Skip to main content

Nexus platform overview

Welcome to Nexus — Yellow.ai's next-generation platform, accessed at nexus.yellow.ai. Nexus brings together smarter agents, a built-in AI copilot, customised widgets, and an AI trust centre — a brand-new experience, distinct from the legacy platform at cloud.yellow.ai.

nexus.yellow.ai studio with the v3 Agents page open — three peer specialised agents (Test Ride, Bike Finder, Order Tracking) listed under the Agents tab, the New agent button in the toolbar, and Status / Category / Updated By / Last Updated filters above the list

What this section covers​

This section documents two parts of Nexus: v3 agents and the Widget Builder. Copilot Nexus and the AI Trust Centre are documented in their own sections.

Am I on Nexus?​

You're on Nexus when you access the platform at nexus.yellow.ai. To move your workspace's agents to v3, open the AI Agents page and click Switch to v3. If you're on a different URL or haven't switched yet, talk to your account team.

What "Switch to v3" does: it moves your workspace's AI agents from v2 to v3 — same orchestrator + super-agent + specialised-agents + tools model you had on v2, but running on the v3 engine. The v3 engine adds a few platform-enforced behaviours (two-stage routing, automatic step execution for numbered procedures, conversation-history compaction, structured execution traces, parent-child agent triggers with auto-return) plus new authoring surfaces (Routing Logic, Widget Builder, AI Trust Centre, versioned configuration). It's a workspace-level change, not per-agent. See the v3 vs v2 comparison for the full list of changes and migration trade-offs.

What's new in Nexus​

Compared to cloud.yellow.ai, Nexus gives you:

  • Copilot Nexus — an in-platform AI copilot that helps you build, debug, and ship faster, right where you work.
  • The v3 agents engine — keeps the same super-agent + specialised-agents shape v2 had, with platform-enforced refinements: two-stage routing (Context Expert + Conversation Agent), automatic step execution for numbered procedures, parent-child agent triggers with auto-return, history compaction, and structured per-turn traces. Plus new authoring surfaces — Routing Logic and versioned configuration.
  • Widget Builder — design customised interactive UI (forms, pickers, multi-step wizards) and drop them into any conversation. Widgets read live agent state and feed user choices back.
  • AI Trust Centre — visibility and controls over how your agents behave, so you can keep them safe in production.

…and more, shipping continuously.

How a v3 super agent decides what to do​

v3 super agent decides between agents using each agent's Trigger and your optional Routing Logic rules

When a user sends a message, the v3 super agent picks one of three things to do: answer the message itself, hand off to one of your agents (an order-status agent, a billing agent, a demo-booking agent, etc.), or call a tool. The primary input is:

  • Each agent's Trigger — the natural-language description at the top of every agent's profile that says when the agent should activate ("When the user wants to book or schedule a test ride…"). The super-agent router reads every Trigger and picks the best match.

For simple bots with clearly distinct Triggers, that's all you need — agent routing works out of the box. You can optionally add Routing Logic rules on the super agent (plain-English overrides like "If the user mentions billing or refunds, hand off to @billing-agent") when you want a deterministic override on top of trigger-based routing.

Tools work the same way: the super agent can call any tool directly, using each tool's description as a routing signal. Attach a tool to a specific agent when it belongs to that agent's job — that also helps routing.

How to use this guide​

Read the pages in this order if you're new:

  1. v3 Agents → Overview — understand the building blocks: super agent, specialised agents, tools.
  2. Configure your Super Agent — set persona, identity, rules, welcome and fallback.
  3. Set up Routing Logic — control how the super agent picks between agents.
  4. Add agents and tools — split work across specialised agents.
  5. Test your agent — interactive and bulk testing.
  6. Widget Builder — when to use widgets, how to build one, how to wire it in.

Already running a v2 agent? The Best Practices page is the fastest way to see what's different in v3.

A note on freshness​

Nexus is shipping rapidly. We update these pages alongside the platform — if something in the studio doesn't match what you read here, the studio is the source of truth. Tell us, and we'll fix the docs.