Blog
Notes from the build.
Answer engine optimisation: getting cited by AI assistants
How AI assistants and AI search choose the pages they cite, what their makers have published about crawlers, and the page and technical changes worth making.WhatsApp AI assistants for business: use cases and limits
What the WhatsApp Business Platform allows for AI assistants in late 2026: the AI provider rule, templates, the 24-hour window, pricing and how to build one.Voice AI agents: how phone automation works now
How voice AI agents handle phone calls in 2026: pipeline or speech-to-speech models, latency budgets, interruptions, telephony, tool calls and consent.LangChain and the alternatives: choosing an LLM framework
What LLM frameworks such as LangChain give you, what they cost, the current options by type, and a decision table for when to use one or none at all.Model Context Protocol (MCP) explained for business teams
What MCP is, the integration problem it solves, where the standard stands in late 2026, the security risks, and when to build your own MCP server.Small language models: when small, private and on-device wins
What small language models do well in 2026, how quantisation and distillation work, running them in a browser or on a phone, and where they fall short.Data privacy with LLMs: retention and what leaves your systems
Where data goes when you call an LLM, what main providers keep by default, and the controls on your side: minimisation, redaction, access, logs, contracts.AI agents on Cloudflare: Workers, Durable Objects, Agents SDK
How to build AI agents on Cloudflare: what Workers, Durable Objects, the Agents SDK, Workflows and Queues each do, the limits, and a support agent design.AI image generation for business: models, uses and limits
What current AI image models do reliably, the business uses that hold up, brand consistency, text in images, the legal position and a review workflow.Workers AI and AI Gateway: running and controlling inference
What Workers AI and AI Gateway each do, how they are billed, when hosted open models beat a model provider, and the limits to know before you rely on them.