Suggestions
Zehra Cataltepe
AI and GenAI solutions for Customer Facing Teams (Marketing) and Legal Teams
Zehra Cataltepe is the CEO and co-founder of TAZI AI Systems, a company that provides an ML platform for creating predictive solutions.1 She is an accomplished professional in the field of artificial intelligence and machine learning, based in the San Francisco Bay Area.1
Key points about Zehra Cataltepe:
-
Professional achievements:
-
Academic background:
-
Previous experience:
Zehra Cataltepe's work focuses on helping companies increase profitability, customer outreach, and retention through AI-driven solutions.1 TAZI AI's platform is recognized as a Gartner Cool Vendor and offers continuous learning, interactive explanations, and human-in-the-loop capabilities.1
Highlights
Here are three thoughts/advice on your AI deployments, after the Crowdstrike update incident:
1- Diverse Models: Have diverse AI algorithms and models, e.g. not just a single LLM provider. Always have a local version of an open-weight model, that you continuously test and verify and use in case when you need it. 2- Have humans on the loop: AI, especially agentic workflows, should always have humans on the loop who can control and correct easily. Failure of a point at the heart of a system (e.g. at operating system kernel level), could be very difficult to remedy. In order to prevent and correct such mistakes, have AI involvement in your business processes at the user level, allowing humans to be on the loop and have ways of intervention. 3- QA testing principles, automated and manual, by human agents, not just AI agents, apply to AI deployments. All discriminative and generative models in your environment should have easy roll back scenarios to an earlier version.
When designing any AI based solution, everybody thinks about data readiness first. IMHO, inclusion of domain experts comes even before data, for ownership, value, efficiency and risk reduction. Humans have to be -in the loop (to make the final decision), -on the loop (to design and monitor those systems), -in command (if needed to stop the AI based solution). I have seen some awesome humans in and on the loop.
Please share your experiences as a human in the loop, on the loop or in command and the difference you made.
Thanks!
https://lnkd.in/gqbFjE97
