Trust
What Tsuno does, and what it refuses to do.
A tool that knows a lot about your teams has to say plainly what it will never do. Here is the rule, the refusals, and what we can or cannot prove.
The doctrine
Two sentences, and nothing behind them that contradicts them.
Tsuno connects the facts.It never judges a person.
The tool reminds. People decide.
-
Facts side by side
Validated training, checklist done or not, incident, hours, agreed goal: dated, sourced, brought up at the right time.
-
Process before blame
When something does not hold, Tsuno looks first at the process, the training, the schedule.
-
A person signs
A decision that affects a person stays with someone who knows them and signs it.
“Instead of putting the blame on the employee, we put it into process, training and skills.”
Page contents
What is allowed
Facts, agreed goals, a next action.
Everything Tsuno does involving people rests on things you can reread, date and discuss with them.
- Dated, sourced facts, with a note of where they come from.
- Training status: what is validated, what is still to validate.
- Goals agreed with the person.
- Hours, as declared or recorded.
- Incidents and tasks, with their history.
- Fact-based one-to-one prep: what happened, what had been agreed, nothing else.
- A suggested next action that you accept, change or ignore.
- Help wording a measurable goal, without setting it for you.
What is refused
Refused by design.
It is not a setting you can switch on or a module you can buy. Tsuno is built not to do it.
Available
- Ranking employees from best to worst.
- Giving a person an overall score.
- Predicting who will resign.
- Recommending who to dismiss.
- Inferring a person's motivation, honesty, loyalty or mood from their messages.
- Sending a sanction automatically.
- Automating an adverse HR decision.
- Cross-referencing sensitive data to profile a person.
From your AI, too
These requests are also refused from your AI (ChatGPT, Claude, Mistral). The refusal does not depend on where you ask the question.
Fictional examples: requests and answers are given for illustration.
“Rank my servers from best to worst.”
I do not rank people. I can show you the dated facts instead, by role and not by person. Would you like to see where a validated training is missing on the evening shift?
“Tell me who is going to quit before summer.”
I do not predict departures and I do not infer anyone's intentions. I can remind you of the facts: overdue one-to-ones, goals still open. Would you like to prepare those one-to-ones?
Data quality
Where each fact comes from.
A summary built on mechanically ticked boxes is worth nothing. Tsuno tells six kinds of fact apart, never equivalent.
-
Automated fact
Recorded without input: a timestamp, an action logged by Tsuno itself.
-
Declared fact
Someone says it is done. Useful, but declared.
-
Attached proof
A photo, a document, a receipt tied to the fact.
-
Manager observation
What a manager saw and noted, with its date and author.
-
AI inference
A link suggested by the AI, flagged as such and never presented as a fact.
-
No data
Nothing was recorded. Tsuno says so instead of filling the gap.
What a summary must flag
When a summary is fragile, Tsuno says so.
- Missing data: what should be there and is not.
- Conflicting data: two sources that do not say the same thing.
- Low reliability: facts declared without proof, or filled in by the batch.
- No proof: a claim that nothing backs up.
Never “all good” when the reality is “no reliable data.”
Actions
Nothing runs without your confirmation.
Whether you work in Tsuno or from your AI, an action always follows the same path.
Available
-
Tsuno prepares
It drafts the action from what you ask.
-
It sums up the effect
What will change, for whom, and where.
-
You confirm
Nothing goes out until you say yes.
-
It runs and logs
The action is carried out, then logged.
Out of reach, even when confirmed: payroll, invoicing and permanent data deletion never go through actions prepared by the AI.
Demos
Never a restaurant's real data in front of an audience.
A demo runs in an isolated environment, with synthetic data.
Available
- Public demo: a separate environment, made-up synthetic data, no real personal data.
- Prospect trial: an isolated environment, built from three of your non-sensitive operational documents.
- No personal data in the trial: no employee or HR documents. Anonymize schedules.
- Access limited to the people you name.
- Deletion on an agreed date, set with you from the start.
- No mixing between clients: what is in your environment exists only in yours.
The full walkthrough is on the Setup page.
Reversibility
Your data belongs to you.
Export and deletion terms are set out in your contract. A question before you decide? Write to us.
For how personal data is processed, see our privacy policy (in French).
Measurement
How we measure what we claim.
Our rule: no figure without a written measurement protocol. No numerical result is published here without its protocol.
What we will publish, with its protocol
The indicators we measure, with a written protocol, from among these:
- Time to autonomy at opening
- Manager reporting time
- Recurring incidents closed
- Food waste
- Time from an anomaly to action
Each result will be published with its protocol, so you can judge the method and not just the number.
Frequently asked questions
Why does Tsuno look at the process rather than the person?
Because Tsuno connects the facts and never judges a person. When something does not hold, it looks first at the process, the training, the expected skill, the schedule. Not for someone to blame. It reminds you of the facts at the right time, without deciding for you what someone is worth.
Can a client switch on employee ranking or scoring?
No. It is not a setting you can switch on or a module you can buy: Tsuno is built not to do it. The refusal also applies when the request comes from your AI (ChatGPT, Claude, Mistral). Tsuno then offers the dated facts, by role and not by person.
Which documents can I send for the trial?
Only non-sensitive operational documents: a recipe, a procedure, a checklist, an anonymized schedule. No personal data: no employee or HR documents, no names, no contracts, no payslips. The environment is isolated, access is limited to the people you name, and it is deleted on the date agreed with you.
Can a confirmed action touch payroll or delete data?
No. Some things are out of reach of actions prepared by the AI, even when confirmed: payroll, invoicing and permanent data deletion. They do not go through that path. Every other action follows the same order: Tsuno prepares, sums up the effect, you confirm, it runs and logs.
Why do you publish no figures from Kajirō?
Because what we observe in our own group is not a measurement protocol. So we draw no figures from it, and nobody should use it to estimate what Tsuno would bring to their business. A result will only be published with its written protocol.
How do I export or delete my data?
Your data belongs to you. Export and deletion terms are set out in your contract. For a question before you decide, write to [email protected]. How personal data is processed is detailed in our privacy policy.
See Tsuno on your own documents.
Send 3 documents. We set up an isolated environment, walk you through it in a meeting, then delete it on the date we agreed together.
Non-sensitive operational documents only (recipe, procedure, checklist, anonymized schedule), with no personal data at all.