One workspace.
Every API.
Invoked is the AI agent workspace for developers. Switch between API surfaces in a keystroke, compose tools, orchestrate multi-step runs, and ship skills that get smarter over time.
Built on open standards · No lock-in
One workspace.
Every API you work with.
Switch between API surfaces without losing context. Built for developers who work across more than one thing.
Full visibility, at a glance.
Monitor tool call volume, latency, error rates, and agent runs in real time. The dashboard surfaces what matters without requiring you to dig.
Every MCP server, agent, and tool set in one place.
Connect multiple surfaces and flip between them in one click. Each surface keeps its own MCP tools, agents, chat history, analytics, and skills, completely isolated. Your GitHub workspace and your Hugging Face workspace live side by side without interfering.
Every tool call, visible.
See the full execution trace for every agent run: model reasoning, tool inputs and outputs, token usage, and timing. No black boxes.
Save what works.
Turn any successful run into a reusable skill. Parameterize inputs, pin a model, lock a tool set. Run it again with one click or share it with your team.
Measure before you ship.
Compare runs side-by-side. Swap models, tweak prompts, and score outputs automatically. Know if your change is actually an improvement.
Direct the agent, don't babysit it.
Describe what you want in plain language. The orchestrator breaks it down, selects tools, and executes while showing you every decision as it happens.
Orchestrate in parallel.
Place agents on a canvas, connect them, and run them in dependency order. Independent agents execute in parallel automatically.
Build your own tool set.
Browse all 17+ built-in tools, configure parameters, and compose exactly the capability surface your agent needs. No bloat, no guessing.
Conversational when you need it.
Sometimes you just want to talk to the model. Chat panel gives you a clean interface with full tool access and the same run tracing you get everywhere else.
Know your top performers.
Track which tools get called most, how fast they respond, and which skills are winning. Built-in performance rankings with latency percentiles across any time window.
Your prompt library, not a scratch pad.
Save prompts that work. Organize them, give them parameters, and run them directly in Chat or Agent mode with a single click. Templates you'll actually reuse.
Export your tools. Import anywhere.
Every tool you configure is exposed as a live OpenAPI 3.0 spec. Download the JSON and import directly into Postman, Swagger UI, or any HTTP client.
Every provider. Your keys, your rules.
Connect Anthropic, OpenAI, Groq, Ollama, and more, or point to a custom endpoint. All credentials stay on your machine. Switch models without touching config files.
The details that matter to developers.
Invoked is dense with the kind of tooling that doesn't make a headline but makes the difference at 11pm when something breaks in production.
Every failed tool call surfaces the full error: stack trace, HTTP status, and raw response body. No guessing what went wrong.
Open a specific workspace on launch, enable verbose logging, or attach DevTools. All via command-line flags, scriptable from CI.
Configure retry attempts per tool, set timeout thresholds, and define fallback models when primary providers are unavailable.
Every run, token count, and tool output is stored locally in SQLite. Query it directly, export it, or leave it alone. Your data stays on your machine.
Your configured tool set is always available as a live OpenAPI 3.0 spec. Import into Postman or Swagger UI in one click.
Stream live logs to your terminal or Console.app. Full request/response payloads, timing breakdowns, and agent decision traces.
Swap the underlying model without restarting a workflow. Pin specific model versions to skills for reproducible results.
⌘K command palette, panel shortcuts, and skill launchers. Everything reachable without lifting your hands from the keyboard.
Set per-run token budgets and get warnings before you overshoot. Cost estimates shown inline before you confirm a run.
How much does “agent-ready” cost without Invoked?
The first build is expensive. But the real cost is everything after: every API change triggers another rebuild cycle.
These estimates already assume your engineers are using AI copilots. Architecture decisions, prompt engineering, and integration testing still take human judgment. AI accelerates the coding, it doesn't eliminate the weeks.
What making one API agent-ready actually takes
| Map API surface & design tool schemas(even with AI assistance) | 2–4 hrs |
| Build custom developer environment(AI speeds coding, not architecture decisions) | 3–5 days |
| Wire AI agent orchestration(prompt engineering + integration testing) | 2–3 days |
| Validate and iterate on tool behavior(real API calls, edge cases) | 4–8 hrs |
| Total per API surface | ~8–13 eng days |
Your data never leaves your machine. Conversations, run history, and API keys are stored locally in SQLite. Only anonymous usage metrics are sent to analytics.
