At Invent, we unlock real-world capabilities by equipping every model with powerful built-in tools.

Think of it as handing your AI a modular toolbox, so it can do more than just chat.

But here’s the inside scoop: These tools aren’t widely documented and operate independently behind the scenes.

Let’s take a peek under the hood!

What are Built-In Tools?

Instead of limiting models to just text, every AI on Invent gets a set of default tools, like:

• Web Search: Fetch up-to-date info from the internet.
• Scrape/Download: Grab content or files from URLs (websites, PDFs, images, etc.).
• Think: Trigger deliberate, step-by-step reasoning.
• Generate Image: Create images from text prompts

These tools work the same, no matter what model you pick.

Tools have their own logic

Each tool is a mini-engine with its own logic. When a tool gets called:

• It can use its own specialized AI models.
• It might download files, scrape websites, or even combine multiple technologies under the hood.

Example:
The “Generate Image” tool doesn’t depend on the conversation model. Internally, it may use Nano Banana (Gemini Flash) to create the image, even if your chat is powered by a different model. The backend for each tool can be swapped or upgraded anytime, without changing the chat experience for users.

How “Auto” selects tools…and even switches models

The real magic happens in the Auto layer, which reviews every user message and:

Decides if, when, and which built-in tool to call (all automatically).
If the primary model (e.g., GPT-4.1) can’t handle your upload (say, a tricky file type), "Auto" seamlessly falls back to a more capable model (like Gemini Flash 2) that supports wider formats.

You never have to worry about compatibility, "Auto" always picks the combination that just works.

Our modular, upgradable tools mean you’re always using the best we have to offer.

Keep following the 'Under the Hood' series for the latest innovations and deep dives.