise’s approach to API operations. I got a demo of what they are up to a couple weeks back and was impressed at how they are using machine learning that differentiates itself from the current hype occurring around artificial intelligence and gets back to the basics of how machine learning is very useful, but in a very performance centered way. While Corewood is focused on using inference models to primarily tackle PII challenges, I see it having much wider potential than that.
- PII Detection - Detecting personally identifiable or PII in development and stage, but also performant enough to do in the runtime.
- PII Masking - Going beyond just detection and actually automating the masking of PII throughout all stages of API development.
- API Mapping - The potential for mapping out the source of truth for APIs within the runtime are pretty significant with this approach.
- Synthetic Data - Transforming requests and responses in useful synthetic data that can be used in API testing and sandboxes.
I get Corewood’s approach at a high level. The demo I got around PII speaks to a challenge I’ve discussed with many folks including Peter Shafton, former Twilio executive and now CTO over at Ngrok, who lead the PII work at Twilio. What Corewood is offering would help shift PII work into the next phase. This is big, but what really gets me excited is the potential for automating the mapping of the API landscape and finding evidence of APIs that exist in production, and contributing evidence to OpenAPIs via Git, helping us stay up on what the source of truth is for our APIs. Beyond the potential, I like Corewood’s approach to differentiating what they do from the rest of the AI hype, but I need your help to validate that their approach to inference models can actually be applied effectively within enterprise API operations.
