# Writing - NosisTech

> All NosisTech articles, newest first.

Canonical: https://nosistech.com/blog/

- [Inside Our Free AI Acceptable-Use Policy: 9 Design Decisions and the Alternatives I Rejected](https://nosistech.com/ai-acceptable-use-policy-design-decisions/) - 2026-10-22 - The nine design decisions behind our free AI acceptable-use policy, the alternatives I rejected, and the frameworks and laws behind each rule. For IT, security, and compliance people.
- [Your Team Already Uses AI. Here Is a Free Policy That Makes It Safe](https://nosistech.com/free-ai-acceptable-use-policy-template/) - 2026-10-20 - Your team already uses AI, often without rules. This free AI acceptable-use policy, in four versions with a one-page staff guide, shows how to make it safe.
- [Your Own Private AI on a $798 Laptop: 2B Model, 91% Pass](https://nosistech.com/local-ai-router-91-percent-pass-798-laptop/) - 2026-06-13 - We built an AI router that keeps 89% of queries free and on-device. 35 tests. Real numbers.
- [4GB GPU Made Laptop an AI Workstation + Smart Cloud Routing](https://nosistech.com/4gb-gpu-made-laptop-an-ai-workstation-smart-cloud-routing/) - 2026-06-12 - A $798 gaming laptop with a 4GB GPU runs private local AI and automatically escalates the hard stuff to the cloud. Here's the test data.
- [Running Local AI on a $798 Laptop: The NosisTech Test](https://nosistech.com/running-local-ai-on-a-798-laptop-the-nosistech-test/) - 2026-06-12 - We installed a local AI on a $798 gaming laptop and ran 71 real-world tasks.
- [You Changed the Context to 64k and Hermes Still Won’t Start](https://nosistech.com/hermes-agent-64k-context-parallel-slots-fix/) - 2026-06-10 - You changed the context to 64k and Hermes still rejects it. Here is the parallel slots problem nobody is writing about yet.
- [Maxing Out Gemma 4 on a Laptop GPU + Hermes Agent](https://nosistech.com/gemma-4-12b-local-optimization-mtp-hermes-agent/) - 2026-06-10 - How I turned a 12GB laptop GPU into a 114 tok/s local AI agent with MTP, Hermes, and three config changes.
- [The Flipped Interaction Pattern: Agent Asks First](https://nosistech.com/flipped-interaction-pattern-agentic-ai/) - 2026-05-27 - Most agents execute on whatever you give them. This one asks questions first. The Flipped Interaction Pattern: an agent that asks before it acts.
- [TaskWeaver Stateful Data Analyst](https://nosistech.com/taskweaver-stateful-data-analyst/) - 2026-05-25 - A LiteLLM rebuild of TaskWeaver showing planner-to-executor handoff, stateful data memory, and safe pandas actions.
- [Aegis Constraint Agent](https://nosistech.com/aegis-constraint-agent/) - 2026-05-24 - A small LiteLLM rebuild showing deterministic constraint gates around model-assisted mission review.
- [Collective Intelligence Consensus Engine](https://nosistech.com/collective-intelligence-consensus-engine/) - 2026-05-24 - A LiteLLM rebuild that turns one model call into a small debate board with proposal, critique, adversarial review, and synthesis.
- [Knowledge Gap Discovery Agent](https://nosistech.com/knowledge-gap-discovery-agent/) - 2026-05-24 - A LiteLLM-native agent that finds product knowledge gaps, ranks evidence, and gates model-written hypotheses.
- [Smolagents Code Agent Rebuild](https://nosistech.com/smolagents-code-agent-rebuild/) - 2026-05-23 - A LiteLLM-native rebuild of the smolagents code-agent pattern with approved tool calls, AST validation, and provider tests.
- [Bayesian Operations Gateway](https://nosistech.com/bayesian-operations-gateway/) - 2026-05-23 - A LiteLLM-native operations gateway with Bayesian-style risk scoring, escalation thresholds, and an audit trail.
- [Three hidden patterns that inflate your AI API bill](https://nosistech.com/three-hidden-patterns-that-inflate-your-ai-api-bill/) - 2026-05-23 - Most AI API overspending comes from three invisible patterns. This post breaks them down with real examples anyone can follow.
