Mend
Try Seam →

How Seam Works

Explain What's Going On. Thimble Will Take It From There.

No menu to click through and nothing to configure. Say what's going on, and Seam figures out whether it's a team conflict, a hard conversation you need to prep for, or a message you want checked, then runs the right kind of conversation.

trymend.solutions/seam/
Anonymity enforced
Conversations
Launch timeline friction
Team dynamics
When did you first notice the handoffs slipping?
A few weeks ago, right after the launch date moved up.
What changed on their end once the timeline shifted?

An actual Seam session, framed for illustration. Colors are the real product UI, not restyled for the marketing page.

Why Mend was built for managers

Mend focuses on the structural forces that cause conflict between people on your team, rather than asking you to referee a personality clash, leveraging current conflict-resolution best practices. Read more about our approach.

Thimble vs. a generic AI chatbot, for managing someone else's conflict

Every manager already has a general-purpose AI chatbot in their pocket. It's a fair question to ask why that isn't enough here. Side by side, these are the places it actually matters:

Where it matters
Thimble
Generic AI chatbot
Whose side is it on?
Thimble
Not simply on your side. It pushes back if your plan would be unfair, vague, or likely to blindside the person who isn't in the room.
Generic AI chatbot
Agrees with whatever framing you type in. It has no reason to question your read on your own report.
Behavior vs. character
Thimble
Pushes back the moment you slide into a character judgment and makes you name the specific, observable behavior instead.
Generic AI chatbot
Happily drafts a script around “they're lazy” or “they don't respect me,” which reads as a character attack once it's actually said out loud.
A real conversation structure
Thimble
Won't call it done until six things are nailed down: the behavior, why it matters, what needs to change, what support you can actually offer, how to open it, and what's already decided elsewhere.
Generic AI chatbot
Gives you exactly what you asked for and nothing more. No structure is enforced unless you already know to ask for it.
Rehearsing the conversation
Thimble
Plays a plausible, realistic version of your report, then offers a second pass with a different reaction, so you're not just prepping for one script.
Generic AI chatbot
No built-in sense of staying in character. Drifts toward whichever reaction is easiest to generate: instant agreement, or manufactured hostility.
A real legal/safety floor
Thimble
Screens every message. If what you describe crosses into discrimination, retaliation, or illegal conduct, it stops coaching the plan and tells you plainly this goes to HR.
Generic AI chatbot
No floor at all. It will help you word a write-up for an unfair or even retaliatory action just as readily as a fair one.
Staying in its lane
Thimble
Never tells you the actual personnel outcome (a PIP, a warning, termination). Says plainly that's HR's call, not its.
Generic AI chatbot
Will confidently tell you what to do, with the false authority of a decision it has no business making.
What happens to what you typed
Thimble
Nothing is stored. The conversation is held in memory for the session and flushed the moment it ends.
Generic AI chatbot
Logged by the provider under whatever retention policy applies to your account. It wasn't built to hold a conversation about someone's job.
An ongoing case, not one chat
Thimble
Tracks a case over time: a running action plan, who's involved by role, and what you tried last, so you can come back to it a week later and pick up where you left off.
Generic AI chatbot
Starts from zero every new chat. No memory of the case, what you already tried, or how it went.

All of this is prompt and application engineering layered on top of the same kind of underlying model a generic chatbot uses. The difference is the structure, the floor, and the escalation plumbing built around it, not a secret smarter model.

Will managers actually use this?

None of this matters if a manager doesn't feel safe using it, because they fear HR or their own boss is “listening in” on a conversation about their own shortcomings as a manager.

That's why Mend is built from the ground up with total anonymity: no persistent record of what's shared. Session content is never truly saved; it's held in volatile memory and flushed as soon as the session ends.

How the Privacy Actually Works

Not a policy promise: the technical mechanics.

What is RecordedNothing. Sessions run in volatile memory enclaves and are never written to disk. We don't "record then delete": it is never saved in the first place.
What Managers SeeZero visibility into voluntary sessions. For sessions formally assigned by a manager, they receive only a binary confirmation: "Resolved" or "Unresolved", never conversation content.
Automatic Safety ProtocolsIf what's being discussed meets the legal standard for discrimination, retaliation, or a physical threat, Thimble escalates on its own and the manager is always told plainly, before anything is sent.

Mend isn't for every company

The deal Mend makes with customers: they're paying for better-handled conflict and stronger teams, not control of the process or visibility into every manager's conversation. Control is exactly what prevents a manager from admitting, honestly, that they're stuck.

Why an AI facilitator works better than a person

  • An AI facilitator is unbiased and has no stake in whether you or your report comes out looking better.
  • Managers are measurably more comfortable admitting they're unsure how to handle something to an AI persona than to another manager or HR.
  • Mend is always available and doesn't pull a peer manager's or HR's time into a problem that's still yours to work through.

The research against “no friction, no dissent”

A team with zero visible conflict isn't necessarily a healthy team you're managing well, it can just as easily be a team that's learned not to disagree with you.

Visible conflict isn't the problem.Noted researchers Lencioni, Edmondson, and Gino all point to healthy disagreement as what separates high-performing teams from quietly dysfunctional ones. That's why destigmatizing conflict matters for the managers running those teams too: only once it's safe to name can it actually get resolved. Left alone, it doesn't disappear, it just sits as an unaddressed wound on the team.

Ready to Try It?

Bring a real situation from your team, or use one of the sample scenarios. Off the record, zero judgment, and completely confidential.

Start with Seam →

Testing Seam? Continue to the survey →

How good a read do you have on your own team's conflicts? Take the 2-minute test →

More on workplace conflict