Fetch.ai

Fetch.ai is an open-access blockchain-and-AI stack that lets anyone create autonomous software agents capable of making decisions, exchanging value, and collaborating without human micromanagement. Founded by Humayun Sheikh, Toby Simpson, and Thomas Hain in 2019, the project pairs an agent runtime (the uAgents framework) with a Cosmos-SDK–based ledger, on-chain identity, micropayments in FET, and a growing suite of developer tools—laying the groundwork for a truly decentralized “Internet of Agents.”

General
Release date	March 2019 (Binance Launchpad IEO)
Founders	Humayun Sheikh, Toby Simpson, Thomas Hain
Type	Decentralized AI-Agent Platform

Core Features

uAgents Framework – Build lightweight Autonomous Economic Agents with the Python SDK, capable of on-chain and off-chain interactions.
Agentverse – A cloud IDE & marketplace for hosting and monetizing agents; explore the docs at docs.agentverse.ai.
DeltaV Chat Interface – A conversational gateway where users can talk to agents via natural language; see the intro blog “Welcome to DeltaV.”
Fetch Ledger – Secure consensus, smart contracts, and FET-denominated fees on the main network.
Almanac Contract – An on-chain registry that lets agents publish metadata and be discovered; details in the Almanac docs.

Tools & Resources

Source & SDKs on GitHub
Quick-start guides in the Agentverse Docs
Live chat with agents via DeltaV
Staking walkthrough: Different Ways to Stake FET
Bridging guide: Native vs ERC-20 FET

Ecosystem & Integrations

Mobility & Smart Cities – Real-time traffic and parking optimization powered by Fetch agents: “How AI Agents Are Revolutionizing Mobility and Smart Cities.”
DeFi & Trading – Agent-based execution on DEXs with the toolkit announced in “Agent-Based Trading Tools for Decentralized Exchanges.”
Supply Chain Automation – Autonomous logistics coordination featured in Moralis Academy’s “Deep Dive on Fetch.ai.”
Energy Management – Peer-to-peer energy-grid balancing showcased in “Fetch.ai Powers Sustainability with AI Agents.”
Artificial Superintelligence Alliance (ASI) – Token merger with SingularityNET & Ocean Protocol; see “Fetch.ai, SingularityNET and Ocean Protocol Unite.”

Fetch.ai continues to expand its toolkit and partnerships, positioning autonomous agents as a practical layer for decentralized AI across finance, mobility, energy, supply chain, and beyond. To start building, dive into the Agentverse docs or clone examples from the Fetch.ai GitHub—then deploy and monetize agents directly on the Fetch Network.

Voni | complex workflows with a voice agent

For example, a Voni agent can: * Book an appointment by checking availability and creating the booking. * Call a new lead within seconds, collect their information, update the CRM, and schedule a meeting. * Help a visa applicant by collecting their information, checking their application, identifying missing documents, updating the application, sending documents, and completing the required steps. * Process a purchase by finding the right product, creating the order, processing the payment, and sending a confirmation. * Handle a support request by finding the customer's account, updating information, solving the issue, and escalating the case when needed. * Handle a fraud report by collecting the details, checking relevant information, creating a case, and notifying the right team. And these are only examples. If a business process can be connected to a tool, API, database, or system, a Voni agent can be built to perform that process. With Voni, a phone call can become a complete business workflow. You tell the agent what you need. It takes care of the work.

BugForge

BugForge — AI-Driven Chaos Engineering Platform The Problem Modern systems fail in compound, unpredictable ways — not just one thing at a time. Yet engineers practicing chaos testing almost always test simple, single-fault scenarios: kill one container, inject one latency. They rarely combine multiple simultaneous failures. The result is a false sense of resilience — systems that pass every individual chaos test but collapse when two or three failure modes hit at once. Traditional chaos tools also lack structure: no scenario review before execution, no automatic fault revert, and no standardized reporting. The Solution BugForge uses an LLM (OpenAI GPT-4o or IBM watsonx Granite) to generate compound, multi-layered fault scenarios that human engineers rarely think to combine — like "inject 300ms latency on the API while the database is frozen while the worker is under 90% CPU stress." These are the scenarios that expose real weaknesses: missing retry logic, absent circuit breakers, connection pool exhaustion. A human-in-the-loop approval step ensures no fault is ever injected without explicit sign-off. Once approved, BugForge executes faults directly against live Docker containers using 6 fault types — container kill, process freeze, network partition, CPU stress, memory stress, and latency injection — all automatically reverted after each step. Real-time container logs and CPU/memory metrics are streamed during execution. A structured report then shows per-step pass/fail results, peak resource charts, and an optional LLM-written resilience score from 1 to 10. Target Users Site reliability engineers, platform engineers, and backend developers operating microservice systems on Docker or Docker Compose environments. What Makes It Unique AI-generated compound scenarios — the LLM designs failure combinations humans would not think to test Human-in-the-loop approval — no fault touches a container without engineer sign-off End-to-end in one platform — design, approve, execute, observe, and report in a single tool The entire workflow — from describing a system to reading the final resilience report — takes minutes, not days.