Skip to content
DEIntro call
← All services

We automate repetitive office processes with AI

We develop custom AI and automation solutions for businesses: we analyse existing workflows, connect the systems you already have and automate time-consuming processes — without companies having to replace their entire IT landscape.

Our goal: less manual work, faster processes and systems that work reliably for your business in the background.

Free intro callSee the offers

Quote process

Example process

  1. Email
    Form
    Document
  2. AI extractsCase recognised, fields filled
  3. CRM matchedCustomer recognised
    Quote createdfrom template
  4. Approvalby a team member
  5. Sending
    Follow-up

See this process in detail: three automations step by step.

AI workshops & training

Where can AI help your business?

We find practical use cases with your team, set priorities and practise using AI in everyday work.

Explore workshops

What we automate

Six areas where recurring work can be taken off your hands most reliably. Each can be implemented on its own — a project usually starts with exactly one of them.

01

AI email and enquiry handling

Automatically recognise, classify and prepare incoming emails and customer enquiries.

Email · one case

An email becomes a reviewable reply draft

Example: A customer asks for a maintenance appointment and attaches a photo.

  1. 01 / Input
    Receive enquiry
    Inbox · email + photoSubject: maintenance requestPreference: next week
  2. 02 / Classification
    Classify request
    Topic
    Maintenance
    Attachment
    1 photo
    Customer
    Check in CRM
  3. 03 / Review
    Review reply
    • Identify responsible team
    • Check available dates

    Missing detail → team

  4. 04 / Outcome
    Present draft
    • Reply with question or date proposal
    • Approval before sending

Status and next step remain visible in the case.

Service details
  • Recognise customer enquiries automatically
  • Extract relevant information from emails and attachments
  • Categorise enquiries by topic or priority
  • Complete CRM data automatically
  • Create draft replies
  • Involve team members for approval
  • Trigger follow-ups automatically
02

Lead and sales automation

Processes from first contact to sales, without enquiries getting stuck.

Sales · one contact

A website enquiry gets a clear next step

Example: A business asks about service for three locations via a form.

  1. 01 / Input
    Receive contact
    Website · enquiry formService: ongoing supportScope: 3 locations
  2. 02 / Classification
    Capture lead
    Need
    Support
    Scope
    3 locations
    CRM
    Match contact
  3. 03 / Review
    Qualify
    • Check company details
    • Identify contact owner

    Unclear → sales team

  4. 04 / Outcome
    Prepare outreach
    • Personal message draft
    • CRM reminder

Appointments and follow-ups follow the agreed rules.

Service details
  • Website enquiries go straight into the CRM
  • Research and qualification in the background
  • Personalised messages instead of boilerplate text
  • Follow-up reminders that are not forgotten
03

Quote automation

Especially interesting for service providers and B2B companies with recurring quotes.

Quote · one draft

From enquiry to approved quote

Example: A service provider receives a service request with a preferred date.

  1. 01 / Input
    Receive request
    Inbox · emailService: on-site visitDate: next week
  2. 02 / Classification
    Prepare details
    Customer
    CRM match
    Service
    On-site visit
    Date
    next week
  3. 03 / Review
    Review quote
    • Check line items and price
    • Team member confirms

    No approval → no send

  4. 04 / Outcome
    Send & follow up
    • Send approved quote
    • Schedule follow-up

The actual price comes from agreed rules and systems.

Service details
  • Prepare quote data from the enquiry
  • Recognise existing customers, create new ones
  • Create a quote or a draft quote
  • Approval by a team member before sending
  • Automatic follow-up after sending
04

Document automation

Read and classify PDFs, invoices and receipts and get them where they belong.

Document · one invoice

A PDF is captured and handed over with context

Example: An email contains an invoice PDF.

  1. 01 / Input
    Receive document
    Inbox · PDFInvoice R-1042Keep the original
  2. 02 / Classification
    Capture fields
    Sender
    Werkstatt Nord
    Amount
    €184.80
    Reference
    R-1042
  3. 03 / Review
    Review details
    • Required fields present?
    • Destination identified?

    Missing → team

  4. 04 / Outcome
    Hand over
    • File document
    • Prepare ERP / CRM data

No automatic posting without agreed checks.

Service details
  • Read PDFs and recognise invoices
  • Categorise documents and extract information
  • Name files automatically and file them correctly
  • Transfer data into CRM, ERP or database systems
  • Detect missing information
  • Create summaries
05

Internal AI knowledge assistant

Make company knowledge searchable in one place — based on your own documents.

Knowledge · one question

An answer points back to its own source

Example: A team member asks about the complaints process.

  1. 01 / Input
    Ask question
    Internal chat · requestHow do complaints work?Approved documents only
  2. 02 / Classification
    Find passages
    Source
    Policy, version 2
    Section
    Complaints
    Match
    Step 3: review
  3. 03 / Review
    Check answer
    • Does it address the question?
    • Is the content current?