- [AI System Governance Intake Agent: Practical Review Desk for AI Systems](https://nosistech.com/ai-system-governance-intake-agent/) - 2026-05-22 - An Educationla demo of AI governance intake agent that flags review areas, gathers evidence, and avoids pretending to give legal advice.
- [MetaGPT Micro Company Agent](https://nosistech.com/metagpt-micro-company-agent/) - 2026-05-22 - A small LiteLLM-native rebuild showing how roles, actions, messages, and a shared environment create an inspectable multi-agent workflow.
- [EvoAgentX Core Agent Rebuild](https://nosistech.com/evoagentx-core-agent-rebuild/) - 2026-05-22 - A LiteLLM-native rebuild of an EvoAgentX-style agent loop with one approved tool and clear execution traces.
- [General Problem Solver Agent](https://nosistech.com/general-problem-solver-agent/) - 2026-05-22 - A LiteLLM agent that verifies claims first, then escalates hard mismatches into structured reasoning.
- [Self-Improving Agent With Guardrails](https://nosistech.com/self-improving-agent-with-guardrails/) - 2026-05-21 - A LiteLLM agent that senses weak KPIs, proposes prompt updates, and routes risky self-improvement through approval gates.
- [SWE-agent Next Action Planner](https://nosistech.com/swe-agent-next-action-planner/) - 2026-05-21 - A LiteLLM educational demo rebuild of the SWE-agent action loop using simulated tools, review gates, and no real execution.
- [OpenHands Sandbox Planner](https://nosistech.com/openhands-sandbox-planner/) - 2026-05-21 - An educational demo inspired by OpenHands, showing review-gated software task planning with LiteLLM.
- [GPT Researcher Micro Research Agent](https://nosistech.com/gpt-researcher-micro-research-agent/) - 2026-05-20 - A safe educational rebuild of GPT Researcher using fictional source packets and a plan, gather, publish research loop.
- [Goose Action Planner](https://nosistech.com/goose-action-planner/) - 2026-05-20 - A safe educational rebuild of Goose as a LiteLLM-native safety-gated action planner for tool selection and review.
- [Hallucination Detection Agent](https://nosistech.com/hallucination-detection-agent/) - 2026-05-20 - An educational demo of an agent that checks AI responses against source context and flags unsupported or contradicted claims.
- [Browser Use Web Task Planner](https://nosistech.com/browser-use-web-task-planner/) - 2026-05-19 - A safe educational rebuild of Browser Use as a fictional page-state planner with local action validation.
- [PentestGPT Authorized Review Agent](https://nosistech.com/pentestgpt-authorized-review-agent/) - 2026-05-19 - A safe educational rebuild of PentestGPT as an authorized scope, review, and evidence planning agent.
- [Promptfoo Redteam Eval Agent](https://nosistech.com/promptfoo-redteam-eval-agent/) - 2026-05-19 - An educational demo of a LiteLLM-native evaluation harness that tests AI responses with visible assertions and compact verdicts.
- [Verification and Validation Agent](https://nosistech.com/verification-validation-agent/) - 2026-05-18 - An educational demo of a LiteLLM-native agent that checks AI-generated claims against trusted data before output is accepted.
- [Fabric Pattern Runner](https://nosistech.com/fabric-pattern-runner/) - 2026-05-18 - An educational demo of the Fabric pattern library idea as a small LiteLLM-native prompt runner.
- [Pydantic-AI Governance Agent](https://nosistech.com/pydantic-ai-governance-agent/) - 2026-05-18 - Educational demo of Pydantic-AI’s core pattern as a LiteLLM-native governance decision agent with structured validation.
- [LLM Guard: Building an AI Safety Gateway](https://nosistech.com/llm-guard-ai-safety-gateway/) - 2026-05-17 - A safe educational rebuild of the LLM Guard pattern as a small LiteLLM safety gateway that scans inputs, redacts data, and checks outputs.
- [Griptape Governance Pipeline](https://nosistech.com/griptape-governance-pipeline/) - 2026-05-17 - An educational demo of the Griptape pattern as a LiteLLM-native governance pipeline with policy matching, task memory, and JSON decision checks.
