Browse the definitive directory of AI tools for builders.
The 2026 AI tool landscape consolidated into real categories: agentic coding (Cursor, Claude Code), orchestration frameworks (LangChain/LangGraph), vector databases, evaluation and observability platforms, vibe-coding builders, and enterprise AI suites. Choosing well matters: tool decisions compound into architecture, cost, and lock-in positions that outlive any single project.
This directory covers 69 tools with what each is actually best for, current pricing, honest pros and cons, and the alternatives worth benchmarking. Every tool links to the skills it rewards and the industries deploying it.
An agentic coding tool (Cursor or Claude Code), one frontier model API (Anthropic or OpenAI), a retrieval stack (a vector database plus an orchestration framework), and an evaluation tool (promptfoo or LangSmith). That set covers the workflow most AI engineering jobs assume.
Open-source infrastructure (PyTorch, vLLM-class serving, Postgres/pgvector, MLflow, promptfoo) runs production at the largest companies. Paid tiers buy managed operations, governance, and support: the build-vs-buy calculus depends on team capacity, not capability gaps.
Instrument token and credit spend per feature from day one, route tasks to right-sized models, use prompt caching and batching, and review seat-based licenses against actual usage quarterly. Most AI budget surprises trace to unrouted premium-model traffic and unused seats.