    No source → no reliable answer

  4. 04 / Outcome
    Show answer
    • Short summary
    • Source to read directly

The team member can check the cited document.

Service details
  • PDFs, Word documents and internal guidelines
  • Google Drive and Microsoft SharePoint
  • Web pages, FAQs and product information
  • Internal documentation

Employees ask, for example

  • “How does our complaints process work?”
  • “Which contract terms apply to this customer?”

The AI answers from existing company information and names the sources it relies on.

06

Automated reporting

Connect existing data sources and have them evaluated regularly.

Reporting · one finding

Connected data becomes a reviewable finding

Example: A team wants to spot changes in support tickets.

  1. 01 / Input
    Fetch data
    Ticket system · reportOpen tickets: 24Previous week: 18
  2. 02 / Classification
    Organise metrics
    Source
    Ticket system
    Metric
    Open tickets
    Compare
    18 → 24
  3. 03 / Review
    Review change
    • Complete data?
    • Explainable change?

    Unclear → team review

  4. 04 / Outcome
    Present finding
    • Finding with source
    • Suggested action to consider

A finding does not make a business decision automatically.

Service details
  • CRM, shop systems and ticket systems
  • Google Analytics and Google Search Console
  • Advertising platforms
  • Excel and databases

How we access your systems — and how we don’t

The questions that always come up in the intro call. That is why the answers are here and not just in the quote.

How the technical implementation works

For our automations we normally do not need access to employees’ computers. Existing systems are connected to each other via secure interfaces.

  • APIs and webhooks
  • OAuth and service accounts
  • Databases
  • n8n as the automation layer
  • OpenAI as well as other AI models and specialised services
  1. Microsoft 365
  2. n8n
  3. AI processing
  4. CRM
  5. ERP
  6. Staff approval
  7. Automated response

Employees can keep working with the applications they are used to.

Secure access to company systems

We work on the principle of only as much access as necessary. Companies do not have to share personal passwords.

  • API access
  • OAuth connections
  • Service accounts
  • Restricted user accounts
  • Read-only access
  • Webhooks
  • Database interfaces
  • SFTP

Only as much access as necessary

  • Service accountonly the “Enquiries” mailbox
  • CRM interfaceread and create cases
  • Storage folderone folder, write only

Every access is named, granted individually and can be revoked individually.

Not needed

  • Employees’ passwords
  • Access to workstations
  • Administrator rights
  • Access to everything “just in case”

Your employees keep working in the applications they are used to.

Access rights are limited as far as possible to exactly the functions the automation needs.

Development with test data

Wherever possible, automations are first developed with test data, anonymised data, sandbox systems and sample documents. Only afterwards are the real company systems connected.

This means large parts of a solution can be developed and tested before any production company data is processed.

Human in the loop

For sensitive or important processes, we rely on human approval first. Only once a process works reliably can suitable steps be automated further.

  1. AI creates quote
  2. Team member checks
  3. Team member confirms
  4. System sends quote

The goal is not to replace employees but to automate repetitive steps and leave important decisions with people.

When systems have no modern API

Older systems can often be integrated too. Which route fits, we clarify together in the process analysis.

  • API connection
  • Webhooks
  • Email processing
  • CSV or Excel imports
  • XML
  • SFTP
  • Database connection
  • RPA — browser or desktop automation

We use RPA only when no more stable interface exists: it depends on the other system’s user interface and breaks as soon as something changes there.

How our projects run

Seven steps from the first mapping to supported operation. After each step you can decide whether to continue.

  1. 01

    Process analysis

    Together we look at the existing workflow.

    • Triggers and input data
    • Decisions and systems involved
    • Manual work and outcome
  2. 02

    Automation concept

    You get an assessment of what is worthwhile — and what is not.

    • Steps that can be automated
    • Required interfaces
    • Possible risks
    • Estimate of time savings and commercial potential
  3. 03

    Development

    The solution is first built with test data wherever possible, not with your production data.

  4. 04

    System integration

    The required CRM, ERP, email, database or cloud systems are connected.

  5. 05

    Test phase

    The automation first runs in a controlled way with human approval.

  6. 06

    Live operation

    After a successful test phase, the solution is integrated into the regular business process.

  7. 07

    Support and optimisation

    If you wish, we take over day-to-day operation.

    • Hosting, monitoring and workflow maintenance
    • Troubleshooting and API adjustments
    • Prompt optimisation
    • Developing existing automations further and building new ones

Our offers

Three ways in — depending on whether you want to check first, implement once or have it run on an ongoing basis.

Automation Audit

We analyse an existing business process and assess which automations make sense technically and commercially.

Result

  • Mapping of the current process: trigger, input data, decisions, outcome
  • Overview of the steps that can be automated — and those that stay with people
  • Possible architecture of the solution
  • Required systems and interfaces, including routes for systems without a modern API
  • Estimate of time savings and commercial potential
  • Implementation proposal with effort and risks

from €890 — Fixed scope, fixed price, plus VAT.

