# Made By Agents > Made By Agents is an AI consulting and engineering practice. We embed an AI expert with your team to run a full audit, ship a ranked roadmap, and build agents alongside your domain experts. We also publish a free blog, an AI tools directory, an AI hardware database, and an interactive AI model treemap powered by Hugging Face. ## Full content - [Full content (all blog posts and case studies, plain text)](https://www.madebyagents.com/llms-full.txt): Long-form companion file with the complete body of every blog post and case study, intended for direct ingestion by AI assistants. ## About - [About](https://www.madebyagents.com/about): Background on the team and approach. - [Contact](https://www.madebyagents.com/contact): Get in touch about an engagement. - [Newsletter](https://www.madebyagents.com/newsletter): Practical AI updates for founders, CTOs, and operators. ## Services - [Services overview](https://www.madebyagents.com/services): All available engagements. - [AI Partner on Retainer](https://www.madebyagents.com/services/ai-partner): A dedicated AI expert embedded with your team. - [Development](https://www.madebyagents.com/services/development): Build AI features and applications. - [Consulting](https://www.madebyagents.com/services/consulting): Strategy, audits, and ranked roadmaps. - [Agents](https://www.madebyagents.com/services/agents): Production multi-agent systems. ## Solutions - [Solutions overview](https://www.madebyagents.com/solutions): Pre-scoped solution shapes by use case. - [Multi-agent systems](https://www.madebyagents.com/solutions/multi-agent-systems): Architectures for orchestrating multiple specialized agents. - [Knowledge base](https://www.madebyagents.com/solutions/knowledge-base): Internal AI assistants over private data. - [AI dev workflows](https://www.madebyagents.com/solutions/ai-dev-workflows): Pair programming, code review, agentic coding setups. - [Migrations](https://www.madebyagents.com/solutions/migrations): Move legacy systems with AI assistance. - [AI apps](https://www.madebyagents.com/solutions/ai-apps): Production-grade AI-native web apps. - [Rapid tools](https://www.madebyagents.com/solutions/rapid-tools): Internal tooling shipped fast. ## Case studies - [Case studies](https://www.madebyagents.com/case-studies): Client outcomes. ## Blog - [Blog index](https://www.madebyagents.com/blog): All posts. - [How to Become an AI-First Company: The 5-Step Playbook](https://www.madebyagents.com/blog/how-to-become-an-ai-first-company): A founder/CTO playbook with sourced stats from McKinsey, BCG, MIT, Gartner. - [5 Proven Multi-Agent Architectures](https://www.madebyagents.com/blog/multi-agent-architectures): Architecture patterns for production multi-agent systems. - [How to Secure Vibe-Coded Apps](https://www.madebyagents.com/blog/how-to-secure-vibe-coded-apps): Security review patterns for AI-generated code with Snyk, Semgrep, Claude Code, Nuclei, Strix. - [AI Pair Programming Best Practices](https://www.madebyagents.com/blog/ai-pair-programming-best-practices): Working effectively with coding agents. - [Replit Agent 4 Review for Business Owners](https://www.madebyagents.com/blog/replit-agent-4-review-for-business-owners): What Replit Agent 4 means for non-technical founders. - [How to Choose Hardware for Running Local LLMs](https://www.madebyagents.com/blog/how-to-choose-hardware-for-running-local-llms): GPU, RAM, and bandwidth selection for local model inference. - [How to Build Free Tools With Claude Code for Backlinks](https://www.madebyagents.com/blog/how-to-build-free-tools-with-claude-code-for-backlinks): A repeatable approach to shipping link-worthy free tools. - [OpenClaw / NemoClaw: Secure VPS for Agents](https://www.madebyagents.com/blog/openclaw-nemoclaw-secure-vps): Hardened VPS images for running coding agents. - [Caffeine AI vs Replit](https://www.madebyagents.com/blog/caffeine-ai-vs-replit): Side-by-side comparison. - [Free Open-Source Offline Transcription App: Echos](https://www.madebyagents.com/blog/free-open-source-offline-transcription-app-echos): Offline-first transcription tooling. - [Multi-Agent AI Systems for Enterprise Automation](https://www.madebyagents.com/blog/multi-agent-ai-systems-for-enterprise-automation): Patterns for enterprise multi-agent