An experimental no-code tool from Google Labs. Describe what you want to build in plain language, and Opal assembles a multi-step workflow that chains prompts, AI model calls, and tools — then runs it as a shareable “mini app.” The assembled logic appears as a chain of step blocks in a visual editor, where you can drag to rewire connections or rewrite individual prompts by hand. Finished apps can be shared by link, and existing apps in the gallery can be remixed into your own version. Opal launched in July 2025 as a US-only experiment, expanded in stages, and is now available in more than 160 countries and regions.
Key Features
- Workflow generation from natural language: Write what you want to do, and Opal builds the flow from accepting input, through AI processing, to output. After it exists, saying “change this part like so” can apply several edits across multiple steps at once
- Manual editing in the visual editor: The generated workflow is drawn as a chain of steps. You can add, delete, and reorder steps or redraw connections, adjusting the logic while seeing the whole picture
- Three basic step types: Combine “User Input” for collecting data from the user, “Generate” for passing a prompt to a chosen AI model, and “Output” for controlling how results are presented (web page, spreadsheet, and so on). You can also attach uploaded images or linked YouTube videos as reference material
- Agent step: Added in 2026 for more autonomous handling. Rather than following a fixed script, it decides which tools and models to use as the situation develops, and can ask the user clarifying questions along the way
- Sharing, publishing, and gallery remixing: Two options — “share” grants specific people view or edit access, while “publish” creates a link anyone can use. Demo apps in the gallery can be copied and freely modified without affecting the original
- Version history and console: Edits save automatically and earlier states can be restored. During a run, the console shows each step’s inputs, outputs, and the model’s reasoning, which makes it easier to narrow down why something isn’t behaving as intended
Pricing
| Plan | Price | Highlights |
|---|---|---|
| Free (experimental Google Labs release) | $0 | All features available with a Google account. No paid plan is offered |
Pricing is current as of August 2026. As an experimental Google Labs product, the terms of availability may change. Check the official site for the latest information.
Pros & Cons
✅ Pros
- Build multi-step AI processes without writing a single line of code
- Moving freely between natural-language instructions and visual editing suits a “rough draft first, refine later” workflow
- Gallery apps can be remixed, so you can start from something that already works instead of a blank canvas
- Finished apps can be distributed by link, making it easy to hand tools to a team
- Free at present, and usable right away with a Google account
⚠️ Cons
- As an experimental Google Labs product, you should expect the possibility of spec changes or discontinuation
- Not suited to full production systems with complex branching or large-volume data processing
- Requires a Google account, and in some cases recipients need one too
- The quality of the generated workflow depends on your prompt; it may not be assembled as intended
- Support is not on par with a fully launched product, so don’t count on help when something breaks
Comparison with Similar Services
| Criteria | Opal | Dify | n8n | Zapier |
|---|---|---|---|---|
| Provider | Google Labs | LangGenius | n8n | Zapier |
| Main use | Building and sharing AI mini apps | Building LLM apps and agents | Business automation workflows | Automating connections between SaaS tools |
| How you build | Natural language plus visual editing | Mostly visual editing | Node-based editor | Trigger-and-action configuration |
| Target users | Mainly non-engineers | Developers and product staff | Technically inclined practitioners | General business users |
| External integrations | Limited | Extensible via plugins | Extensive | Very extensive |
| Pricing | Free (experimental) | Free tier / paid plans | Self-hostable / paid plans | Free tier / paid plans |
Who Is It For
- People with work they’d like to automate with AI but who would rather not write code
- Anyone repeatedly copy-pasting a set of prompts — the whole sequence can become an app
- Teams that want to prototype a small internal tool and gauge the reaction first
- Anyone who wants to learn how a chain of model calls fits together by seeing it as a diagram
- AI beginners who want to grasp what an “AI workflow” is by playing with working examples
Summary
Opal packages multi-step AI processing into something app-shaped that you can hand to someone else. Being able to move freely between natural-language instructions and visual editing lowers the barrier to getting started considerably. That said, because it is an experimental Google Labs product, it makes more sense for prototyping, internal sharing, and validating ideas than for anything load-bearing in your business. It is free to try, so remixing a single app from the gallery is the quickest way to see what it can do.