Answers to your most common questions about Ziflow’s revolutionary generative AI collaboration tool.
Common questions and answers:
What is ZiflowAI?
ZiflowAI is a chatbot powered by artificial intelligence built inside of Ziflow’s Proof Viewer. Any of your proof’s reviewers can ask it a question and in seconds, ZiflowAI will deliver an answer that can help your team arrive at a creative solution to your creative issue.
Generative AI tools use large, multimodal language models based on a broad range of data captured from the internet to respond to questions in a conversational way. ZiflowAI uses the same technology but for the specific application of creative collaboration.
Why should my team use ZiflowAI?
A proofing process is usually restricted to the review team’s own experiences and imaginations. Your team can increase the creativity levels within that project simply by asking for an outside opinion that’s powered by artificial intelligence: ZiflowAI.
What is the generative AI engine that powers ZiflowAI?
We use ChatGPT (v. 3.5) by OpenAI via their API to power ZiflowAI. As a result, OpenAI acts as an outside provider and is named on Ziflow’s list of sub-processors. We prime the model for creative and marketing use cases to give the chatbot a narrower context when it receives queries to help it provide responses that are more informed.
What data does ZiflowAI have access to, and when does it get access to it?
ZiflowAI will not have access to any of your team’s information until the moment one of your reviewers takes specific action to prompt the chatbot in a given proof. From there, ZiflowAI will receive only the text of the query that your reviewer makes within their comment, along with the information from your Proof Brief to give the AI additional context.
ZiflowAI doesn’t have access to any other information from within your instance of Ziflow (user names, email addresses, company name, etc), nor will any of this information be accessible from proof to proof, meaning: all interactions between your team and ZiflowAI are siloed within an individual proof.
What does ZiflowAI do with the data or questions I’ve submitted to it? Will other companies or my competitors have access to sensitive data?
According to OpenAI’s API terms of use and FAQs: “data submitted through the OpenAI API is not used to train OpenAI models or improve OpenAI’s service offering.” In addition, OpenAI states that they do not share content with third parties for marketing purposes.
For each request delivered through OpenAI’s API, the only content ZiflowAI sends to ChatGPT is the content of a Proof Brief and the query in the comment thread itself. Ziflow does not send any personal information such as the name of the reviewer or their email address.
Are the responses my team will get from ZiflowAI brand-safe?
ZiflowAI does not contain any additional filtering for trademarks or creative commons licenses, nor for language that may be offensive to other cultures or sub-groups. As a result, we encourage your reviewers to verify separately any response from ZiflowAI prior to including it in your final, published materials.
Are the responses my team will get from ZiflowAI verifiably true?
Artificial intelligence and machine learning is a rapidly evolving field. ZiflowAI is built on similar technology, and as result, some interactions with the chatbot may result in output that does not accurately reflect facts. We encourage your reviewers to verify the accuracy of any response from ZiflowAI prior to including it in your final, published materials.
Will ZiflowAI respond with language that’s offensive?
Having been built on a large multimodal language model, ZiflowAI will respond to any question you ask it—regardless of the content of the question itself. As a result, if a reviewer asks ZiflowAI a question where you’d expect an offensive response, there’s a reasonable chance that you’ll get one. Given that Ziflow is not responsible for the questions your reviewers may ask it, Ziflow is not responsible for the answers that ZiflowAI delivers in return.
As a general guideline, we would encourage your team to abide by your company’s accepted rules of behavior and decorum when interacting with the chatbot. So, if your reviewer wouldn’t feel comfortable asking a colleague a question in a company-wide Slack or Microsoft Teams channel, then they should avoid asking the same question of ZiflowAI.
How can I find out more about ZiflowAI?
Ziflow’s website has a dedicated page about this exciting feature. In addition, there’s an article in our Help Center that comprehensively covers this topic, and we have an episode of our ZiflowPro series that discusses best practices on the use of the tool.
If you have any additional questions about ZiflowAI that aren’t answered in this FAQ, don’t hesitate to reach out to your customer success manager, or use the help widget to ask a question.