Toasted

toasted

FAQs

FAQs

Why use Toasted?

We're often advised to write for our audience, but that's easier said than done, especially when we have little in common with our readers. Toasted estimates how strongly your writing will appeal to your audience and makes suggestions on how you can improve your writing. This way, you can write for an accountant even if you've never made a spreadsheet.

What do Toasted's predictions mean?

When Toasted predicts, for example, that your job description is fairly effective for an audience of software engineers, it does not imply that every single software engineer will be receptive to it. Rather, Toasted predicts that compared to other job descriptions, software engineers will — on average — be more receptive to yours. Not everyone in your audience will feel exactly the same way, and this is especially true for large and diverse demographics like men, women, retirees, and college students.
What can I use it for?

You can use Toasted for any kind of writing! Businesses primarily use it for writing job descriptions and ad copy. For example, if you were a tech company trying to hire more engineers, you might run your job descriptions through Toasted before posting them.

How does it work?

Toasted uses natural language processing (NLP), a subfield of AI. It is based on a semi-supervised approach, combining generic and widely available corpora with data specific to the audience and document type. Some types of text are overrepresented among certain demographics — for example, ad copy for cosmetics products is primarily written for women. Because of this, there's a fine line between capturing age-, gender-, and race-specific biases and capturing genuinely meaningful patterns in the data. To walk this fine line, Toasted uses a combination of human- and machine-based oversight.

How can I share my document?

Click the Link icon (the chain at the bottom) to generate a url to your text.

None of the suggested words are a good fit. What do I do?

Sometimes, when a word has multiple meanings, Toasted's suggestions won't be a good fit. In such cases, we recommend editing the word yourself to see how much you've improved.

How can I save my document?

To respect your privacy, Toasted does not permanently store any of the text that you analyze on our servers (excluding shortlinks, which are anonymously mapped to a compressed version of the text they are linking to). To permanently save your text, we recommend copying it to a file on your computer.

How can I contact you?

We'd love to hear from you! Please send us an email at info@isittoasted.com.
Close
Ethics
Ethics

We're often advised to write for our audience, but this is easier said than done, especially when our audience does not look the same, speak the same, and have the same life experiences as we do. Learning to write for different audiences, however, is sound business sense: it helps us consider outlooks that we might have otherwise overlooked and prevents others from overlooking us. Toasted's goal is to help companies communicate with everyone. By increasing the diversity of companies' customers and workforces, Toasted aims to revolutionize the way businesses do business.

Given the degree to which AI has been fraught with ethical concerns, we as a company abide by the five following principles:
1. Protecting your Privacy
  Privacy is one of our top priorities. Save for any shortlinks you create, we do not persist any of the data that you analyze on our servers. We save the document that you're currently analyzing as a server-side cookie and then delete it once your session is over.
2. Using Sound Science
  Toasted uses state-of-the-art methods in natural language processing to extract meaningful patterns from large corpora. Our team members have years of expertise in NLP and are quick to enhance our model as developments are made in the field.
3. Maintaining Transparency
  AI can often be a black box, meaning that its reasoning is opaque to users. At Toasted, we care about the 'why' more than we do the 'what', which is why we provide detailed explanations for why a document is assigned a specific score. This also means that we will always be transparent about the limitations of what Toasted can do.
4. Ensuring Fairness
  Even well-intentioned AI systems can become prejudiced, in part due to implicit bias buried in the training data. Toasted uses a combination of human-and machine-based oversight to address this problem.
5. Kaizen
  At Toasted, we adhere to the Japanese business philosophy of continuous development. We continuously incorporate customer feedback and new training data into product development so that Toasted gets fairer, more accurate, and easier to use over time.
We've trained Toasted on a combination of generic corpora and audience-specific data. This means that certain types of writing are overrepresented among certain demographics — for example, ad copy for cosmetics products is primarily written for women. Because of this, there's a fine line between capturing age-, gender-, and race-specific biases and capturing genuine patterns in the data that can improve your writing. To walk this fine line, we use a combination of human-and machine-based oversight: algorithms flag potential biases in our model that we then manually review.

We recognize the capabilities of the technology behind Toasted and work to ensure that it is not used for applications that cause harm or unfairly disadvantage certain demographics. As Toasted continues to grow, we will continue to work with representatives of all backgrounds to ensure that businesses can use Toasted fairly and effectively.
Close
Examples

I'm writing for

Terms of Use

Careers