简体中文
English
Website review
This review may include AI-generated content and is not individually verified. Domain, certificate and other facts come from the observed data on this page.
What is What Brandon Thinks?
What Brandon Thinks is an AI chat-analysis service that reads your exported WhatsApp or iMessage conversation and returns a written report about the relationship behind it. You upload the chat, and the reported output covers who cared more, what changed over time, and blunt commentary on the people in the thread.
What Brandon Thinks
What it actually does
- Takes a chat export as input.
- Produces a narrative report rather than charts or sentiment scores.
- Report angles named on the site: who cares more, what changed, and the writer's honest opinion of each participant.
- A testimonial mentions a follow-up report with "what to do about it," so advice appears to be a second step rather than part of the first read.
Who it's for
The site's own examples show the range it's aimed at: situationships, partners and spouses, exes, crushes, talking stages, best friends, family, group chats, trip chats and work chats. In practice that means two very different audiences:
- One-to-one romantic or ambiguous threads — people rereading messages after a breakup or during a stall, trying to work out whether they were imagining the drift.
- Group chats — long-running friend or family threads where the appeal is more social than therapeutic. One testimonial describes a fifteen-year boys' group chat where a per-person "awards" section got screenshotted and argued over for a week.
What you're trading off
- Narrative over evidence. You get a story with a confident voice, not a breakdown of message counts or response times you can verify. Useful when you want the pattern named; weaker if you want to check the reasoning.
- Bluntness is the product. The framing is deliberately opinionated, including about third parties who never agreed to be analysed. Testimonials suggest it lands as funny among friends; that's a different thing from it being kind.
- Emotional timing matters. Several quoted users describe reading it late at night after a breakup. A report that confidently assigns blame is easy to over-trust in that state, so treat it as one outside reading rather than a verdict.
- Privacy. You're handing over an entire intimate conversation. Decide before uploading whether the other people in it would be fine with that.
Next step
Export one specific chat you're actually unsure about — not your whole message history — and read the report against your own memory of the timeline. If the report names a turning point you'd already half-noticed, that's a useful signal. If it contradicts your memory, go back and check the messages yourself before acting on it.
How does uploading a WhatsApp or iMessage chat work to get an AI report?
Uploading a chat means exporting your conversation from WhatsApp or iMessage as a file, then handing that file to the site. What Brandon Thinks then produces a written report about the relationship dynamics inside it — who initiates, who pulls away, what changed and when — rather than a generic summary of topics.
The practical flow
- Export the chat. In WhatsApp, open the conversation, use the export chat option, and choose whether to include media. In iMessage, you export from the conversation on a Mac or via your Apple account. Media adds size and rarely changes the analysis, so text-only exports are usually the cleaner choice.
- Upload the file. The exported transcript goes to the site, which reads the messages and writes the report.
- Read the output. Expect observations about patterns over time, shifts in tone or effort, and a blunt character read on the people involved.
What makes it work — and where it struggles
The quality depends almost entirely on the export. A chat with timestamps, clear speakers and months of history gives the analysis something to work with. A short thread, heavy on reactions and voice notes, gives it very little.
Two things to weigh before uploading:
- Privacy. A full chat export is one of the most personal files you own. It contains other people's messages, and they did not consent to being analysed. That is a real trade-off, not a technicality.
- Interpretation. The report is one reading of a transcript, not a verdict. It cannot hear tone, see a face, or know what happened offline. Treat it as a second opinion that is unusually direct, not as evidence.
A concrete scenario
Someone rereading three weeks of messages at 3am after a breakup is the natural user here. The value is not in learning new facts — they already lived them — but in having the timeline laid out plainly: here is the week the replies got shorter, here is who stopped asking questions. That is a framing exercise, and it is genuinely hard to do on your own when you are inside the situation.
Decision criterion
Upload if you want a structured, opinionated read on a long text history and you are comfortable with the privacy cost. Skip it if the conversation is short, mostly voice notes, or if the other person would be badly hurt by knowing it happened.
