Local AI on Linux #22 — AskMyDocs | THE CAPSTONE — Everything You Built, in One Product

Master IT skills with Dargslan - No Filler, Just Knowledge.

🚀 Get our 300+ Tech & IT eBooks: https://dargslan.com

In this video, we dive deep into professional IT workflows to help you learn faster and code smarter. Whether you're a DevOps engineer, a SysAdmin, or a curious developer, our "No Filler" approach ensures you get the facts without the fluff.

🏆 **THE CAPSTONE. The moment.** This part isn't new knowledge — **it's the assembly** of everything you built across Parts 11-21: tuned RAG (Part 15) + voice (Part 18) + OpenAI API (Part 19) + memory (Part 20) + monitoring (Part 21) → **ONE real product**: **AskMyDocs**. Your documents (PDF, MD, TXT). Your machine. Your assistant. **100% local. $0/query. ~95% accuracy.** *"This used to be impossible without an OpenAI API key."*

The 16 slides cover: a 5-phase recap of the entire 24-part series (*"The runtime / The brain / The memory / The product / The launch"*), **AskMyDocs Lite vs Pro** (two-tier construction: ship Lite tonight in ~80 lines, layer in Pro features only as you need them), an **architecture diagram** (Documents → Chunker → Chroma → Tuned RAG → Ollama, plus Pro add-ons: memory.db, ai.log, /v1 API), the `askmydocs/` project layout, **`ingest.py` in 9 lines** (chunker + embedder + Chroma upsert, idempotent), **`rag.py` from Part 15** (BM25 + RRF + cross-encoder reranker + cited prompt), the **`ask.py` CLI**, a **real run** (`askmydocs -q "When does my contract with NagyKft end?"` → cited answer in 11 seconds), **"What just happened" in 5 steps** (every Part wired together), **3 Pro upgrades** (Open WebUI, memory + facts that remembers *"Laszlo runs Dargslan"*, voice loop), **deploy + share** with Cloudflare Tunnel + Bearer token auth, and the emotional peak: **"Take a moment. Look at this."** (4 stats: 100% / $0 / ~95% / YOUR).

📚 IN THIS PART YOU'LL LEARN

✓ What you've built across 22 parts — 5 phases in one image
✓ AskMyDocs Lite vs Pro: ship Lite (~80 lines) tonight, add Pro as needed
✓ Architecture diagram: how Parts 11-21 wire together
✓ Project layout: `askmydocs/` — 6 files for Lite, 3 more for Pro
✓ `ingest.py` — chunker + embedder + Chroma upsert (idempotent)
✓ `rag.py` — the Part 15 tuned pipeline (BM25 + RRF + reranker + cited prompt)
✓ `ask.py` — CLI with argparse, --json pipe-friendly mode
✓ Real run: `askmydocs -q "When does my contract with NagyKft end?"`
✓ "What just happened" — 5 steps tying together Parts 11, 12, 14, 15, 5+7+13
✓ Lite → Pro: Open WebUI, memory + user_facts, voice, monitoring
✓ Deploy: uvicorn + cloudflared tunnel + Bearer token auth
✓ The pedagogical peak: "Take a moment. Look at what you built."

⏱️ TIMESTAMPS

00:00 Intro — THE CAPSTONE
00:30 Look at what you've already shipped (5-phase recap)
01:00 Lite vs Pro — two tiers
01:30 Architecture — how it all connects
02:00 Project layout — askmydocs/
02:30 ingest.py — point at your folder, run
03:00 rag.py — the tuned pipeline (Part 15)
04:00 ask.py — the CLI you'll actually use
04:30 Real run — your docs, your questions
05:00 What just happened, in 5 lines
05:30 Lite → Pro — three drop-in upgrades
06:00 Deploy + share — public URL in 30 seconds
06:30 Take a moment. Look at what you built.
07:00 Recap + cheat sheet
07:30 Up next — Where Next

⚡ IF THIS WAS HELPFUL
👍 Like — strongest signal to YouTube
🔔 Subscribe + bell — only 2 parts left!
💬 Comment — questions, requests, feedback
📤 Share — with anyone learning AI

#localai #ollama #askmydocs #rag #capstone #ragassistant #localfirst #privacy #ai #linux #aitutorial #2026ai #dargslan #aifor beginners

The 3rd emotional peak in the series (Part 5 = first chat, Part 13 = "RAG works on YOUR docs", Part 22 = full assistant shipped).

🔔 Subscribe for weekly IT insights:
https://m.youtube.com/channel/UCv2QLrkCMSBljYG5XKd8IEA

About Dargslan:
We are dedicated to sharing high-quality, practical IT knowledge. Our mission is to provide the most efficient learning resources for the modern tech industry.

#DevOps #ITMastery #Dargslan #Linux #CloudEngineering #NoFillerTech

Видео Local AI on Linux #22 — AskMyDocs | THE CAPSTONE — Everything You Built, in One Product канала Dargslan