Or try paginated search →
First I socially engineer our own support agent into leaking its full system prompt and another customer's data. Then we lock it down with four layers...
Read articleWe've built agents, tools, a knowledge base, and streaming. Now how do you test it without hitting OpenAI on every run? In this episode we test the en...
Read articleYour AI platform works in development, but what happens when the provider goes down at 2am or traffic spikes? In this episode we add provider failover...
Read article
The agent can answer from our knowledge base, but not about things that change by the minute like shipping delays. In this episode we add the WebSearc...
Read article
A string isn't an email until something validates it. Aegis scaffolds the Value Object that fixes that, with validation, Eloquent casting, and a test...
Read articleThe agent's been talking to us through JSON. In this episode we build a real-time chat widget with Livewire, Alpine, and server-sent events so respons...
Read article
Last episode we built semantic search from scratch. This time we let the AI provider host the vector store and handle the search. Upload PDFs, full po...
Read article
Your agent looks up orders and remembers conversations but invents policies when customers ask about returns or shipping. In this episode we build a s...
Read articleYour agent looks up orders and pulls customer history but forgets everything between messages. Ask about order 1042, follow up with "can I return it?"...
Read article
Your support agent talks, classifies tickets, but when a customer asks "where's my order?" it makes something up. In this episode we give the agent to...
Read article
Your support agent can talk, but it responds in prose. You can't route a ticket or track sentiment from a paragraph. In this episode we build a Ticket...
Read article
Laravel 13 ships with a first-party AI SDK. In the first episode of Ship AI with Laravel we install it, scaffold a support agent, and get it respondin...
Read article