- [GDPR and CCPA Compliance Agent – Educational Demo](https://nosistech.com/gdpr-ccpa-compliance-agent/) - 2026-05-17 - An educational demo of a small privacy screening agent that flags GDPR and CCPA concerns from fictional data activity examples.
- [Prompt Injection Defense Agent: Screening Build](https://nosistech.com/prompt-injection-defense-agent-screening-build/) - 2026-05-16 - educaitonal demo of small prompt injection screening agent that flags suspicious text before it reaches a main AI model.
- [Agency Swarm Orchestration: Framework Makes Architecture Visible](https://nosistech.com/agency-swarm-3-agent-content-team/) - 2026-05-16 - Three-agent content pipeline using Agency Swarm, a CEO agent delegates research and writing to specialized workers through a structured communication chart.
- [Compliance-Driven Agent for AI Code](https://nosistech.com/compliance-driven-agent-ai-code/) - 2026-05-16 - AI coding agent that plans, tests, writes, and reviews generated code through visible compliance rules.
- [Explainable AI Agent: Show Your Work](https://nosistech.com/explainable-ai-agent-show-your-work/) - 2026-05-15 - AI agent that scores loan applications and explains exactly why it made each decision, using SHAP, LIME, and counterfactual analysis in pure Python.
- [AutoGen: Three Ways AI Agents Talk to Each Other](https://nosistech.com/autogen-three-ways-ai-agents-collaborate/) - 2026-05-15 - Three AutoGen agents showing how AI can hold structured conversations, run group debates, and self-correct code without human input
- [CrewAI: 3-Agent Content Crew](https://nosistech.com/crewai-3-agent-content-crew/) - 2026-05-15 - Three-agent CrewAI crew where one agent researches, one writes, and one reviews, all routed through LiteLLM with no provider lock-in.
- [Conflict Resolution Agent: When Two AIs Disagree](https://nosistech.com/conflict-resolution-agent-when-two-ais-disagree/) - 2026-05-14 - Multi-agent system - two AI reviewers independently evaluate the same claim, a third layer detects when they disagree, and a human decides what happens next.
- [Chain-of-Agents: Market Intelligence System](https://nosistech.com/chain-of-agents-market-intelligence-system/) - 2026-05-14 - Multi-agent market intelligence system - three specialist agents research any company and a manager synthesizes their findings, with built-in conflict detection.
- [Building a Self-Hosted AI Agent That Runs on a $5 VPS](https://nosistech.com/building-self-hosted-ai-agent-lessons/) - 2026-05-14 - I started this project a while ago with a simple idea: build a personal AI assistant that I could talk to, that would remember our conversations, and that would never
- [DSPy: Stop Writing Prompts, Start Compiling Them](https://nosistech.com/dspy-stop-writing-prompts-start-compiling/) - 2026-05-13 - DSPy replaces hand-written prompts with compiled programs that optimize themselves.
- [Langflow: I Built a Flow](https://nosistech.com/langflow-visual-ai-flow-builder-rfp-analyzer/) - 2026-05-13 - Installed Langflow Desktop, built an RFP analyzer by connecting four visual nodes, and called it via API.
- [Why KORVIN Removed Docker Sandboxing and What We Learned](https://nosistech.com/korvin-removed-docker-sandboxing-honest-ai-architecture/) - 2026-05-13 - Why we removed Docker sandboxing from KORVIN before this series ended, what that decision revealed about building an open-source AI tools, and what KORVIN actually does today.
- [Physical World Sensing Agent: AI That Runs a Building](https://nosistech.com/physical-world-sensing-agent-ai-building-automation/) - 2026-05-12 - AI agent that reads sensor data across three building zones, makes HVAC decisions in pure Python, and writes facility reports.
- [Planning Agent: How I Built AI That Sequences Tasks](https://nosistech.com/planning-agent-ai-task-sequencing-dag/) - 2026-05-12 - Building a planning agent that breaks any business goal into a dependency-ordered task sequence using a DAG and topological sort.
- [AnythingLLM: Build a Private AI Brain for Your Business](https://nosistech.com/anythingllm-private-ai-brain-business-documents/) - 2026-05-12 - Python client that connects to a self-hosted AnythingLLM instance, giving it private document intelligence without sending data to the cloud.