EnquireSee the scope

AI Automation Sprint

We automate one clearly defined end-to-end process — from the enquiry to the follow-up.

Result

  • One well-defined process, fully implemented
  • Connection of the systems involved via APIs, webhooks, databases or file transfers
  • Development first with test data, anonymised data or sandbox systems
  • n8n as the automation layer, AI processing only where it adds something
  • A test phase in which every case is approved by a member of staff
  • Handover with documentation of the process and the access that was set up

from €3,900 — Starting price, exact amount after the analysis, plus VAT.

EnquireSee the scope

AI Operations Support

After implementation, we can take over the operation and further development of your automations.

Result

  • Hosting of the automation layer
  • Monitoring that reports when a process stalls
  • Fixing errors during operation
  • Adjustments when an interface or access changes
  • Optimising prompts and process logic
  • A contact person for questions from your team

from €290 / month — Cancellable monthly, plus VAT.

EnquireSee the scope

As a ready-made product

AI Employee — the digital back-office employee

Instead of a custom project: a system that is already running, reads customer mail, decides and hands over what it cannot do safely — configured rather than programmed.

See the AI Employee

AI Employee

Customer mail → checks → decision

One case. Every step traceable.

  1. 01Input

    Customer mail

    Request with a documentEmail · document

    Create a caseCustomer + order reference

  2. 02Classification

    Read & understand

    Request
    Credit note
    Customer
    Rebenhof
    Amount
    from document

    Identify the relevant information

  3. 03Cross-check

    System context

    • Customer details
    • Order details
    • Previous activity

    Cross-check existing information

  4. 04Policy

    Your rules

    • Amount limit
    • Responsibility
    • Confidence

    Can the case proceed?

  5. 05Outcome

    Case A · within policy

    Complete the case€184.80
    or

    Case B · approval needed

    Involve your team€1,840.00
Continuous audit trail
  1. Source
  2. Checks
  3. Decision
  4. Outcome
With reasons and responsibility
Explore the decision

Two cases, two routes

The amounts are examples. Your agreed rules determine what can be completed independently and what your team reviews.

Within your rules

The data and system context match. The permitted action can be prepared and carried out.

Outside your rules or unclear

The case is handed over with context and reasons. Your team decides the next step.

Schematic workflow. System access and permitted actions are configured for your business.

Typical sprints

  1. Customer enquiry
  2. CRM
  3. Quote
  4. Approval
  5. Sending
  6. Follow-up
  1. Email
  2. Document
  3. Data extraction
  4. Filing
  5. ERP / CRM
  6. Staff approval

The marked illustrations use fictional businesses and sample data. They show possible workflows, not real customer results.

Our approach

We do not automate for the sake of automation. The central question is:

Which recurring process is costing your business unnecessary time or money today?

If an automation offers no commercial benefit, we do not recommend it. Our aim is to change existing processes so that employees spend less time on repetitive work and can focus more on tasks that create value.

Frequently asked questions

Where should a business start?

With a frequent, clearly defined process that noticeably costs time today — typically sorting enquiries, preparing quotes or structuring documentation. A pilot like that can be tested within a few weeks and rolled back again. Large all-in-one solutions fail more often because nobody adopts them in everyday work.

Does this have to involve AI?

No. A considerable part of what eats up time is simple forwarding and double entry — a clean workflow without a language model is enough for that. AI is worthwhile where unstructured text or images have to be processed. We say so when the simpler solution is the better one.

Do you need access to our employees' computers?

Usually not. It is the systems that are connected, not the workstations — via API access, OAuth connections, service accounts, webhooks, database interfaces or SFTP. We do not need personal passwords for this, and permissions are limited as far as possible to exactly the functions the automation needs.

What if our software has no modern interface?

Older systems can often be connected too: via email processing, CSV or Excel imports, XML, SFTP or directly at the database. If none of these is available, RPA is an option — browser or desktop automation. We deliberately use it last, because it depends on the other system’s user interface and breaks as soon as something changes there.

What happens to our data?

We clarify that before the first connection: which data flows where, how long it is stored and who has access. Where personal or client data is involved, we choose services with a data processing agreement and document the processing. Development beforehand uses test data, anonymised data or sandbox systems wherever possible.

How do we measure whether it helps?

Before the start, we record how long the process takes today and how often it occurs. Afterwards the same figure is measured again. If a pilot makes no recognisable difference, it is stopped rather than expanded — that is explicitly a possible outcome.

Which process costs you the most time today?

30 minutes, free and without obligation — we look at a specific process and tell you openly whether automating it is worthwhile.

Book a call

More services

Process digitalisation & AI operations

Connect systems — and look after the automations that are running.

See more

Custom software & web apps

Tailor-made systems instead of off-the-shelf software that doesn’t fit.

See more

GEO & SEO — visibility in AI answers

Be found — on Google and in ChatGPT, Gemini and the like.

See more

Websites & digital infrastructure

Digital foundations that sell instead of templates.

See more