How It Works

From messages to a pattern you can actually read

ChatMend takes a conversation you already have and turns it into a clear, evidence-based report. Here is exactly what happens — and how your data is handled at every step.

01

Bring a conversation

Paste a text thread or upload a file (plain text, CSV, or JSON export). ChatMend reads and parses the messages right in your browser to detect who said what.

Parsing happens on your device. Nothing is sent yet.
02

Map the two people

Confirm which two speakers to analyze and give them names you recognize. If more than two people appear, you choose the pair.

You stay in control of what gets analyzed.
03

Choose depth and analyze

Pick the relationship type and how deep you want to go. Only the selected messages are sent over an encrypted connection for AI analysis.

Sent securely over HTTPS/TLS, processed in real time.
04

Read a clear report

Get a structured report: a health read, attachment signals, the conflict cycle, evidence linked to messages, and grounded next steps.

Saved to your account so you can revisit or delete it.
Under The Hood

What the analysis actually looks at

ChatMend reads communication behavior — not personalities — and grounds every read in attachment-informed research.

🧩

Attachment signals

Patterns that may suggest secure, anxious, avoidant, or fearful-avoidant tendencies — described as signals, never diagnoses.

🔄

Conflict cycles

Recurring loops like pursue-withdraw: who moves toward connection under stress, and who moves away.

🔥

Emotional tone

Where warmth, tension, defensiveness, or shutdown appear — and which topics shift the temperature.

🤝

Repair attempts

How often repair is tried after friction, who initiates it, how long disconnection lasts, and whether warmth returns.

📈

Communication shifts

How the dynamic changes over time — response speed, message length, and engagement across the timeline.

📋

Evidence linking

Each finding ties back to specific patterns in the thread, so insights are grounded rather than vague.

Where Your Data Goes

Privacy is part of the flow

You are analyzing conversations you would never want exposed. Here is the honest path your data takes.

1

Read on your device

Files and pasted text are parsed locally in your browser. Speaker detection happens before anything leaves your device.

2

Only the selected messages are sent

When you run the analysis, the chosen messages are transmitted over encrypted HTTPS/TLS to our cloud function for processing.

3

Not retained after analysis

The conversation text sent for analysis is processed in real time and isn't kept on our servers after the response is returned.

4

Results live in your account

Your finished report is saved to your account so you can revisit it — and you can delete any report at any time.

Choose Your Depth

Three levels, same evidence-first approach

Pick how thorough you want the read to be. You can always run a deeper analysis later.

Snapshot

Up to ~500 messages

A quick read: health grade, attachment labels, and the key strengths and risks.

Popular

Deeper Dive

Up to ~5,000 messages

Detailed profiles, conflict themes, and therapeutic recommendations.

Comprehensive

Up to ~10,000 messages

Full analysis with quadrant mapping, suggested scripts, and a communication timeline.

See the kind of report you get

Walk through a complete, anonymized example before you analyze anything of your own.

View the Sample Report →

ChatMend supports reflection and therapy preparation. It does not diagnose and is not a replacement for a licensed professional. AI analysis can be wrong and cannot see the full context of your relationship.

Ready to see your pattern?

Bring a conversation and get a clear, evidence-based read in minutes.