Introducing Jemima, your AI powered Assistant in Brandkit

Meet Jemima, our built in AI powered virtual assistant. She provides helpful assistance and utilises powerful AI tools like OpenAI’s API(using the same Large Language Model which powers ChatGPT), and our own proprietary automations.

Enabling you to save time, be more accurate, and get through your day with less effort.

Why “Jemima”

Jemima is named after a real person. Jemima Phelps is from Brand Tasmania and responsible for managing their Tasmanian Partner Toolkit and who was the subject of a public panel discussion (in a really good way) at the recent City Nation Place Global Awards conference in London on 9th November 2023. We’ll tell that story in depth at a later date.

Setting Up

You will need to turn on Auto-tagging your settings (Admin > Settings > Automation).

Please note that Auto Tagging is on by default in new accounts.

Once that is done you have some options.

• Use AWS Image Rekognition - this is the previous generation Machine Vision based auto tagging.
• Use Open AI - new generation AI everyone is talking about.
• OpenAI API Key - you can use your own Open AI account
• OpenAI Organisation ID - you can use your own Open AI account
• Description Prompt - this allows you to manipulate results to suit your brand.
• Tagging Prompt - this allows you to manipulate results to suit your brand.

For Prompts, we have specific advice on how to set these up for best results after our testing. It’s also part of our onboarding process for new accounts.

How to use

When tagging an Asset you will now see a Jemima branded Auto-tag button. Click this button and Jemima will do 3 things for you.

1. Generate Tags from EXIF/IPTC/XMP keywords embedded in your uploaded file.
2. Call OpenAI’s API to inspect the image and generate Tags using OpenAI’s Machine Vision.
3. Copy these to the Tags field, ready for you to edit and save.

Afterwards

After Jemima has completed her task, several new Tags will appear.

You can then Edit the list of Tags and Save.

Note that Tags are now visually differentiated by origin. More on Tag orgin here

Batch auto-tagging

If you have Auto-tagging turned on in your account settings. Jemima will now run the same process as above automatically on file upload (on ingestion).

Images only at this stage

Please note that auto-tagging will be applied to images only at this stage. The exception being that embedded metadata will also be extracted from video and possibly other files if they are compatible.

OpenAI Token Costs

Using OpenAI is far superior to the older AWS Image Rekognition service but is not free.

Every Asset tagged or described, costs a certain amount of OpenAI tokens. We report the tokens used in our Reporting module (Admin > Resports > Open AI Calls report)

For now we are covering the OpenAI token costs, however in the future we will start billing separately for OpenAI tokens (or release new plans that include these costs). Alternatively you can configure your account to use your own OpenAI paid account (Admin > Settings > Automation).

Thanks Jemima!

Enjoy :)