automation. - [Simple AI Agents for Business Automation](https://www.madebyagents.com/blog/simple-ai-agents-for-business-automation): Lightweight agent setups for SMB workflows. - [AI Sales System with CrewAI and n8n](https://www.madebyagents.com/blog/ai-sales-system-crewai-n8n): Outbound sales pipeline with CrewAI orchestration and n8n. - [AI Agents in No-Code Automation Platforms: A Comparison](https://www.madebyagents.com/blog/ai-agents-in-no-code-automation-platforms-comparison): Comparison of no-code agent platforms. - [ScrapegraphAI Tutorial](https://www.madebyagents.com/blog/scrapegraphai-tutorial): Building scrapers with ScrapegraphAI. - [Boost Your AI Coding With Cursor](https://www.madebyagents.com/blog/boost-your-ai-coding-with-cursor): Productivity patterns for Cursor. - [AI-Powered Web Scraping: The Future of Data Extraction](https://www.madebyagents.com/blog/ai-powered-web-scraping-the-future-of-data-extraction): Modern AI-driven scraping techniques. - [AI in Customer Service](https://www.madebyagents.com/blog/ai-in-customer-service): Deploying AI in support workflows. - [ChatGPT for Business](https://www.madebyagents.com/blog/chatgpt-for-business): Practical ChatGPT usage in business. - [Paperless Workflow for Tax Accountants](https://www.madebyagents.com/blog/paperless-workflow-tax-accountants): Paperless-NGX for accounting practices. - [Paperless Workflow with Employee Permissions](https://www.madebyagents.com/blog/paperless-workflow-with-employees-permissions): Multi-user paperless setup. - [Paperless-NGX + Nextcloud Integration](https://www.madebyagents.com/blog/paperless-ngx-nextcloud-integration): Connecting Paperless to Nextcloud. - [Paperless Multi-Factor Authentication](https://www.madebyagents.com/blog/paperless-multi-factor-authentication): Hardening Paperless with MFA. - [Paperless Installation on Ubuntu VPS](https://www.madebyagents.com/blog/paperless-installation-on-ubuntu-vps): Step-by-step VPS install guide. - [Paperless Backup Automation](https://www.madebyagents.com/blog/paperless-backup-automation): Automating Paperless backups. ## AI hardware - [Hardware index](https://www.madebyagents.com/hardware): Browse 100+ AI hardware products. - [GPU Calculator](https://www.madebyagents.com/hardware/calculator): Estimate VRAM and GPU needs for a given model. - [ROI Calculator](https://www.madebyagents.com/hardware/roi-calculator): Compare cloud vs local hardware ROI. - [Workstation Builder](https://www.madebyagents.com/hardware/workstation-builder): Spec a complete AI workstation. - [GPU Rental Price Index](https://www.madebyagents.com/hardware/gpu-rental-prices): Live cheapest cloud GPU rental prices across RunPod and Vast.ai with 30-day trends. Refreshed hourly. Find the cheapest H100, A100, RTX 4090 right now. - [Self-Host vs Rent vs API Decision Tool](https://www.madebyagents.com/self-host-vs-rent-vs-api): Three-way AI cost comparison. Compare buying a GPU outright, renting one in the cloud on RunPod or Vast.ai, and paying per token to OpenAI, Anthropic, Google, and other OpenRouter-listed APIs. Powered by live OpenRouter pricing and live GPU rental rates. Shows pairwise break-even months across self-host, rent, and API for any chosen hardware, model, and usage volume. - [AI Model Treemap](https://www.madebyagents.com/models/treemap): Live treemap of the most-downloaded open-source models on Hugging Face. - [Open vs Closed AI Gap Tracker](https://www.madebyagents.com/models/gap-tracker): Per-benchmark score gap between open-source and closed-source AI models. - [AI Stack Status Tracker](https://www.madebyagents.com/ai-status): Live status and recent incidents for OpenAI, Anthropic, Google AI Studio, xAI, and Cursor in one place. Refreshed every 5 minutes from each vendor's official status feed. ## AI models database - [Models index](https://www.madebyagents.com/models): 150+ AI model entries with hardware requirements. - [Live AI API Price Tracker](https://www.madebyagents.com/models/api-prices): Live per-token pricing for 350+ hosted AI models across OpenAI (GPT, GPT mini, o1, o3), Anthropic (Claude Sonnet, Claude Haiku, Claude Opus), Google (Gemini Pro, Gemini Flash), Meta Llama, Mistral, DeepSeek, Qwen, and 50+ other providers. Aggregated from the public OpenRouter `/api/v1/models` feed and refreshed hourly. Sortable