InMusings of a software architectbyTony Hickman·Oct 29The Tools in Practice — Watching a Triage Conversation Flow Through watsonx Orchestrate and InstanaMy previous post, “Seeing the Whole Conversation — End-to-End Observability Across watsonx Orchestrate, Podman, and IBM Z”, described the…
InMusings of a software architectbyTony Hickman·Oct 24Seeing the Whole Conversation — End-to-End Observability Across watsonx Orchestrate, Podman, and…This is the next post in the conversational triage series. In this post I focus on a pet topic of mine — Day 2 operations with…
InMusings of a software architectbyTony Hickman·Oct 22Replacing a Mac with IBM Z: How I Hosted Laya capability and Granite on Spyre AI Accelerators and…An investigation into running the inference layer of my watsonx Orchestrate triage agent backend entirely on IBM Z — the architectural…
InMusings of a software architectbyTony Hickman·Oct 19One Agent, Many Journeys — Turning the Orchestrate Triage Agent into a Front DoorThis is my next post in the conversational triage series. This time I am focusing on the Orchestration Agent in watsonx Orchestrate.
InMusings of a software architectbyTony Hickman·Oct 16Authoring DEL Journeys in the Content Creator — From Canvas to Published TriageThis is a follow on from my “Testing DEL + Laya on watsonx Orchestrate — From Mac Mini to the Cloud” experiment, this time the focus is on…
InMusings of a software architectbyTony Hickman·Oct 11Testing DEL + Laya on watsonx Orchestrate — From Mac Mini to the CloudThis is the sixth post in the conversational triage series. The previous five are: “Configuration Driven Interactions”, “Agentic Triaging…
InMusings of a software architectbyTony Hickman·Oct 6Replacing the LLM Intent Classifier with LayaAdding a non-autoregressive encoder to the DEL triage PoC — faster classification, calibrated confidence, and a Bayesian session-opening…
InMusings of a software architectbyTony Hickman·Aug 28Teaching IBM Bob to Imagine Its Own Banking CustomersHow I took my team’s synthetic persona system — originally built around a watsonx.ai hosted model and custom tools — and migrated it…
InMusings of a software architectbyTony Hickman·Aug 4From Rules to Epistemic Logic: Rethinking Conversational TriageWhat happens when you replace a rules engine with a branch of modal logic — and let an LLM compile the journey for you?
InMusings of a software architectbyTony Hickman·Jun 4Conversational Document Transformation: Building an AI-Friendly Document Processing SolutionIntroduction