What does the report reveal about who cares more or what changed in a relationship?
The report is built to answer two questions directly: who is putting more into the relationship, and when the dynamic shifted. According to the page, Brandon reads every message and writes what he really thinks about everyone in it — so the output is opinionated rather than neutral, closer to a blunt friend's read than a therapy summary.
H3 What it tends to surface
- Effort imbalance — who initiates, who replies late, who keeps the conversation alive. One testimonial describes a situationship report laying out "who initiated, who disappeared, what I said yes to."
- Timing of change — a specific turning point rather than a vague vibe. Camille's report "found the exact week he started pulling away."
- What went unsaid — the subtext under the messages. Chloé's report "put into words what each of us was actually saying underneath."
- A verdict plus a next step — the page mentions a second report telling you what to do about it.
H3 Where it works better and worse
| Chat type | What you get | Trade-off |
|---|---|---|
| Ex or fading partner | Turning point, who pulled away | Painful to read at 3am; can confirm what you already suspect |
| Situationship | Pattern of initiation and avoidance | Ambiguity may get flattened into a cleaner story than reality |
| Long friendship or family | Shift in tone over years | Twelve years of messages is a lot to interpret; framing may feel presumptuous |
| Group chat | One-page-per-person read, plus awards and predictions | Playful and shareable, but it's entertainment as much as insight |
The "who cares more" answer is the headline, and it's also the part to treat with care. A report can only see what was typed; silence, phone calls, and things said in person don't show up. If you want a second opinion before acting on it, a real conversation, a therapist, or a neutral friend will weigh context the chat can't.
A sensible next step: upload one specific chat, read the "what changed" section first rather than the verdict, and note whether the turning point matches your own memory. If it does, the rest is worth taking seriously. If it doesn't, you've learned the limits of reading a relationship from text.
For a different angle on the same habit, The Gottman Institute publishes research-based material on conflict patterns and repair.
Can I use What Brandon Thinks for group chats or work chats, not just romantic relationships?
Yes. What Brandon Thinks is built for any chat you can upload, and its own examples include friend groups, family, trip chats and a work chat alongside exes, crushes and situationships.
H3 What it handles beyond romance
- Friend groups: the site's example "The Boys" chat and a testimonial from Hugo describe a 15-year group chat where each member got roughly a page, plus an awards section and predictions the group argued over.
- Best friends and family: examples include a best friend, "Mom" and a family chat; Chloé's testimonial covers 12 years of messages with a best friend and a shift in the relationship.
- Work and trip chats: "Startup team" and a trip chat are listed as upload options, so a team thread or holiday planning group is within the intended use.
H3 What changes in a group or work setting A one-to-one chat gives the tool two voices to weigh. A group gives it many, so the useful output shifts from "who cares more" toward roles, patterns, who initiates, who goes quiet and how each person comes across. Hugo's observation is that the report named all seven members without anyone taking offence — worth noting, because a work chat carries more risk than a friendship group if a colleague reads a blunt character summary.
H3 Practical next step Before uploading, decide who will see the report and whether every member would be comfortable with it. If it is a work chat, consider whether you are using it for your own read on the dynamic rather than as something to share. If it is a friend group, sharing is clearly part of the appeal — the testimonial describes the awards section being screenshotted into the chat within a minute.
For a second opinion on the same material, you could compare notes with a general AI assistant such as ChatGPT, though that means writing your own prompt rather than getting a structured report.
How accurate or honest is Brandon's opinion compared to a therapist or friend's advice?
Brandon is built to give you a blunt, opinionated read of a chat log, not a clinical assessment. The site's own framing is that he "reads every message and writes what he really thinks about everyone in it" — that's a deliberate persona choice, closer to a sharp-tongued friend than a neutral analyst. So "accuracy" here means pattern-spotting and plain speaking, not diagnosis.
