AI Workflows

Moltis (131 HN points) showed the world what an AI assistant with memory, tools, and self-extending skills looks like. This workflow builds the same architecture — a LangGraph agent that creates its own tools, grows its skill set, and persists everything across sessions.

A real-world AI agent in 2026 ran a DN42 scan loop that bankrupted its operator's cloud account in hours. This workflow builds a self-healing cost control system that enforces token budgets, detects cost anomalies, and circuit-breaks runaway agents before they burn cash.

Pipelex (122 HN points) introduced a declarative language for repeatable AI workflows. This build walks through creating production-ready declarative agent pipelines — define your workflow in YAML, run it with a single command, and reuse pipelines across any agent task.

Goose (53,000+ GitHub stars) is an open-source extensible AI agent that goes beyond code suggestions to install dependencies, execute commands, edit files, and run tests autonomously. Build a LangGraph workflow that extends Goose with custom tools and multi-model orchestration.

NanoBot (47,000+ GitHub stars) is the ultra-lightweight, self-hosted AI agent framework in Python with WebUI, tools, memory, MCP, and multi-agent orchestration. Build a LangGraph workflow that extends NanoBot with persistent memory and MCP tool chaining.

Headroomlabs' Headroom compresses tool outputs, logs, files, and RAG chunks before they reach the LLM — achieving 20% fewer tokens for coding agents and 60-95% fewer for structured data. Build a LangGraph workflow that wraps any agent pipeline with Headroom compression for instant cost and latency savings.

WebLLM by mlc-ai runs 27B parameter models entirely in-browser via WebGPU. Build a LangGraph agent workflow that processes sensitive data clientside with zero server costs and zero data egress.

Google's Gemini 3.8 Flash Cyber scored 863 points on Hacker News with its cyber-security-first architecture. Build a LangGraph workflow that autonomously scans codebases, enriches CVE data, and generates verified patches.

Meta's Muse Spark 1.3 hit 429 HN points with real-time text-to-image at 512x512 in 0.8 seconds. Build a LangGraph workflow for prompt engineering, iterative refinement, and production visual asset validation.

Agentic web research is replacing manual search-and-copy workflows in enterprises. This workflow builds a production pipeline using Firecrawl for reliable web scraping, LangGraph for multi-stage orchestration, and structured synthesis with source verification. Results: 71 percent faster research cycles with citation-verified outputs.

Most agent architectures are request-response and synchronous. This workflow builds a streaming agent architecture using WebSockets for real-time client push and Kafka for event-driven agent-to-agent communication. Achieves sub-100ms end-to-end latency for real-time agent applications.

Single-vector RAG hits a ceiling at approximately 72 percent answer accuracy. This multi-agent pipeline combines three retrieval agents — vector search, Cross-Encoder reranking, and knowledge graph traversal — with a judge agent that selects the best answer. Achieves 52 percent higher accuracy than single-vector RAG in production benchmarks.