A Haystack pipeline can be perfectly wired and still unsafe. The retriever returns documents. Every component did its job. But if untrusted text moved through the pipeline as ordinary context, the trust boundary was lost. That is the problem this post is about. Not bad Python. A valid component connection only says: this value fits the next component It does not say: this value is safe to influen
[05] When to Pull the Trigger on FIRE — Monte Carlo Says You're Already Free This is Part 5 of a 6-part series: Building Investment Systems with Python "You need 25x your annual expenses." That's the standard FIRE rule. For ¥9.6M annual expenses, that's ¥240M. Most people see that number and think: "I'll never get there." But the 25x rule assumes a fixed 4% withdrawal rate, zero income, zero ada
[04] The 90/10 Portfolio — Dividend Core + Growth Satellite with a Live Simulator This is Part 4 of a 6-part series: Building Investment Systems with Python In the manifesto, I described a 90/10 portfolio philosophy: 90% in dividend-growing core positions, 10% in a deep-value satellite aiming for 3-5x. Today we build both sides — the dividend snowball model for the core, and a live interactive s
[03] Designing a Personal Commitment Line — Two Loans, One Defense System This is Part 3 of a 6-part series: Building Investment Systems with Python Every major corporation maintains a revolving credit facility — a pre-arranged borrowing line they can draw from instantly during a crisis. They pay a commitment fee for the privilege of having this standby capacity, even when they don't use it. The
[02] Stress Testing Your Life — What Happens at -30%, -50%, -60%? This is Part 2 of a 6-part series: Building Investment Systems with Python After the 2008 financial crisis, regulators required banks to run stress tests — hypothetical scenarios where markets crash 30%, 40%, 60% — and prove they could survive. Your personal balance sheet faces the same risks. If you hold a securities-backed loan,
Comparison: Haystack 2.0 vs. RAGatouille 0.3 for Building High-Accuracy RAG Pipelines for Developer Docs Retrieval-Augmented Generation (RAG) has become the standard for building LLM-powered tools that answer questions using private or domain-specific data. For developer documentation (dev docs) — which includes technical jargon, versioned APIs, code snippets, and structured reference material —