Where it can genuinely help
- It reads the whole thread at once, so it can point to things people miss in the moment: who initiated, who pulled back, when the tone shifted. Testimonials on the site describe exactly this — Camille says it identified "the exact week he started pulling away," and Valeria describes seeing "who initiated, who disappeared, what I said yes to" laid out on one page.
- It says the uncomfortable thing without social cost. A friend may soften a verdict to protect the friendship; a paid report has no such incentive.
- It can be useful for group chats too, not just romance — Hugo's testimonial describes one-page reads on all seven members of a fifteen-year group chat.
Where it differs from a therapist or a friend
| Brandon | Therapist | Close friend | |
|---|---|---|---|
| Basis | The messages you upload | Your account plus history, context, questions | Shared history and knowledge of you |
| Tone | Blunt, verdict-style | Neutral, exploratory | Variable, filtered by loyalty |
| Can ask follow-ups | No | Yes | Yes |
| Accountability over time | None | Ongoing | Ongoing |
| Cost of being wrong | Low stakes, but you may act on it | Low, corrected in session | Can damage the friendship |
A therapist's value isn't that they're more honest — it's that they can question your framing, notice what you left out, and stay with you as things change. A friend's value is that they know you outside this one conversation. Brandon only sees the text, and text is a partial record: it can't show what was said in person, what was going on in someone's life, or what you chose not to send.
Honesty is not the same as accuracy
The persona is designed to sound certain. That's engaging, and it's also the risk: a confident verdict about someone's motives reads as insight even when it's inference. Treat the report as a set of hypotheses to test against what you already know, not a verdict.
A practical way to use it
If you're deciding whether to try it, pick a chat where you're genuinely unsure what happened, and read the output asking one question: does it point to specific messages, or just assert a mood? Specific claims you can check are useful; sweeping character judgments are entertainment.
If the report confirms something painful — a partner pulling away, a friendship fading — use it as a prompt to have the actual conversation, or to bring the question to a therapist who can help you decide what you want. The site is at What Brandon Thinks. For the underlying relationship question, a licensed therapist or a trusted friend who knows both people will still give you something a chat log cannot.
Is it private or safe to upload my personal chat messages?
Whether it's safe depends on how the service handles your data, not on how the tool feels. What Brandon Thinks is built around uploading a real WhatsApp or iMessage chat, so the messages you send include other people's words, not just yours — that's the core trade-off to weigh before you start.
What the site actually shows
The page invites you to upload a chat and get an AI report on "who cares more, what changed, and what Brandon honestly thinks of everyone in it." Testimonials describe people uploading breakups, situationships, group chats and family threads. There's no visible privacy policy, retention statement or deletion control on the page evidence, so you can't currently confirm from the site itself where your file goes, who can read it, or when it's deleted.
The real risk isn't you
Your own messages are yours to share. The problem is that a chat contains other people's messages, names, phone numbers, and often health, money or relationship details they never agreed to hand to a third party. In group chats that's a dozen people at once. If the report is generated by an AI system, the content is also processed by that provider, which is a second party beyond the site itself.
Practical ways to decide
- Check before uploading: look for a privacy policy, a stated retention period, and a delete-my-data option. If none exists, treat the upload as effectively permanent.
- Redact first: replace full names with initials, strip phone numbers and addresses, and cut anything medical or financial.
- Start small: upload a short, low-stakes thread — say a talking-stage chat — rather than twelve years with your best friend.
- Ask the other people: if the chat is with a partner, friend or family member, they're in the file too.
- Prefer on-device or self-hosted tools if the content is genuinely sensitive; those keep the transcript on your machine.
If you want a second opinion on what a chat means without exposing it, a therapist, a trusted friend, or reading it yourself with a clear head carries zero data risk. For general guidance on how services should handle personal data, see UK Information Commissioner's Office and US Federal Trade Commission.
A reasonable next step: open the site, search the page for "privacy," "delete" or "data," and if you can't find them, either redact heavily or don't upload.
vercel-dns.com
Google Workspace
User reviews (0)