What the first build gives you, in 60 seconds
- A native desktop app your whole team can install and use immediately
- Browse, search, and invoke every endpoint interactively
- An AI agent that calls the right tools automatically
- Full run history and trace visibility across your team
- API changes? Edit a config file, no rebuild, no redistribution
vs. building it yourself with an AI coding agent
Invoked Team ($25/seat/mo)
At $100/hr, payback = 15 minutes of one engineer's time. Every month after is capacity back on the roadmap.
Get Team — $25/seat/mo →Conservative estimate · assumes engineers are already using AI copilots · 60% utilisation (integration work is rarely one person's full focus) · iteration savings: 4 hrs per update cycle vs. full rebuild · blended rate default $100/hr
Download it. Use it. Love it.
Install the full app. Not a sandbox, not a demo. Your trial gives you 10 days and 50 runs against the pre-configured Hugging Face workspace — no credit card, no API key required to start. Subscribe when you're ready to go unlimited.
- ✓Full app, not a demo or web sandbox
- ✓Hugging Face workspace pre-configured
- ✓Bring your own API key (Anthropic, OpenAI, Groq…)
- ✓Run history, evals, flow canvas
- ✓All runs saved locally to SQLite
- –50 run limit on the HF workspace
- –Unlimited runs & all providers after trial
- ✓Everything in the trial, no run limits
- ✓Unlimited runs across all workspaces
- ✓Connect Anthropic, OpenAI, Groq, Ollama + more
- ✓Build & author your own skills
- ✓Multi-agent canvas with parallel execution
- ✓Full eval harness with LLM judge scoring
- ✓Solo and Team plans available
Simple pricing.
Try the full app free. Pay when you're ready to go unlimited.
10 days · 50 runs · No credit card.
- Full native app, Mac and Windows
- HF workspace pre-configured
- No API key required to start
- Run history and full trace
- Skills and prompt workspace
- 10 days · 50 run cap
Or $149/year — save $31. Two months free.
- Everything in the trial
- Unlimited runs
- All 5 LLM providers
- Full evals and model comparison
- Multi-agent flow canvas
- Skill authoring and export
- API and OpenAPI integration panel
- Priority support
Or $249/seat/year — save $51/seat. Two months free.
- Everything in Solo
- Shared workspaces
- Team skill library
- Config and credential sharing
- Role-based access control
- Org-wide analytics
- Dedicated support
White-label Invoked under your brand and distribute to your customers or internal teams.
- Everything in Team
- Your logo, colors, and app name
- Custom domain deployment
- Pre-configured surfaces and tools
- SSO / SAML passthrough
- Org-wide admin analytics
- Managed update channels
- Dedicated support & SLA
One agent execution or chat turn with tool calls. Browsing the UI, editing skills, and reviewing run history don't count.
Immediately on signup. The free trial gives you 10 days and 50 runs — no credit card, no API key required to start. Free API key options are available from Anthropic, OpenAI, Groq, and HuggingFace if you want to go deeper. Upgrade to Solo or Team when you're ready for unlimited runs and full feature access.
Runs are stored locally in SQLite on your machine. Anonymous usage metrics go to Tinybird for product analytics. No run content ever leaves your device.
Ready to go beyond the demo?
Connect your own API keys, build your own skills, and run agents on your own data. Fully local, no account required.
macOS 13+ · Windows 10/11 · No account required
The skill marketplace
is almost here.
Browse skills built by the community. Install in one click, fork to make them your own, or publish what you've built and let others put it to work. The best agent workflows shouldn't live on one machine.
Also on the roadmap
Shared workspaces, team skill libraries, and role-based access. Publish your tool config once and every teammate pulls the same setup instantly.
Token spend, run frequency, skill adoption, and model cost breakdowns across every member of your team in a single view.
Tell Invoked what you want to automate. It scaffolds the skill, wires the tools, and writes the prompt. You review and publish.
A first-party plugin API for tools, providers, and exporters. Build integrations for any service and publish them to the marketplace.