On‑chain AI agent platforms bring together agent engineering, blockchain verification, and commercial infrastructure so autonomous agents can act, prove their behavior, and generate revenue. This topic covers agent marketplaces that list and transact with agents; frameworks for building stateful, verifiable agents; decentralized compute and model‑serving stacks; AI data platforms for training and auditing; and governance tools that define rights, liabilities and upgrade paths. Relevance in 2026 stems from maturing toolchains and composable architectures: engineering platforms such as LangChain (including stateful graphs like LangGraph) are used to design, debug and deploy reliable agent workflows; no‑code and low‑code entry points like AgentGPT let non‑developers instantiate goal‑directed agents quickly; developer‑centric environments (Windsurf, Replit, Tabby) accelerate agent development, iteration and hosting with integrated model serving. Specialist building blocks — smart, AI‑aware crypto wallets (e.g., Aptos‑focused offerings) — enable agents to hold assets, transact and participate in tokenized revenue models (fees, subscriptions, bounties). Key trends: hybrid on‑chain/off‑chain architectures that cryptographically attest agent decisions; marketplaces that standardize interfaces and payment rails; self‑hosted and privacy‑preserving model serving for sensitive workloads; and growing emphasis on governance and auditability to manage risk and legal exposure. For practitioners, evaluating platforms means balancing verifiability, latency and cost, plus available tooling for testing, monitoring and model updates. Together these categories form an ecosystem enabling verifiable, revenue‑generating agents that can be discovered, contracted with, and governed within decentralized environments.