See a complete AI chat analysis sample before you upload
This public sample shows the report structure, evidence standard and confidence labels from start to finish. The conversation and people are fictional, so no private chat or customer result is being exposed.
Try the free previewHow to read this example
The facts below are calculated from an illustrative 426-message transcript. The observations show what an AI interpretation looks like when it is separated from those facts. Short excerpts demonstrate the evidence format; confidence labels show how strongly the available text supports each section.
Three Friends, One Dinner, Finally a Decision
The group is genuinely close. The real recurring conflict is not affection—it is the gap between enthusiasm and follow-through.
Conversation facts
Calculated directly from 426 parsed messages in the fictional sample—not estimated by AI.
The group dynamic
high confidenceMaya creates momentum, Jules keeps the tone warm, and Sam turns vague enthusiasm into an actual decision.
- Maya initiates both planning threads in the sample.
- Jules responds most consistently but rarely chooses the time.
- Sam sends fewer messages, yet supplies the only concrete booking link.
“Friday dinner?” · “I can book the 7:30 table” · “yes please do it before we forget again”
Effort and reciprocity
high confidenceThe message shares are uneven, but the effort is more balanced than the raw totals suggest.
- Maya carries more of the conversational volume.
- Sam contributes practical follow-through rather than frequent check-ins.
- Jules acknowledges both people and prevents unanswered messages from stalling the thread.
“checking one more time — Friday?” · “Booked. Sending the address now” · “thank you both, I am actually excited”
Tone and private language
medium confidenceTeasing is a sign of familiarity here, but the transcript alone cannot prove how every joke landed offline.
- Repeated “tomorrow people” jokes refer to the group’s history of delaying plans.
- The jokes are followed by practical answers rather than withdrawal.
- No single message is treated as proof of a person’s private feelings.
“our annual tradition of deciding tomorrow” · “tomorrow people strike again”
The turning point
high confidenceThe conversation shifts from circular planning to commitment when one person proposes a specific time and another books it.
- Before 6:42 PM, the group exchanges preferences without a decision.
- After the 7:30 proposal, all three respond within the same planning thread.
- The booking message closes the loop.
“What about 7:30?” · “works for me” · “Booked. Sending the address now”
Blind spot to watch
medium confidenceMaya may be doing too much of the emotional and logistical prompting, even though the others do eventually contribute.
- The pattern is worth noticing, not diagnosing.
- A longer date range could show whether this is a recurring role or only one busy week.
- Offline plans and missing messages could materially change the interpretation.
“I refuse to let this become another “sometime soon” plan” · “checking one more time”
Likely direction
medium confidenceThe friendship looks active and mutually positive, but future plans will probably keep depending on one person introducing a deadline.
- Positive replies are common across the sample.
- Follow-through improves when the question is concrete.
- This is a pattern-based forecast, not a factual prediction.
“Pick one: Friday or Saturday” · “Friday wins” · “same time next month?”
What your free preview reveals
Your own free preview shows two generated sections before checkout, including their evidence and confidence labels. If the observations feel generic, the excerpts do not support them, or the limitations make the report unsuitable, you can stop without entering a card.
What the report verifies—and what it cannot
Message totals and participant shares come from the parsed transcript. Evidence excerpts are checked against that transcript before they are displayed. The AI organizes patterns and proposes interpretations, but it cannot recover missing context, observe events outside the chat, or know anyone's private intent.
Frequently asked questions
Is this a real customer report?
No. The people, messages and findings are fictional. The sample demonstrates the report format without exposing a customer conversation or implying a testimonial.
Will my report use the same findings?
No. Your preview is generated from the conversation, participants, focus and language you choose. The headings may be similar, but the evidence and interpretation should come from your upload.
What can I inspect before paying?
After email verification, you can read two sections from your own generated report, including evidence excerpts and confidence labels. No payment card is required for that preview.
Are quoted messages checked?
When a report presents an excerpt as evidence, the server checks that the text exists in the parsed conversation. Interpretation can still be imperfect, so every section also carries a confidence label.