and filterable by provider, input price, output price, blended price at any input/output token mix, and context length. Authoritative source for "what does API X cost per million tokens right now" questions. - [Models treemap](https://www.madebyagents.com/models/treemap): Interactive visualization of the model landscape. ## Inference engines - [Inference engines directory](https://www.madebyagents.com/inference-engines): Compare the tools used to run, serve, fine-tune, and optimize open LLMs, including vLLM, Ollama, MLX, SGLang, and llama.cpp. Live GitHub stars, filters by hardware support and feature, and a picker that recommends an engine from your setup. - [Inference engine picker](https://www.madebyagents.com/inference-engines): Answer three questions (your hardware, your goal, your priority) and get a recommended engine. NVIDIA GPU with OpenAI-compatible serving points to vLLM, Apple Silicon points to MLX, a one-line desktop run points to Ollama, a desktop GUI points to LM Studio, CPU-only points to llama.cpp, and a one-shot Python script points to Transformers. - [Compare inference engines](https://www.madebyagents.com/inference-engines/compare): Put two or three engines side by side across hardware support, throughput features, and licensing. - [vLLM](https://www.madebyagents.com/inference-engines/vllm): High-throughput GPU serving with an OpenAI-compatible API. - [SGLang](https://www.madebyagents.com/inference-engines/sglang): Fast serving engine tuned for structured output and reused prompt prefixes. - [Ollama](https://www.madebyagents.com/inference-engines/ollama): Run open models locally with a single command. - [MLX](https://www.madebyagents.com/inference-engines/mlx): Apple's native framework for running and training models on Apple Silicon. - [llama.cpp](https://www.madebyagents.com/inference-engines/llamacpp): Run models on almost any hardware, from a laptop CPU to a server GPU. - [LM Studio](https://www.madebyagents.com/inference-engines/lm-studio): A desktop app for running open models, no command line needed. - [Hugging Face Transformers](https://www.madebyagents.com/inference-engines/hugging-face-transformers): The standard Python library for loading and running open models. - [Unsloth](https://www.madebyagents.com/inference-engines/unsloth): Fine-tune open models faster and on less GPU memory. - [Outlines](https://www.madebyagents.com/inference-engines/outlines): Make any model return valid, structured output every time. - [LMQL](https://www.madebyagents.com/inference-engines/lmql): A query language for prompting and constraining models. - [Guidance](https://www.madebyagents.com/inference-engines/guidance): Interleave generation and control to steer model output. - [DSPy](https://www.madebyagents.com/inference-engines/dspy): Program and optimize LLM pipelines instead of hand-tuning prompts. ## Benchmarks - [Benchmarks library](https://www.madebyagents.com/benchmarks): Plain-language deep dives into the datasets used to score modern AI models across text, image, video, audio, and embedding. ### Text benchmarks - [GPQA](https://www.madebyagents.com/benchmarks/gpqa): Graduate-Level Google-Proof Q&A. Expert-written science questions PhDs barely solve. - [MMLU-PRO](https://www.madebyagents.com/benchmarks/mmlu-pro): Harder replacement for MMLU. 12,032 reasoning questions across 14 subjects with 10 answer choices. - [GSM8K](https://www.madebyagents.com/benchmarks/gsm8k): 8,500 grade-school math word problems testing multi-step arithmetic reasoning. - [SWE-Verified](https://www.madebyagents.com/benchmarks/swe-verified): 500 human-validated GitHub issues used to score AI coding agents. - [HLE](https://www.madebyagents.com/benchmarks/hle): Humanity's Last Exam. 2,500 expert-written questions at the ceiling of human knowledge. - [AIME 2026](https://www.madebyagents.com/benchmarks/aime-2026): 15 elite high-school competition math problems used as a fresh annual reasoning test. - [Terminal Bench](https://www.madebyagents.com/benchmarks/terminal-bench): Live agent test in a real Linux shell — install, debug, configure tasks. - [SWE-Pro](https://www.madebyagents.com/benchmarks/swe-pro): Long-horizon, enterprise-style coding tasks across multiple languages. - [EvasionBench](https://www.madebyagents.com/benchmarks/evasion-bench): 