Glossary
Every Wingman word in one or two sentences, with the page that teaches it. For the connected story, read Core concepts in Get started.
How this list is ordered
Terms are alphabetical-ish: product first, then the things you make, then hardware and plumbing.
| Term | Definition | Learn more |
|---|---|---|
| Axelera Wingman | The product: one account across a desktop app, hosted chat, mobile web, and hosted MCP for AI CLIs. | Get started - What is Axelera Wingman |
| Agent Dev | The desktop mode where an agent builds and runs on your machine — your repos, your terminals, and the Axelera card in front of you. | Agent Dev - Agent Dev overview |
| Chat (hosted chat) | The hosted conversation surface, in the browser and embedded in the desktop app. It plans, reviews, and searches knowledge; it cannot touch your hardware. | Chat - Wingman Chat overview |
| Wingman Home | The home screen, and the rounded button at the top center of the desktop window that returns to it from any mode. Home is where you switch between Chat and Agent Dev. | Get started - Wingman Home: your command center |
| Run mode | The mode Wingman starts in — Chat, Agent Dev, or Zendesk — picked on Wingman Home. | Get started - Core concepts |
| Topic | One saved conversation in chat. Start a fresh one with Alt + N; organize, share, and export them from the topic sidebar. | Chat - Organize, share & export topics |
| Agent | A configured assistant: instructions, documents, opening questions, and model defaults, reusable across conversations. | Chat - Create & customize agents |
| Group | A team-style conversation where several agents work together, each with its own role and settings. | Chat - Group chats |
| Knowledge base | Indexed sources — files, documents, repositories — that agents can search and cite. Knowledge is what Wingman looks things up in. | Agent Dev - Build your Knowledge Base |
| Memory | Preferences and facts Wingman learns about you and reuses across conversations. Memory is learned; knowledge is indexed. | Chat - Memory |
| Workspace | The folder the desktop app has open. Agent Dev reads and edits inside it. | Agent Dev - Your first Agent Dev session |
| Project | A pipeline-level workspace in the dashboard that collects the documents an agent writes for one effort, browsable as folders and files. | Agent Dev - Projects |
| Work order | Retired. A repository-backed job that ran in a sandboxed branch or worktree. Removed because it reported phases for runs that never touched a repository. | Agent Dev - Projects |
| Approval policy | When the Agent Dev agent must ask before acting: never, on-request, on-failure, or untrusted. | Agent Dev - Approval policies & sandbox modes |
| Sandbox mode | What the agent may touch while it works: read-only, workspace-write, or full access. | Agent Dev - Approval policies & sandbox modes |
| Readiness indicator | The Agent Dev provider status below the composer. Green is ready, amber opens diagnostics, and an expanded Fix now button starts guided repair. | Agent Dev - Your first Agent Dev session |
| Metis | Axelera’s first AIPU generation, inside the Embedded 110m, 113m and 111c, the Edge 130p, and the Server 150p. | Devices - Hardware generations & product names |
| Europa | Axelera’s second AIPU generation, inside the Edge 232p and the Server 250p PCIe cards. | Devices - Hardware generations & product names |
| Voyager SDK | Axelera’s SDK for building, deploying, and running pipelines on Axelera hardware. Wingman Home verifies your install automatically. | Devices - Set up the Voyager SDK & your card |
| MCP | Model Context Protocol: the open standard Wingman uses to expose its tools to AI clients. | MCP & Skills - Connect AI tools: overview |
| Hosted MCP | Wingman’s cloud MCP endpoint. It plugs Wingman tools into Claude Code, Codex, and Gemini CLI with a browser sign-in. | MCP & Skills - Connect any MCP client & manage servers |
| Skill | A packaged, runnable workflow an agent can invoke by name to do one job well. | MCP & Skills - Run Wingman skills |
| Zendesk mode | The support-triage run mode that connects Zendesk knowledge through MCP. | MCP & Skills - Zendesk support triage mode |
| Axelera Earth 1.0 | Chat’s default model, the deepest Axelera AI model for multi-step work. Pick another model from the model menu and Chat keeps it. | Reference - Models reference |
| Evaluation Lab | The workbench for benchmarking your agents against test sets before you rely on them. | Chat - Evaluation Lab: benchmark your agents |
| Credit | The only balance your account keeps. Axelera Venus 1.0, Axelera Earth 1.0 and agent work spend credits; Axelera Mercury 1.0 chat spends none. Nothing is priced to you in currency. | Chat - Chat settings |
| Plan | What your account is on, and how many credits it receives each month. Launchpad is the plan every account starts on; the paid plans are arranged with Axelera. | Chat - Chat settings |
| Grant | One bucket of credits: your plan’s monthly credits, the one-off welcome gift, or a top-up. Each has its own expiry date, except the welcome gift, which has none and stays until you spend it. The Credits page lists them in the order they are spent, soonest to expire first. | Chat - Chat settings |
| Top-up | A block of extra credits added to your account. It is spent only after your plan credits, and stays valid for months. | Chat - Chat settings |
| Fair-use ceiling | The daily bound on free Axelera Mercury 1.0 chat. It resets at 00:00 UTC, is the same on every plan, and is set high enough that ordinary use never meets it. | Chat - Chat settings |
| Seat cap | A monthly credit limit your account administrator can set on one seat. The account still has credits; that seat may not draw more of them this month. | Chat - Chat settings |