How to review your digital twin's conversations

Because it speaks on your behalf. A digital twin serves people with your face, your voice and your brain, and whatever it tells a client or a prospective client is, for all purposes, your business speaking. Reviewing its conversations is how you find out whether it does so the way you would.

One point before starting: session notes and follow-up for each client's twin are in development, as the table at the end shows. This guide explains what to check when you have a conversation in front of you and what you can already fix today.

It is also how you improve it. Every conversation shows what your clients really ask, what is unclear in your materials and which rules you are missing. A professional who reviews their twin's conversations ends up knowing their clients better, not just their twin.

The same things, in the same order, every time. A fixed checklist keeps the review quick and makes two reviews comparable.

It depends on the stage. At the start, every conversation: the first weeks teach the most, because each new question uncovers a gap in the brain. Once reviews stop finding faults, a regular sample is enough, plus always reviewing the conversations that end in a sale, a complaint or an appointment with you.

Note the date of each review and what you changed afterwards. ISO 30401, on knowledge management systems, is a reminder that useful knowledge is maintained, reviewed and improved; keeping a record of those reviews is part of the job.

And who reviews? Ideally you do, at least at the start, because nobody knows your way of working better. If you have a team, someone you trust can do the first pass with the same checklist, but you approve the changes to the Book of Criteria: they are your rules, and the twin speaks in your name.

In the brain, not in the conversation. If the twin explained your method badly, the problem is in the materials; if it answered something it should have passed to you, a rule is missing. The fix goes where the error came from.

In Digital Twin by Legathum, rules live in the Book of Criteria: each with its condition, its action, its reason and its source, reviewed and approved by you. Fixing a rule means editing it in the draft and activating a new version, with a note explaining what changes and why. The previous version stops applying and there is a record of which version was in force at each moment, which also helps if you ever need to explain why the twin said what it said.

If something serious goes wrong and you cannot fix it straight away, you have an immediate way out: the twin can be revoked with one click and stops speaking at once.

Two basic things. The first is to inform. Article 13 of the GDPR requires that, when data is collected from a person, they are told at that moment who the controller is, what the data is used for and on what legal basis, as well as how long it will be kept. If the twin's conversations are stored for review, your client has to know.

The second is not to keep more than you need. Article 5 of the GDPR requires data to be kept for no longer than necessary for its purpose. Set a reasonable retention period for conversations and stick to it.

If you are a lawyer or work with confidential information, also check your profession's rules on recordings and secrecy before keeping any conversation.

Nothing that has not been checked is presented as available. This table separates the two.

• Today: you fix the brain in the Book of Criteria, in versions and with a note; you can export the Book in force; the twin says it is an AI and can be revoked with one click.
• In development: session notes, next steps and follow-up for each client's twin.
• In development: automatic logging of unanswered questions and notifying the professional.