Make a symbol for any word: AI symbol generation in Sayella

Every AAC user has a handful of words that no symbol library has ever drawn. The name of the family dog. A grandparent only this child calls "Nanny Pat." A local food with a name that means nothing outside one town. The specific brand of crisps that ends every shopping trip in negotiation. These are not edge cases. They are often the most important words a child wants to say, because they are the words attached to the people and things they love most.

And yet these are exactly the words that traditional symbol sets cannot include. A library, however large, can only draw the words its creators thought of in advance. The personal, local and specific always fall through the gaps.

This article is about how Sayella closes those gaps. AAC, which stands for Augmentative and Alternative Communication, only works when a person can say what actually matters to them. Sayella can generate a clear symbol for any word you type, or turn a photo you take into a symbol, so the words that matter most are never the ones left out.

Why symbol libraries always leave gaps

Symbol libraries are genuinely valuable. A good one gives you thousands of clear, consistent images for the everyday words a communicator needs. Sayella draws on established open libraries for exactly that reason, and most of the vocabulary a child uses every day is well served by them.

But a library is a fixed collection. It was drawn once, by people who could not possibly anticipate every name, place and object in every user's life. So there will always be words it does not cover:

• Names. Pets, people, characters, favourite toys with names only your family uses.
• Local and cultural words. Foods, places, traditions and treats specific to a region or a household.
• Hyper-specific objects. A particular comfort item, a piece of equipment, a one-of-a-kind something that matters enormously to one child.

Historically, filling these gaps meant hunting for a clip-art image, cropping a photo, or asking a therapist to draw something. It was slow, fiddly, and often produced a symbol that did not match the visual style of the rest of the board. So the easiest thing was simply to leave the word out, which quietly told the child that some of the things they care about most are not worth saying.

Generate a symbol for any word

Sayella takes a different approach. When you need a word that no library covers, you can type the word and have Sayella generate a symbol for it. A clear, simple image that fits the look of the rest of the board appears, ready to drop straight onto a button.

That means the family dog "Biscuit" gets a symbol. The local bakery treat gets a symbol. The made-up name for a favourite teddy gets a symbol. Anything you can name, you can give a button, without leaving the app to go hunting for a picture.

The point is not novelty. The point is that a child's vocabulary should be shaped by their actual life, not limited to the words a library happened to include. AI symbol generation removes the excuse for leaving a meaningful word off the board.

Turn a photo into a symbol

Sometimes the clearest possible symbol for a word is the real thing. For very personal vocabulary, especially names and specific objects, a photo can be far more recognisable to a child than any drawing.

Sayella's photo-to-symbol feature lets you take a picture, of a pet, a person, a cherished object, and turn it into a symbol that sits cleanly on a button. The child sees the actual dog, the actual cousin, the actual cup, rather than a generic stand-in. For early communicators in particular, that recognisability can make the difference between a button that gets used and one that gets ignored.

How this fits the rest of the board

A custom symbol is only useful if it does not break everything else about how the board works. Sayella's design protects that in two ways.

First, every button pairs a symbol with text, and the modified Fitzgerald colour key, the colour-coding of buttons by part of speech, applies just the same to a custom button as to a library one. A generated symbol for "Biscuit" still sits in the right colour for a noun, with the word written underneath. It belongs visually rather than looking like a patch.

Second, and most importantly, adding a custom symbol does not disturb the words your child has already learned. Sayella grows vocabulary through an approach called Progressive Language, where new words are revealed inside a grid that stays put. The layout does not reflow, and a learned button never moves. A new custom symbol fades into the grid around the words your child already knows, so personalising a board never costs them the muscle memory they have built.

Where this sits, and what it costs

AI symbol generation and photo-to-symbol are premium features in Sayella, included with the paid tiers. They are part of what you get on the monthly, yearly or one-off lifetime options, alongside cloud sync, multi-device access, premium UK voices and analytics. The free tier gives you the full fixed-position core engine, modelling tools and offline speech on one device, and the AI symbol tools are an upgrade on top of that.

It is worth being plain about why this matters competitively. No established AAC app offers on-demand symbol generation and photo-to-symbol like this. It is a genuinely new capability, and it exists to solve a real, everyday frustration rather than to add a gimmick.

An honest note on status

Sayella is in pre-launch. The app builds and runs on native iOS and the web today. The AI symbol features are built but not yet live in production, as we work through the final deploy steps ahead of a public launch around AAC Awareness Month in October. We would rather tell you exactly where things stand than overstate what is shippable this minute.

Frequently asked questions

What kinds of words is AI symbol generation actually for?

Mainly the words no library covers: names of pets and people, local foods and places, and very specific objects that matter to one child. For common everyday words, Sayella's built-in symbols already do the job.

Is a photo or a generated drawing better?

It depends on the word. For a specific person, pet or object, a photo is often the most recognisable, especially for early communicators. For more general or abstract words, a clean generated symbol usually sits more consistently with the rest of the board.

Will adding custom symbols move the buttons my child has learned?

No. Because the grid never reflows and a learned button never moves, a new custom symbol fades into the layout without shifting the words your child already relies on.

Do I need an internet connection?

The AI symbol features are part of the premium, cloud-connected experience. Core speech works offline, so your child's existing vocabulary keeps talking even without a connection.

---

Sayella is in pre-launch and we are opening early access now. If you want to be among the first to give your child a button for every word that matters to them, including the names and local words no library has ever drawn, you can join the early access list today.