Blog
Insights on AI, security, software architecture, and building what's next for ambitious businesses.
2167 articles · Page 47 of 109
Using Haiku 4.5 for Tool Use and Function Calling: Patterns and Pitfalls
Production-grade patterns for deploying Haiku 4.5 on tool use and function calling. Prompt design, validation, cost optimisation, and failure modes.
Migrating from Mode to Superset for Enterprise Organisations
Enterprise guide to migrating from Mode analytics to Apache Superset. Covers scoping, governance, cost benchmarks, cutover patterns, and real outcomes.
Model Deprecation Risk: A 2026 Mitigation Pattern
Repeatable framework for managing model deprecation risk in 2026. Built for engineering teams to re-run on every major model release through 2027.
A Quarterly Model Refresh Cadence That Survives Contact
Models ship faster than review cycles. The eval set, the gate, and the rollback plan that let you upgrade without re-testing everything by hand.
SOC 2 in Australian Healthcare: A Practitioner's Walkthrough
SOC 2 compliance for Australian healthcare orgs: control patterns, audit timeline, common pitfalls and real evidence requirements.
SOC 2 for Insurance Startups: The Australian Path
Get SOC 2 audit-ready in 90–120 days. PADISO + Vanta Fast Track for Australian insurance startups. Scoping, evidence collection, post-audit rhythm.
Sonnet 4.6 vs Gemini 2.5 Flash: Where Each One Breaks
Latency, cost per million tokens, and tool-use reliability side by side, plus the workloads where the cheaper model stops being cheaper.
Throughput Benchmarks: TPS by Frontier Model and Region
Repeatable framework for measuring transactions per second (TPS) across frontier AI models by region. Built for engineering teams to re-run benchmarks on every major model release.
The 7-Day Model Migration Plan
Repeatable framework for migrating AI models in 7 days. Built for engineering teams to re-run on every major model release through 2027.
AI Agents in Production: Streaming Tool Use Outputs
Real patterns for streaming tool use outputs in production AI agents. Architecture, code, operational quirks, and the patterns that scale.
AI Automation Consulting Melbourne: What Buyers Actually Need in 2026
Essential guide for Melbourne leaders evaluating AI automation consulting. Covers pricing, scope, red flags, and what to demand in vendor scoping calls.
Apache Superset on Fly.io: Reference Deployment Pattern
Step-by-step production deployment of Apache Superset on Fly.io. Covers networking, storage, secrets, autoscaling, and operational best practices.
Apache Superset for Hospitality Revenue: A Reference Dashboard Set
Build revenue dashboards for hospitality with Apache Superset. Pre-built schema, metrics, drilldowns, and scalable patterns for hotels, restaurants, and venues.
Superset Chart Render Time: Finding the Real Bottleneck
Render time is usually the warehouse, not the browser. How to tell which, and the caching and query changes that fix each case.
Superset on Trino: Where Query Time Actually Goes
Split planning, worker memory, and pushdown decide performance far more than Superset config. What to measure and which knobs move the number.
Apache Superset + Trino: A D23.io Reference Architecture
Production-ready Apache Superset + Trino architecture. Connection patterns, query performance, caching, and operational deployment from D23.io customer experience.
Benchmarks That Actually Matter for New Model Releases
A repeatable framework for benchmarking AI models that delivers real outcomes. Built for engineering teams to re-run on every major release through 2027.
Choosing a Default Model for Code Generation
Framework for selecting a default code generation model. Repeatable evaluation criteria for engineering teams to re-run on every major model release through 2027.
Claude in Production: Agent Coordination
Master Claude agent coordination in production. Patterns, code, failure scenarios, and architecture for multi-agent systems at scale.
Claude Token Economics: The 2026 Cost Lever You Are Underusing
Master Claude token economics to cut AI costs by 30–60%. Real benchmarks, code patterns, and implementation strategies for 2026.