Wordable: Generate Marketing Style for Businesses

Proposals Round 11
#1

Project Statement
One statement: Generating a cohesive brand language with font, design palettes, etc for any digital brand given it’s context.

Descriptive
Right now the generative AI landscape is fragmented based on modality. My target here would to be to create a model ensemble for generating everything a business suggesting with fonts combinations, image assets, etc with each one being in synchronisation with each other.

Key component: Every model is node and rechecked after state change, eg: Font types and color of assets would be checked for compatibility score.

Deliverables
Implementing the font similarity module and synchronisation mechanism to adjust all models.

Squad
Amit Singh (metamyth)
Applied AI, Paperplane Communications
Twitter: https://twitter.com/not_amyth?t=BSpbbssQAIkrRAgg1ZTUIQ&s=09
Discord: metamyth #8558
USDC Wallet Address: 0x0159af752e0220ed3eef439bef36f982cc0a6fbf

1 Like
#2

Overall, the project proposal seems interesting and addresses a practical problem faced by businesses. However, the proposal lacks details about the technical implementation and the specific algorithms and models that will be used to achieve the goals. More elaboration on the architecture and methodology would help to better understand the project’s feasibility and impact.

Additionally, it would be beneficial to provide more specific deliverables and a timeline for completing the project, as well as an explanation of how the grant funds will be used. This will help to ensure that the project stays on track and meets its objectives.

Finally, the proposal could benefit from including information about the team’s prior experience and qualifications in the field, as well as any existing work or prototypes that have been developed. This will help to establish credibility and confidence in the project’s success.

1 Like
#3

Hey Mark thanks for review. The algorithms, for font similarity module will be traversing across font embeddings and compatibility score will be mix of cosine distances and user feedback loss.
Reason I didn’t mention this is because, user feedback loss is bit vague and needs some mathematical grounding first.

As for experience, I’m a published as a researcher in AI and CogSci (read my papers by clicking my twitter bio).
And, as an engineer, my models are serving over 5000 doctors to write prescriptions, generate marketing text and identify patient’s intent behind their queries sent to the doctor. They’re all deployed in real time as I write this. Visit https://www.paperplanetech.co/ to know more about our startup.

1 Like