Blog · 15 articles
The latest model releases, standards, and rules — and what they change for your business. Plus the practical engineering behind agents, RAG, voice AI, and automation that pays back.
Written by the XISLABS engineering team · Updated for September 2026
Anthropic shipped Fable 5.1 and Mythos 5.1 on September 1 and made Fable 5 legacy after 12 weeks. How to tier models and plan for lifecycle churn.
Latest
Three columns, newest first. Every piece ends with what to do this quarter and how we'd approach it.
OpenAI released GPT-6 Astra on September 3, 2026, the fourth major release this year. A practical evaluation harness so you can decide in days, not quarters.
Stop picking one winner. A three-tier model strategy (frontier, workhorse, self-hosted) with routing by task, data residency, and eval-driven switching.
Chat, voice, and handoff are merging into one service layer. What a phone agent can do, where it fails, and how to roll one out by call type.
AI is shifting from chat tools to robot controllers and workflow operators. What that means for safety, latency, and where operations teams should start.
Live video try-on, consent as a selling point, AI-content labeling duties, support agents with approval gates, and a rollout order for Shopify merchants.
Why industrial sites keep video on-site, what modern computer vision does beyond recording, and how to roll it out across hundreds of cameras.
Most agent pilots die for boring reasons: no baseline, no eval set, no owner, no exit criteria. The production playbook that gets the other 12% shipped.
Pick one workflow, count volume, time each step, find the leak, name the metric, set exit criteria. The baseline method behind honest AI ROI.
Cost per task, model tiering, prompt caching, routing, batch processing, and hard caps: how to run LLM features cheaply without a quality regression.
Agent, automation, or chatbot? A decision matrix by workflow type, a cost and risk ladder, and where to start in sales ops, support, and back office.
Why RAG demos impress and production RAG fails, and the engineering that closes the gap: permission-aware retrieval, hybrid search, citations, and evals.
What the 2026-07-28 MCP spec changes mean operationally, what the adoption numbers say, and how to expose internal systems as MCP servers safely.
Three controls turn an AI agent from a liability into an operable system: approval gates, hard budget caps, and structured audit trails. How to build each.
The Digital Omnibus pushed high-risk AI Act deadlines to 2027 and 2028. Transparency, GPAI, and prohibited-practice rules still apply. Here is what to do now.
If an article here matches a problem you have, we'll scope the fastest path to a working system on a free call.