- [Data Analysis Agent: Math First, AI Second](https://nosistech.com/data-analysis-agent-math-first-ai-second/) - 2026-05-11 - Building an agent that runs real statistical analysis in Python before the AI sees a single number.
- [Marketing Content Agent: Built-In Brand Compliance](https://nosistech.com/marketing-content-agent-brand-compliance/) - 2026-05-11 - Building multi-role marketing agent that enforces brand rules automatically, rejecting non-compliant drafts and retrying until the output passes.
- [Memory, Knowledge, Tools: The Phidata Architecture](https://nosistech.com/phidata-agent-memory-knowledge-tools-architecture/) - 2026-05-11 - Rebuild the three core pillars of the Phidata agent framework from scratch in pure Python. Memory, knowledge, and tools working together.
- [Letta MemGPT: I Built an AI That Never Forgets](https://nosistech.com/letta-memgpt-ai-that-never-forgets/) - 2026-05-11 - An AI assistant persistent memory across sessions without a database or server.
- [Document Intelligence Agent: AI That Reads Invoices](https://nosistech.com/document-intelligence-agent-ai-that-reads-invoices/) - 2026-05-10 - AI agent that reads invoice images, extracts structured data, and automatically flags low-confidence results for human review.
- [Knowledge Retrieval Agent: AI That Only Knows What You Tell It](https://nosistech.com/knowledge-retrieval-agent-ai-that-only-knows-what-you-tell-it/) - 2026-05-09 - building a RAG agent that forces an AI to answer only from your private documents, and discovered a bug.
- [Memory-Augmented Agent: How I Built an AI That Remembers](https://nosistech.com/memory-augmented-agent-ai-that-remembers/) - 2026-05-09 - Building an AI agent that remembers past interactions, plans multi-step workflows, and decides when to act alone or escalate to a human. Here is how it works.
- [PraisonAI: Building a Two-Agent AI Workforce](https://nosistech.com/praisonai-two-agent-ai-workforce/) - 2026-05-09 - Building a two-agent AI workflow where a Researcher and Writer work together automatically. What I discovered about model-agnostic multi-agent architecture.
- [Flowise: Build AI Agents Without Writing Code](https://nosistech.com/flowise-build-ai-agents-without-writing-code/) - 2026-05-09 - Building a working PDF knowledge agent in Flowise using drag-and-drop blocks, no code required. Here is how it works and what I learned.
- [Async Standup Summarizer: AI Team Reports Without Meetings](https://nosistech.com/async-standup-summarizer-ai-team-reports-without-meetings/) - 2026-05-08 - building an AI agent that reads standup updates from a file and produces a consolidated team report in seconds, no meeting required.
- [KORVIN Remembers: How Persistent Memory and Voice Make an AI Agent Actually Useful](https://nosistech.com/korvin-persistent-memory-voice-ai-agent/) - 2026-05-08 - KORVIN remembers context across sessions and responds to your voice — making it the first genuinely hands-free AI security agent built for personal use.
- [Financial Advisory Agent: Market Data + News in One Briefing](https://nosistech.com/financial-advisory-agent-market-data-news-briefing/) - 2026-05-08 - Agent that combines live stock data and news headlines into a four-section advisory briefing, without generating a single financial opinion of its own.
- [Financial News Agent: AI Briefings From Live Headlines](https://nosistech.com/financial-news-agent-ai-briefings-live-headlines/) - 2026-05-08 - I built an agent that fetches live financial news for any stock and returns a plain-English sentiment briefing, without inventing a single opinion of its own.
- [Market Data Agent: Compliance-First Stock Pipeline](https://nosistech.com/market-data-agent-compliance-first-stock-data-pipeline/) - 2026-05-07 - AI agent that pulls live stock data and summarizes it in plain English, with every number traceable to a real data source, never the language model.
- [Autonomous Decision-Making Agent: AI That Triages Customer Issues Without Human Input](https://nosistech.com/autonomous-decision-making-agent/) - 2026-05-07 - Built an AI agent that scores customer issues mathematically and decides autonomously whether to resolve them or escalate to a human.
