Generative AI
Generative AI, without the mythology
A generative model continues a pattern it was trained to continue. That sentence explains both the useful demos and the failures that show up in production.
Kingson Jebaraj·10 Nov 2026·2 min read
BREAKING
Explore. Exchange. Evolve.
The desk
Publication dates are when CloudExchangers published the piece. Briefings are short. Longer formats carry the technical argument.
Generative AI
A generative model continues a pattern it was trained to continue. That sentence explains both the useful demos and the failures that show up in production.
Kingson Jebaraj·10 Nov 2026·2 min read
Kubernetes and Containers
The object that serves traffic is less interesting than the requests, the probes, the identity and the question of what happens when it is killed.
Kingson Jebaraj·10 Nov 2026·2 min read
Hybrid and Multi-Cloud
More than one environment is justified by residency, latency, or a system you cannot move yet. It is not justified by a fear of choosing.
Kingson Jebaraj·10 Nov 2026·2 min read
Robotics and Automation
A robot in a video has not met your floor, your safety case, or your exception path. Those are the product.
Kingson Jebaraj·10 Nov 2026·1 min read
Quantum Cryptography
One project replaces public-key algorithms with new classical ones. The other uses quantum physics to share a key. They solve different problems.
Kingson Jebaraj·9 Nov 2026·2 min read
RAG, KAG and Knowledge Graphs
Retrieval-augmented generation only helps if the right passage arrives and the model is willing to admit when it did not.
Kingson Jebaraj·9 Nov 2026·2 min read
Cybersecurity
NIST’s definition is about removing implied trust. A product that uses the phrase has not, by itself, done that.
Kingson Jebaraj·9 Nov 2026·1 min read
AI Agents and Multi-Agent Systems
Splitting work across several model roles can clarify a task. It can also create a committee that nobody can debug.
Kingson Jebaraj·8 Nov 2026·2 min read
AI Infrastructure and GPU Computing
Training and inference do not want the same cluster. The platform questions are utilization, isolation, failure and who can spend the hardware.
Kingson Jebaraj·8 Nov 2026·2 min read