enaudo
AI automation with human approval
AI can extract documents, classify enquiries and prepare text. We connect these tasks to your workflows and professional review. Models, processing arrangements and approvals are defined for the project.
The principle: machine drafts, humans decide
For each AI task, we define the input data, review method and point where processing stops. Plausible text or a value is not yet a checked result. Detected uncertainty receives a review note; other errors may remain even in an apparently clear output.
- AI prepares: text, classification, summaries, categories
- Rules decide the clear cases: duplicates, thresholds, required fields
- People approve anything that goes outside: send, booking, payment, commitment, activation, access removal
- Uncertain fields are flagged: unchecked status plus a concrete note on what to check
-
Workflow scenarios
Connected flows with step sequences: enquiry to invoice, documents and HR.
-
Tools and interfaces
Possible connections between email, CRM, accounting and internal apps.
-
n8n as the connecting layer
Triggers, failure paths and approvals live in n8n. AI is one layer inside, not the whole flow.
Agree on models, data and operating locations
We assess the model, provider and processing systems. EU processing and open-source models may meet project requirements; the actual deployment, involved services and contractual terms matter. Data scope, operating and backup locations, access, retention and any required processing agreement are clarified before use.
- Assess EU processing and open-source options against project requirements
- Define model providers, involved services and processing locations
- Credentials live in the instance, not in chats
- Access by person, revocation after the project
- Clarify data use and possible training use by the selected services before deployment
Four generic examples of AI in a workflow
These generic scenarios illustrate possible uses, not approved client references. Each pattern connects preparation with professional review before binding steps.
-
Capture
Audio enters the flow, split into chunks automatically.
-
Transcribe
Speech recognition writes the conversation with speaker separation.
-
Summarise
AI drafts an internal and a neutral external version.
-
Approve
ApprovalA person checks and approves before anything is sent.
-
File
The reviewed version and its approval status are assigned to the case.
-
Note arrives
Message or file from day-to-day work.
-
AI reads along
The model pulls out people, companies and requests.
-
Check existing records
Matched against contacts and companies on file.
-
Create or enrich
ApprovalNew only what is missing. Otherwise enriched.
-
CRM is current
Contact, company and history sit in one place.
-
Document arrives
Identify identical files by checksum. Check other possible duplicates using suitable invoice details.
-
Read
Text recognition reads amounts, dates and numbers.
-
Classify
AI proposes a category and tax. Detected uncertainty is flagged; professional review is still needed.
-
Review and approve
ApprovalDraft in accounting, a person approves.
-
Book
Only after approval does the document become binding.
-
Occasion is set
Onboarding or a ticket supplies the context.
-
Draft
AI writes the text from the case, with the open points.
-
Approval
ApprovalA person checks tone and content.
-
Send
Only then does the message go out.
Where AI deliberately does not decide
Some steps stay manual with system support. Binding quotes, payments, cancellations and anything hard to undo need a person. AI delivers the draft; responsibility stays with you.
- No automatic send to customers without approval
- No booking or payment without review
- No commitments on dates or prices without a person
- No access removal without approval
- Every case stays traceable with time and status
Frequently asked questions
Which AI models do you use?
The choice depends on the task, data and operating requirements. We assess suitable models and providers, including EU processing and open-source options. The actual selection, processing locations and involved services are agreed for the project.
How is training use of our data handled?
Permitted data use is defined before deployment. We check the terms and settings of selected providers, possible training use, retention and deletion. Confidential data is provided only through the agreed project method.
What happens when the AI is unsure?
Detected uncertainties receive a review note, for example about the invoice number or due date. Plausible information that has not been flagged may also be wrong and requires the agreed professional review.
Can the AI send without approval here?
Only where the risk allows it and you agreed in writing, for example purely internal notes. Anything that goes outside or moves money stays tied to approval.
Where do I see what the AI did?
Every case gets a log entry with time and status. Drafts, approvals and sends stay verifiable.
Sources and documentation: GDPR, Regulation (EU) 2016/679AI Act, Regulation (EU) 2024/1689
If a process is eating time in your team, describe it. We will say honestly whether automation is worth it.