The Generator Chat Just Got a Lot Smarter to Navigate
ChemIllusion's generator chat now shows a context-aware skill chip ribbon that surfaces the right actions for what is on your canvas—no more hunting through menus or remembering which skill does what.
One of the most common things we heard from new users goes like this: they knew ChemIllusion could do something—highlight atoms, predict a mechanism, change a bond color—but they could not remember how to ask for it. The skill system has always been powerful, but finding the right skill required you to already know its name or open the full library and scroll.
That friction is gone. The generator chat panel now shows a ribbon of skill chips directly above the input field. The chips update in real time based on what is on your canvas and what you have selected. Click one, and either the action runs immediately or the chat input fills with a ready-to-edit template. Either way, you never start from a blank box again.
The generator chat now surfaces relevant actions for whatever is on the canvas — no typing required to get started.
The problem with a blank input box
Most chat interfaces for tools give you a text box and nothing else. That works fine once you know what the tool can do. It is terrible when you are new, when you are in the middle of a workflow and have forgotten a skill name, or when you are a student who has never used the tool before.
ChemIllusion has dozens of skills across molecule creation, styling, mechanism work, protein visualization, educational workflows, and more. No one holds all of those in their head. The blank input box was forcing people to either memorize skills or stop and browse the library every time they wanted to do something non-obvious.
The chip ribbon is our answer to that. Instead of asking you to remember, it looks at the canvas and tells you what is most useful right now.
How the chip ribbon works
The ribbon shows up to four chips at once, ranked by the current canvas state. The ranking logic considers what types of objects are present, whether anything is selected, whether you have a reaction on the canvas, and what the most relevant next step tends to be for each combination.
Chips come in two flavors, marked visually with different icons:
⚡ Action chips — these run immediately when you click them. The prompt is pre-set. One click and the skill executes against your current canvas context. Use these when you know what you want and the chip matches it exactly.
✎ Template chips — these fill the chat input with a starting prompt you can edit before sending. Use these when you want to customize the request—change which atom to highlight, specify a different color, adjust the scope.
A first-time hint banner explains the distinction the first time you see the ribbon. After you dismiss it, the chips stay clean.
If you want to see more options than the four in the ribbon, a ⋯ button at the end opens the full skill library, already filtered to what is relevant for your current canvas.
What chips appear when
The chip set changes meaningfully across different canvas states:
Empty canvas — starter chips for the most common first moves: create a molecule from a name, draw from SMILES, or add a protein structure.
One molecule selected — chips for single-molecule actions: highlight specific atoms, predict a reaction mechanism, change bond colors, generate a structure animation.
Reaction on the canvas — mechanism-oriented chips surface when a reaction arrow and two or more molecules are detected together.
Protein selected — chips for protein-specific workflows: search BindingDB for ligand targets, change the rendering representation, or export the view.
Image uploaded — when you drop a photo or PDF onto the canvas, chips appear for identifying molecules from the image and extracting SMILES.
Molecule-context chips like Highlight bond appear automatically when a structure is selected.
On an empty canvas, starter chips like Create molecule give new users an immediate first move.
The scope strip: knowing what the AI sees
Above the chip ribbon, a narrow scope strip shows exactly what context the AI is working with before you send anything. It shows one of three states:
Selected items — the AI will act on what is highlighted on the canvas
Whole canvas · N items — the AI has visibility into everything on the canvas
No canvas context — the AI is answering as a general chemistry assistant with no canvas state attached
A small change dropdown lets you switch scope without clicking away from the chat. You can also use it to bulk-select all molecules, all proteins, or all text elements at once—useful when you want to act on a specific object type across a complex canvas.
This addresses a real source of confusion. When users sent a message like "make it red," they could not always tell whether the AI was about to act on the selected molecule or redraw everything. The scope strip makes that explicit before the message is sent, not after the wrong thing happens.
The skill library now recommends
Skills route directly to the right chemistry tool — the chip ribbon makes the first step visible before you type anything.
The full skill library modal has always had the complete catalog. Now it opens with a Recommended section at the top, ranked by the current canvas context using the same logic as the chip ribbon. If you need something that is not in the ribbon—a less common skill, or one that applies to a combination of objects the ribbon did not anticipate—the library is still one click away and now leads with the most likely options rather than alphabetical order.
Why this matters for students
Instructors who share ChemIllusion with students consistently tell us the same thing: students do not know where to start. They draw a structure, open the chat, and stare at the empty input box. Some type something vague. Most close the panel.
Feature discovery for students happens in the moment — the chip ribbon shows what to do next without requiring prior knowledge of the skill system.
The chip ribbon changes that. A student who draws a ring system immediately sees that they can highlight the pi bonds, predict an electrophilic aromatic substitution mechanism, or ask the AI to explain what is on the canvas. That first move is no longer blocked by not knowing what to ask.
For instructors building lesson sequences, the chip ribbon also reinforces what skills are relevant at each stage of a drawing exercise. If a lesson involves mechanism drawing, the reaction-context chips surface the mechanism tools automatically when the student places an arrow. No instruction needed.
What we did not change
The chip ribbon does not replace typing. If you know exactly what you want to say, the input still works the same way. The chips sit above the input without covering it, and they disappear while the chat is sending so they do not clutter the interface during a response.
The underlying skill routing is also unchanged. Whether you trigger a skill by typing its name, clicking a chip, or selecting it from the library, the same router and executor run the same way. The chips are entry points, not a different code path.
Try it now
Open the generator, draw any molecule, and select it. The chips that appear are the ones most likely to be useful for that structure right now. Try the action chips first—they are the fastest path from molecule to result.
The chip ribbon is live in the generator now.
Open ChemIllusion Generator →