- [Tool Invocation Agent: Building an AI That Takes Real-World Actions From Scratch](https://nosistech.com/tool-invocation-agent-how-i-built-ai-real-world-actions/) - 2026-05-07 - I built an AI agent that doesn't just answer questions -- it takes real actions by calling tools. Here's how it works and why it changes everything.
- [KORVIN’s Live Dashboard: Real-Time Infrastructure Awareness From Anywhere in the World](https://nosistech.com/korvin-live-dashboard-infrastructure-monitoring/) - 2026-05-04 - KORVIN exposes real-time system health, memory, and security status through a live public dashboard secured by Cloudflare — accessible from any device, anywhere
- [KORVIN Delivers On-Demand Security Research Through Voice and Telegram](https://nosistech.com/korvin-ai-security-research-voice-telegram/) - 2026-05-01 - KORVIN answers security questions in real time through voice and Telegram, making on-demand threat intelligence accessible without a technical team.
- [KORVIN: An Open-Source AI Security Agent Built for the Real World](https://nosistech.com/korvin-open-source-ai-security-agent/) - 2026-04-29 - Meet KORVIN — a voice-controlled, open-source AI security agent built to work on modest hardware without enterprise complexity.
- [Connecting DeepSeek V4-Pro to Hostinger VPS Using LiteLLM and Open WebUI](https://nosistech.com/deepseek-v4-pro-hostinger-vps-litellm-open-webu/) - 2026-04-26 - how to run DeepSeek V4-Pro on Hostinger VPS with LiteLLM and Open WebUI
- [AI Assistant Tuning: What Broke and Finally Worked](https://nosistech.com/i-voice-tuning-modelfile-ollama-gemma-open-webui/) - 2026-04-25 - Fixing a local AI voice assistant that ignored its system prompt -- the Modelfile approach that worked
- [I Built a Private AI Voice Assistant – Runs Entirely on My Laptop](https://nosistech.com/jarvis-private-ai-voice-assistant-ollama-open-webui-windows/) - 2026-04-25 - Building a private self-hosted AI voice assistant with Ollama and Open WebUI on Windows.
- [Fixed Persistent WordPress Admin Bar Leak And Automated the Cache](https://nosistech.com/fix-wordpress-admin-bar-cloudflare-cache-litespeed/) - 2026-04-23 - WordPress admin bar visible to guests? Here is how I fixed it at the code level using WPCode, LiteSpeed, and a Cloudflare scoped API token.
- [How I Built a Google Search Console Automation With n8n](https://nosistech.com/n8n-google-search-console-automation/) - 2026-04-22 - I built an n8n automation to ping Google Search Console daily and notify me on Telegram. Here is what broke and how I fixed it.
- [Cloudflare Cache Rules, WAF, and the REST API Leak I Almost Ignored](https://nosistech.com/hardening-nosistech-cloudflare-cache-waf-rest-api/) - 2026-04-22 - Real hardening decisions on a live WordPress site -- cache rules, WAF, and a REST API leak I almost missed.
- [Securing my website Against AI Scrapers, Bots, and Brute Force Attacks](https://nosistech.com/securing-nosistech-ai-scrapers-bots-brute-force/) - 2026-04-22 - Real WordPress security hardening on Cloudflare's free plan -- what I found, what I fixed, and what I got wrong.
- [From Laptop to Cloud: Migrated My Discount Monitor to a VPS](https://nosistech.com/gcp-discount-monitor-vps-migration/) - 2026-04-21 - A few days ago I built a simple but useful automation: an n8n workflow that checks the Google Cloud certifications blog every morning . It uses a local AI model
- [I Built a Local AI Discount Monitor with n8n, Ollama, and Telegram](https://nosistech.com/n8n-ollama-gcp-cert-discount-monitor/) - 2026-04-20 - I built a local AI workflow with n8n, Ollama, and Telegram to monitor GCP cert discounts daily, here's every mistake I made and how I fixed them.
- [Securing a Self-Hosted AI Lab with Cloudflare Zero Trust](https://nosistech.com/securing-self-hosted-ai-lab-cloudflare-zero-trust/) - 2026-04-20 - Within hours of going live, my home AI lab was being probed from Bulgaria, the Netherlands, Singapore, and the US. Here is what I did about it and what I learned about securing self-hosted infrastructure the right way.
