AI doesn't have to be faceless. The Bots AI Agent Avatar Illustrations Collection brings every chapter of the Bots series together in one place: a growing cast of bot avatars, each an everyday object with a tiny face. A coffee cup, a shopping cart, a strawberry, a padlock, a tent. Give each assistant, chatbot and automation its own face, and people stop wondering which bot just replied.

Every avatar is an editable vector in SVG, PNG and AI, drawn in black outline with neutral greys, and every one has eyes you can animate. New chapters join the collection as they're released, so the cast keeps growing.

One face, drawn across every chapter

The collection works because the drawing never changes. Each piece uses the same thick black contour at one steady weight, with a slight hand wobble and rounded ends. Inside the line there are three flat greys, white speckles and a highlight streak on anything shiny. About half the objects stand on a solid black oval shadow, which keeps them grounded when they're cropped into a circle.

The faces are what tie the series together: two close-set oval eyes with a white glint and, almost always, no mouth. On the darkest objects the eyes flip to white circles with dark pupils so they still read at 32 px. That restraint keeps the whole cast calm and a little deadpan, so a server stack, a cupcake and a dinosaur all look like they came from the same studio. Also if you are missing an illustration you can always request it from the Custom Work form and we will make sure to include it or if you're in a hurry we will custom make it for you.

Why the set works better than single packs

Bots multiply. A team starts with one support assistant, then adds a research agent, a scheduler, a coding helper and a store assistant. If those avatars come from different sources, the product starts to look stitched together. With the whole series, every new agent gets a face that already matches the rest.

The subjects also cover a wide spread of roles, from work and tech objects to food, shopping, money, nature, weather, travel and play. That range lets you match an avatar to what the bot actually does: a magnifier for search, a piggy bank for budgets, a map pin for travel, a speech bubble for support.

A simple habit helps here: name the bot, then pick the object. A bot called Scout gets the magnifier, Penny gets the piggy bank, Chef gets the cupcake. Once a name and a face line up, people remember which agent does what, and your team stops writing "the second assistant" in tickets. With the full collection you rarely have to settle for a face that only half fits.

And because each chapter ships in the same palette and on the same square artboards, there's no clean-up when you mix them. Drop them into one Figma library and they behave as one set.

Animate the faces

Every avatar in the collection has eyes built from simple vector shapes, so a static bot profile picture can react. Squash the eyes into a line and it blinks. Slide them sideways and it glances at the input box. Widen them when something breaks, or flatten them into arcs for a happy squint.

For an AI agent, those expressions double as status: idle, thinking, error, done. People read a face faster than a spinner. Figma Smart Animate, After Effects, Rive and Lottie can all start from the SVG, and on the web an inline SVG with a few lines of CSS is enough for a looping blink. You have to take a look at our illustration packs, seriously you will be impressed.

Formats and technical details

• SVG files, editable and sharp at any size

• PNG files for tools that won't accept SVG

• AI source files for Adobe Illustrator

• Square artboards, one avatar per artboard

• One neutral palette across the series: black line, three greys and white

Licence

The GetIllustrations commercial licence covers personal and commercial work, client projects, apps, websites, marketing, video and print, and you can recolour, resize and edit the art. You can't resell or redistribute the files on their own, bundle them into templates or themes for resale, sub-license them, or add them to AI training datasets. Attribution isn't required.