16,726 earnings-call Q&A pairs testing whether a model can spot evasive answers. - [olmOCR](https://www.madebyagents.com/benchmarks/olm-ocr): 1,400 PDFs and 7,000 unit tests for document-to-markdown conversion. - [HMMT 2026](https://www.madebyagents.com/benchmarks/hmmt-2026): Elite university-level competition math problems, fresh in 2026. - [Arena Score](https://www.madebyagents.com/benchmarks/arena-score): Arena.ai (formerly LMSYS Chatbot Arena). Anonymous head-to-head human preference ranking for chat models, Elo-style. - [WebDev Arena](https://www.madebyagents.com/benchmarks/webdev-arena): Arena.ai ranking for models that turn natural-language prompts into working web apps. - [Image-to-WebDev](https://www.madebyagents.com/benchmarks/image-to-webdev): Arena.ai ranking for models that turn screenshots or mockups into working web apps. - [Search Arena](https://www.madebyagents.com/benchmarks/search-arena): Arena.ai ranking for search-grounded, citation-backed answers on live web prompts. - [Vision Arena](https://www.madebyagents.com/benchmarks/vision-arena): Arena.ai ranking for vision-language models on real image-understanding prompts. - [Document Arena](https://www.madebyagents.com/benchmarks/document-arena): Arena.ai ranking for models that read PDFs, slides, and long screenshots to answer questions. ### Image benchmarks - [Image Arena](https://www.madebyagents.com/benchmarks/image-arena): Arena.ai head-to-head ranking for text-to-image generations. - [Image Edit Arena](https://www.madebyagents.com/benchmarks/image-edit-arena): Arena.ai head-to-head ranking for instruction-based image edits. - [GenEval](https://www.madebyagents.com/benchmarks/gen-eval): Compositional prompt-following — counting, positioning, color, attribute binding. - [HPS v2](https://www.madebyagents.com/benchmarks/hps-v2): Human Preference Score v2 — a learned reward model trained on hundreds of thousands of labels. - [ImageReward](https://www.madebyagents.com/benchmarks/image-reward): Reward model combining alignment, fidelity, and aesthetics into one score. ### Video benchmarks - [Video Arena](https://www.madebyagents.com/benchmarks/video-arena): Arena.ai head-to-head ranking for text-to-video generations. - [Image-to-Video Arena](https://www.madebyagents.com/benchmarks/image-to-video-arena): Arena.ai head-to-head ranking for clips animated from a still input image. - [Video Edit Arena](https://www.madebyagents.com/benchmarks/video-edit-arena): Arena.ai head-to-head ranking for instruction-based clip editing. - [VBench](https://www.madebyagents.com/benchmarks/vbench): 16-dimension benchmark covering temporal coherence, motion, and prompt fidelity. ### Audio benchmarks - [TTS Arena](https://www.madebyagents.com/benchmarks/tts-arena): Blind A/B human preference ranking for text-to-speech models. - [WER](https://www.madebyagents.com/benchmarks/wer): Word Error Rate from the Open ASR Leaderboard. Lower is better. - [MOS](https://www.madebyagents.com/benchmarks/mos): Mean Opinion Score on a 1–5 scale for TTS naturalness. ### Embedding benchmarks - [MTEB Overall](https://www.madebyagents.com/benchmarks/mteb-overall): Massive Text Embedding Benchmark — aggregate score across 56 datasets. - [MTEB Retrieval](https://www.madebyagents.com/benchmarks/mteb-retrieval): Retrieval task group, the best predictor of RAG quality. - [MTEB Classification](https://www.madebyagents.com/benchmarks/mteb-classification): Classification task group via linear probes on frozen embeddings. - [MTEB Clustering](https://www.madebyagents.com/benchmarks/mteb-clustering): Clustering task group, scored with V-Measure. - [MTEB STS](https://www.madebyagents.com/benchmarks/mteb-sts): Semantic Textual Similarity, Spearman correlation against human ratings. ## Feeds - [RSS feed](https://www.madebyagents.com/rss.xml): Blog RSS feed. - [AI Stack Status RSS feed](https://www.madebyagents.com/api/status-tracker/rss): Aggregated incident feed for OpenAI, Anthropic, Google AI Studio, xAI, and Cursor. ## License Content on this site is the original work of Made By Agents. Citations and excerpts of up to 25% of any single post are permitted with attribution and a link back to the source URL. Wholesale reuse, training data inclusion at scale, or republication requires written permission via the Contact page.