What Brandon Thinks about your chat.
Try a small AI sentiment model in your browser, then explore fictional conversations and the messages behind each sample finding.
A small model, a closer look.
Paste a short English message or use the fictional example. A sentiment model estimates whether its wording is positive or negative. It cannot read tone of voice, understand your relationship or tell you what someone really thinks.
Privacy: This Space runs the model in your browser and does not send the text you enter to our server or an AI API. The first run downloads model and runtime files from Hugging Face and jsDelivr (about 65 MB for the model); your browser may cache them. Network providers can see those file requests. Avoid including sensitive or identifying details.
This is a limited English-language sentiment model trained on movie reviews, not a relationship analysis model. Results can be misleading, particularly for sarcasm, mixed feelings, short texts and other languages.
0 / 1,000 characters · not saved by this app; clear when finished
Model loads only when you choose Analyze sentiment.
This is the model’s confidence in its label, not a measure of someone’s feelings or the accuracy of the interpretation.
Meet the conversations.
Pick a scenario, explore Brandon’s read, then find the messages that support it. Everything here is fictional.
A reading, not a verdict. Text alone cannot prove anyone’s intentions.
Curious about your own conversation?
Start a personal preview on whatbrandonthinks.org, or explore the complete fictional report first.
A clearer read, with room for nuance.
What Brandon Thinks turns supported WhatsApp and iMessage exports into an AI-assisted report. This Space includes a separate local sentiment demo and fictional conversations with prewritten observations.
What the full product does ↗Follow the evidence
A useful finding gives you a place to look in the conversation. Read the surrounding messages too.
Make space for another read
The same words can mean different things. Offline context, tone and timing matter.
Choose what happens next
Use a report to form a better question, not to decide what someone secretly feels.
The small print,
made simple.
A few things to know before bringing your conversation anywhere.
What is What Brandon Thinks?
What Brandon Thinks is an AI-assisted chat analysis service for supported WhatsApp and iMessage exports. Its reports discuss communication patterns and link important findings to messages. Learn more on the product review page.
Is this demo analyzing my chat with AI?
The optional sentiment demo runs a small English classifier in your browser on text you enter. The conversation examples and their findings are fictional and prewritten. A personal report on the main website is a separate AI-assisted experience.
Does this Space collect chat records?
This Space has no chat-file upload. The optional text input is processed in your browser and is not sent to our server or an AI API. The model and runtime files are downloaded on first use. Visiting the main website is a separate experience with its own privacy and safety information.
Can I try What Brandon Thinks for free?
The main website offers a personal preview before payment. See the current free preview details and the instructions shown there.
Can a report know what someone really thinks?
No. It can suggest interpretations of the text you provide, but it cannot know private thoughts or replace the context of your relationship. It is not therapy or professional advice.