I saw the video on IBM AI integration and downloaded the iPaaS Buyers Guide. I was formerly a Principal Architect for Lion Bioscience. We acquired Netgenics which had been founded by Nobel Laureate Walter Gilbert. Dr. Gilbert had a vision of creating an integration platform for bio and chem informatics APIs/apps. Lion took this on calling it "Discovery Center" (see attached). Dr. Gilbert saw personalized medical research in this. A patients DNA can be used in drug discovery to find a treatment specific to that patient. Erroneous thinking seems present in this "one size fits all" approach to medicine.
Unfortunately Lion failed. The biggest hurdle was Big Pharma. For one they didn't want to share data. That gene sequence can lead to a billion dollar drug. Today we probably have satisfactory ways to handle this like we do with smart contracts and crypto currency but this was close to 2000. The other issue from Big Pharma, although no one mentioned this, it was a huge paradigm shift. They are used to discovering a drug like Viagra and sitting back and watching the money flow in. This was unfamiliar ground. Most medical research comes from Big Pharma so without their support - no go.
Things may be different today. We had a bad pandemic. Many powerful people are pushing for preparedness. This integration may have cured cancer for all we know by now. Experts seem to feel it's most worthy. Dr. Gilbert is considered by many to be the father of modern biology too. I think most of the data issues I saw then are pretty well addressed by Doug Cutting - Hadoop and data lakes, we're getting good at handling "messes" of data.
The applications however are different. Chemistry is well known the integration there is far simpler. But bioinformatics is a bowl of spaghetti. The apps are excellent for the scientists who use them but there are so many of them. Following the links from this https://en.wikipedia.org/wiki/List_of_bioinformatics_software shows many apps. Many were developed in universities using now obscure languages. The iPaaS LLM approach seems very applicable.
I was also thinking of smaller "observer models". These models would pair with agents to mine logs, observe the apps when being invoked/responding and helping to derive the integration plan from that. The LLM of iPaaS would still be the overall coordinator and collate the results to obtain the final plan.
------------------------------
John Harby
CEO
Autonomic AI, LLC
Temecula CA
9513835000
------------------------------