New: Narrator

When your best operator retires, the knowledge stays in the plant

Experienced employees talk freely into a voice message. No login, no forms, no follow-up questions. AI agents turn it into a knowledge base their successor can ask.

Book a demo
Voice note3:42

Maintenance, 34 years on the job

"... and on press 3 you warm the hydraulics first, otherwise the film tears ..."

being structured in the background
No reply. No follow-up. Just listening.
TalkA voice note, as long as they want
Agents structurePress 3 setup, exception captured
Successor asksAnswer with a link to the original recording
The problem

Every retirement walks 30 years of experience out the gate

The person who knows why the line behaves differently in damp weather retires in two years. That knowledge is in no manual and in no work instruction. It sits in a head that will soon not be in the plant.

Documentation projects do not fix this. They ask for effort from the one person who is already busiest. And they ask for knowledge the holder does not know is special. After a three month handover the successor has a fraction of it.

The solution

Talking instead of documenting

Narrator inverts the task. The experienced employee writes nothing. They talk whenever it suits them, for as long as they want, with no structure and no template. A voice message in the messenger they already use.

The system never replies. It does not ask, it does not correct, it does not judge. The work happens in the background: AI agents transcribe, assign topics, link recordings, and build a knowledge base that grows with every recording.

How it works

Three steps. No project.

From free talking to a knowledge base you can query, without anyone filling in a form.

1

Talk

The employee sends a voice note over WhatsApp or Telegram. No login, no app, no template. They say whatever comes to mind.

WhatsAppTelegram
2

Agents structure

In the background, AI agents transcribe the recording and place it: asset, procedure, exception, context. Every new recording is linked to what is already there.

3

The successor asks

The successor asks in their own words. The knowledge base answers and links to the original recording, so they can hear the source themselves.

Capabilities

Built for people who do not want a software project

Voice first

Talking is ten times faster than writing and carries more. Photos, video, and short text work too, whenever they fit better.

The system only listens

Narrator does not reply and does not ask questions. No interview, no time pressure, no assessment. That is what lowers the barrier for people who have never handed knowledge to software.

The recording stays canonical

The raw recording is never changed. Every structured artefact is derived from it and regenerated as transcription and models improve. Your knowledge base gets better retroactively.

Dialect, jargon, and nicknames

The system learns the language of your plant: what the machines are really called, which abbreviations apply, and what a term means specifically here.

Knowledge that connects

Each recording is linked to the existing body of knowledge. Contradictions and gaps become visible, so you can follow up while the holder is still there.

Your knowledge stays yours

Recordings and knowledge base belong to you and export in open formats. Hosted in Germany or on premise on your own infrastructure.

The knowledge store

A body of knowledge that improves as the technology does

Classic documentation starts ageing the day it is written. A store of original recordings gets more valuable with every better model.

1

Raw store

Every recording is kept unchanged. Nothing is lost, not even the side remark whose meaning nobody understands yet.

2

Derived artefacts

Transcript, summary, work instruction, glossary, and knowledge graph are generated from the raw store and fully recomputed when needed.

3

Knowledge on demand

The successor asks in their own words and gets an answer with its source. The same base later feeds onboarding, maintenance, and audits.

Workable in practice

Built for the realities of a European plant

Voluntary and co-determination ready

Participation is voluntary. Every employee sees their own recordings and can delete them. Purpose, scope, and retention can be agreed cleanly with the works council.

On premise or cloud in Germany

Recordings are personal voice data. They stay where you decide: in our cloud in Germany or on your own infrastructure.

GDPR, roles, and audit trail

Encryption in transit and at rest, role based access control, and a complete audit trail over every access to a recording.

Start with three people

No rollout and no change project. You start with the people whose knowledge is closest to leaving.

Frequently asked questions

What is Narrator?

Narrator is a knowledge retention solution. Experienced employees talk freely into voice messages, with no structure and no forms. AI agents transcribe and organise the recordings in the background and build a knowledge base that successors can query later.

How does knowledge transfer before retirement work with Narrator?

Instead of a handover phase full of documentation tasks, the departing employee talks over many months whenever something comes to mind. Every recording is structured and linked to the previous ones. The successor later works with a knowledge base instead of a folder.

Why does the system not reply?

Because a reply forces a conversation and a conversation creates expectations. People who do not want a software project talk more freely when nobody asks, corrects, or judges. Structuring happens in the background, not in a dialogue.

How is works council co-determination handled?

Voice recordings of employees are personal data and usually subject to co-determination. Narrator is designed for that: voluntary participation, access and deletion rights for every speaker, a clearly bounded purpose, and no performance or behaviour monitoring. These points belong in a works agreement, and we support the preparation.

What happens to the recordings after the employee has left?

You define retention and deletion periods, usually in the works agreement. Technically the raw recording remains the canonical store for as long as it is kept. All derived artefacts can be regenerated from it at any time.

How is Narrator different from a wiki or a document database?

A wiki requires somebody to write, structure, and maintain. That is exactly where knowledge projects fail. Narrator only requires talking. Structure is produced by agents and recomputed as the technology improves, instead of ageing from the day it was written.

How does Narrator relate to Fieldcast?

Both rest on the same idea: people report or talk through the messenger they already use, and AI agents turn it into something usable. Fieldcast produces actions from daily operations. Narrator produces a body of knowledge from experience. Both run on the same platform.

Who retires next at your plant?

See in 30 minutes how free talking becomes a knowledge base your successor can query.

Book a demo