<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Sqlite on Start AI Tools - Presented by Intent Solutions</title><link>https://startaitools.com/tags/sqlite/</link><description>Recent content in Sqlite on Start AI Tools - Presented by Intent Solutions</description><generator>Hugo</generator><language>en-US</language><copyright>Intent Solutions. All rights reserved.</copyright><lastBuildDate>Thu, 09 Apr 2026 22:22:51 -0500</lastBuildDate><atom:link href="https://startaitools.com/tags/sqlite/index.xml" rel="self" type="application/rss+xml"/><item><title>Intent Mail: Full Email Platform in a Day — Gmail, Outlook, and a Rules Engine</title><link>https://startaitools.com/posts/intent-mail-full-platform-gmail-outlook-rules-engine/</link><pubDate>Tue, 23 Dec 2025 10:00:00 -0600</pubDate><guid>https://startaitools.com/posts/intent-mail-full-platform-gmail-outlook-rules-engine/</guid><description>&lt;p&gt;35 commits in one repo. By the end of December 23rd, intent-mail was a working email platform with two provider connectors, full-text search, threaded send, attachment support, and a rules engine that could automate triage without touching the inbox until you said go.&lt;/p&gt;
&lt;h2 id="the-storage-layer-sqlite--fts5"&gt;The Storage Layer: SQLite + FTS5&lt;/h2&gt;
&lt;p&gt;Email data has two access patterns: exact lookup (give me this thread) and fuzzy search (find all emails about the deployment incident). SQLite handles the first pattern natively. FTS5 — SQLite&amp;rsquo;s full-text search extension — handles the second.&lt;/p&gt;</description></item><item><title>Local RAG NEXUS Upgrade and Nixtla Validation Infrastructure</title><link>https://startaitools.com/posts/local-rag-nexus-upgrade-nixtla-validation-infra/</link><pubDate>Mon, 22 Dec 2025 10:00:00 -0600</pubDate><guid>https://startaitools.com/posts/local-rag-nexus-upgrade-nixtla-validation-infra/</guid><description>&lt;p&gt;December 22nd was a 35-commit day across four repos. The headline: local-rag-agent shipped a complete hybrid cloud architecture in 17 commits and a versioned release. Nixtla hardened its validation infrastructure. git-with-intent migrated its grading system to SQLite.&lt;/p&gt;
&lt;p&gt;One of those days where every project moved forward.&lt;/p&gt;
&lt;h2 id="nexus-hybrid-cloud-upgrade"&gt;NEXUS Hybrid Cloud Upgrade&lt;/h2&gt;
&lt;p&gt;The local-rag-agent had been locked to a single LLM provider since its inception. Anthropic&amp;rsquo;s Claude, hardcoded. That works for a personal tool. It doesn&amp;rsquo;t work when you want to run the same RAG pipeline against different providers for cost comparison, latency testing, or compliance reasons where certain data can&amp;rsquo;t leave a specific cloud tenant.&lt;/p&gt;</description></item><item><title>Lumera-Emanuel Launch and git-with-intent Open-Source Release</title><link>https://startaitools.com/posts/lumera-emanuel-launch-gwi-open-source-release/</link><pubDate>Sat, 20 Dec 2025 10:00:00 -0600</pubDate><guid>https://startaitools.com/posts/lumera-emanuel-launch-gwi-open-source-release/</guid><description>&lt;p&gt;Two projects shipped on December 20th. One was brand new. The other opened its doors.&lt;/p&gt;
&lt;h2 id="lumera-emanuel-privacy-first-agent-memory"&gt;Lumera-Emanuel: Privacy-First Agent Memory&lt;/h2&gt;
&lt;p&gt;Lumera-Emanuel went from empty repo to working system in 11 commits. The goal: a privacy-first memory layer for AI agents that encrypts everything at rest and gives the agent structured recall without leaking user data to the host.&lt;/p&gt;
&lt;h3 id="encryption-layer"&gt;Encryption Layer&lt;/h3&gt;
&lt;p&gt;The core storage is SQLite with AES-256-GCM encryption. Every memory entry gets encrypted before it touches disk. The encryption key derives from a user-provided passphrase via PBKDF2 with a unique salt per entry. No plaintext ever persists.&lt;/p&gt;</description></item></channel></rss>