- [Self-Hosted AI Lab Accessible From Anywhere](https://nosistech.com/ollama-open-webui-cloudflare-tunnel-windows-setup/) - 2026-04-20 - A complete beginner-friendly guide to building a self-hosted AI lab on Windows Ollama, Open WebUI, and Cloudflare Tunnel configured as permanent Windows services that survive every reboot.
- [I Chose the Wrong Tool](https://nosistech.com/lm-studio-vs-ollama-always-on-local-ai/) - 2026-04-20 - I ran the benchmarks. LM Studio was faster. So I built my home AI lab on it and discovered that speed is the wrong metric when your server needs to run without you.
- [LM Studio vs Ollama on a 12GB Laptop GPU](https://nosistech.com/lm-studio-vs-ollama-rtx-5070-ti-benchmark/) - 2026-04-18 - I have an Acer Predator Helios Neo 16S with an Intel Core Ultra 9 275HX, an NVIDIA RTX 5070 Ti with 12GB GDDR7 VRAM, and 32GB of DDR5 RAM running
- [Qwen3, Sonnet, Opus, and Gemini Pro Compared](https://nosistech.com/running-local-ai-vs-cloud-giants/) - 2026-04-18 - I pulled a model, it worked, and weeks later I actually read about the architecture. Here's my honest comparison of local AI vs Claude, Opus, and Gemini Pro.
- [I Ran a Good Model With Decent Hardware](https://nosistech.com/running-local-ai-gaming-laptop-setup-2/) - 2026-04-18 - I saw a benchmark showing Qwen3.6 outperforming Claude Sonnet on my laptop. I didn't believe it. So I ran it myself — mistakes, PowerShell errors, and all.
- [I Pushed My Local AI’s Memory to the Limit](https://nosistech.com/local-ai-context-window-limit-testing/) - 2026-04-18 - Testing context window limits on local AI — what worked, what slowed down, and what crashed.
- [Running Local AI on a Gaming Laptop: Set Up](https://nosistech.com/running-local-ai-gaming-laptop-setup/) - 2026-04-18 - I set all of this up and do not fully understand everything I did. But it works, it is fast enough to be useful. That is the point of documenting
- [I Built My Own AI Server on a VPS](https://nosistech.com/ai-server-hostinger-vps-litellm-gemini-open-webui/) - 2026-04-18 - I set up my own AI server on a Hostinger VPS running Ollama, proxying Claude through LiteLLM, and eventually adding Gemini. Here's what that actually looked like.
- [How Transformer LLMs Work – What I Retained](https://nosistech.com/how-transformer-llms-work-what-i-learned/) - 2026-04-18 - Transformer LLMs, reflection, and tool use — what I took away from a technical course as a non-developer.
- [I Made My Local AI Accessible From Anywhere](https://nosistech.com/local-ai-accessible-anywhere-cloudflare-tunnel/) - 2026-04-18 - I didn't want to pay for a VPS just to access my local AI setup remotely. Cloudflare Tunnel solved it for free — and it took about ten minutes to set up.
- [VRAM vs RAM in Local AI Models](https://nosistech.com/vram-vs-ram-local-ai-models-explained/) - 2026-04-18 - When my AI model didn't fit in GPU memory, it spilled into RAM. Here's what that actually means for performance — explained without the jargon.
- [I Ran AI Models on My Laptop](https://nosistech.com/running-local-ai-models-laptop-ollama-open-webui/) - 2026-04-18 - I installed Ollama and two AI models on my Acer Predator laptop. It wasn't smooth — but it worked. Here's an honest look at what happened and what I learned along the way.
- [Agentic AI](https://nosistech.com/agentic-ai/) - 2026-04-16 - Completed the Agentic AI course by DeepLearning.AI. Key takeaways: multi-agent system design, reflection loops, tool use architecture, and what it all means for AI governance and compliance consulting.
- [AI for Everyone — For Business Leaders](https://nosistech.com/ai-for-everyone-deeplearning-ai-review/) - 2026-04-15 - Completed AI for Everyone by Andrew Ng (DeepLearning.AI). Here's what the course covers and why AI fundamentals matter for governance, compliance, and strategic decision-making.