Done-for-you AI automation

AI workflows built around your business— and managed for you.

We map repetitive work, connect the tools you already use, and deploy reliable AI automations with human approval where it matters.

  • Client-owned workflows
  • Clear fixed scope
  • Human approval controls
  • No vendor lock-in
Enquiry → booked meetingExample workflow
  1. 1New enquiry arrivesWeb form or shared inboxtrigger
  2. 2Classify the requestAI step: type, urgency, languageAI
  3. 3Enrich company dataDomain, size, existing CRM recordlookup
  4. 4Update the CRMNo duplicates, owner assignedwrite
  5. 5Human approvalReply draft reviewed by your teamapproval
  6. 6Book the meetingCalendar link matched to owneraction
  7. 7Send summaryTo the account owner, loggednotify

Demonstration using sample data. No live customer data is processed on this page.

Fixed scope

Every package states what is included, how long it takes and what changes the price. Written acceptance criteria before we build.

Client ownership

Production accounts, workflows, prompts and documentation belong to you. If we part ways, everything keeps running.

Human controls

Approval gates on sensitive steps, confidence thresholds, exception routing and a kill switch are designed in from day one.

No vendor lock-in

n8n, Make, Zapier, Power Automate or plain code—chosen per use case, documented, exportable.

Where the hours go

The hidden workload between your tools

Most repetitive work is not inside one application. It sits in the handoffs: the copying, checking, chasing and re-typing that no single tool owns.

  • Copying form and email data into a CRM
  • Re-typing invoice fields into accounting software
  • Sorting a shared inbox and forwarding to the right person
  • Chasing approvals that sit in someone’s inbox for days
  • Preparing the same weekly report from three spreadsheets
  • Sending the same follow-up for the fortieth time

First we map, then we automate

Automating a broken process only makes the problem faster. The first thing we do is observe how the work actually happens, capture a baseline (volume, hours, errors, response time) and decide together which steps deserve a machine, which deserve a human and which should simply be removed.

This is why every engagement starts with discovery, and why we publish what one “automation” means before you buy one.

Outcomes

What changes when the handoffs are automated

Faster response

Enquiries, tickets and documents are classified and routed within minutes instead of waiting for someone to get to the inbox.

Fewer manual errors

Validated fields, no duplicate records, and exceptions routed to a person instead of silently going wrong.

Recovered staff capacity

Skilled people spend their time on judgement, relationships and exceptions—not on copying data between systems.

Growth without matching overhead

Handle more volume with the team you have, and hire for the roles that actually need people.

What can be automated

Five places we usually start

Each card shows a real input, the automated action and the output. Every one of these can begin as a single pilot workflow.

Lead and CRM automation

Speed-to-lead, clean records, follow-ups that actually happen.

InputForm submission, inbound email, LinkedIn lead
ActionClassify, enrich, de-duplicate, assign owner, draft reply
OutputCRM record, booked call, owner summary

Customer support automation

Triage, drafting and routing with a human on every uncertain case.

InputTicket, email, order-status question
ActionClassify intent, look up order, draft answer, route
OutputApproved reply, tagged ticket, escalation when unsure

Document and finance admin

Invoices, receipts and contracts captured, coded and approved.

InputPDF invoice in a mailbox or folder
ActionExtract fields, match supplier, suggest coding, request approval
OutputBooking-ready record with audit trail

Reporting automation

The weekly report that builds itself, with a short written summary.

InputCRM, accounting, ads and support data
ActionCollect, reconcile, summarise changes, flag anomalies
OutputDashboard refresh and Monday-morning brief

Knowledge assistants

Answers from your own documentation, with sources and permissions.

InputQuestion in Slack, Teams or a web widget
ActionRetrieve relevant documents, answer with citations, log
OutputSourced answer or hand-off to a person

Not sure where to start?

Run the free Opportunity Scan. Nine questions, an honest fit score, and a package suggestion—no sales call required.

How it works

Five steps, one owner, no surprises

  1. Discover

    We observe the real process, capture the baseline (volume, hours, errors, response time) and agree on the one workflow to start with.

  2. Design

    Inputs, outputs, approval points, fallback paths and written acceptance criteria. You know exactly what “done” means.

  3. Build and test

    Built in client-owned accounts on the simplest reliable platform. Tested on normal, edge, failure and privacy scenarios.

  4. Launch and train

    Gradual rollout with monitoring, cost caps and a kill switch. Recorded handover, runbook and an ownership map.

  5. Monitor and improve

    30/60/90-day comparison against the baseline. Expand only after the first workflow has proven useful.

Read the full delivery method →

Packages

Clear prices, from one workflow to enterprise systems

Implementation is a one-time fee. Ongoing care is optional and priced separately. Software and AI usage stay in your accounts.

One Workflow Pilot

€1,450

One clear process you want to test before anything bigger.

  • Delivery: 10–15 business days
  • 14 days post-launch support
  • 2 credits

Solo

€3,400

Solo operators and very small companies with one department to fix.

  • Delivery: 2–3 weeks
  • 30 days post-launch support
  • 5 credits

Business

From €27,900

30–200 employees or multi-department scope with custom integrations.

  • Delivery: 6–12 weeks
  • 60 days post-launch support
  • 30 credits

Enterprise

Discovery from €5,000Implementation from €40,000

Complex, regulated or large organisations that need governance first.

  • Delivery: 10–24+ weeks
  • Defined in the SLA

All prices exclude VAT. Third-party software, AI model usage, telephony and hosting are billed separately.

Compare every package, credit definitions and care plans →

Do the maths

Estimate the payback before you talk to anyone

Every assumption is editable and the formula is shown. It is a planning estimate, not a promise.

Automation ROI calculator

Change every assumption. The formula is shown below the results.

Salary plus employer costs, divided by productive hours.
Conservative default. Rarely above 70% for processes with exceptions.

Estimated payback period
7.7 months
Hours potentially recovered per month
46.8 h
Monthly gross time value
€1,638
Monthly net estimated benefit
€1,538
Estimated net value over the horizon (12 months)
€6,556
Estimated ROI over the horizon
55 %
Formula
hours/month = people × hours per person per week × 52/12 × automatable %
gross value = hours/month × loaded hourly cost
net benefit = gross value + error/rework cost − monthly software & care
payback     = implementation fee ÷ net benefit
net value   = net benefit × horizon − implementation fee
ROI         = net value ÷ implementation fee

This is an editable planning estimate, not a guarantee. Actual results depend on process quality, adoption, volume, exceptions and implementation scope.

Check if the process is a good candidate Discuss this estimate

Security and governance

Built so you stay in control

We design privacy, access controls, human oversight and documentation into each workflow, then identify where specialist legal or security review is required.

  • Client-owned production accounts
  • Least-privilege access, separate dev and production credentials
  • Human approval for sensitive actions
  • Run logs, error alerts and cost monitoring
  • Fallback path and kill switch
  • Documented handover, backups and export

How we handle data, access and the EU AI Act →

Evidence

Why the gap is implementation, not interest

Three numbers from public research. They describe the market, not our results.

9 in 10

of surveyed organisations use AI in at least one business function — but only 44% report enterprise-wide scaling.

Source: McKinsey, The State of AI (as of 2026)

70.9%

of EU enterprises that considered but did not adopt AI cited a lack of relevant expertise.

Source: Eurostat, Use of AI in enterprises (as of 2025)

65%

of surveyed SME users said generative AI improved employee performance; effects on workload were far more common than job cuts.

Source: OECD, Generative AI and the SME Workforce (as of 2025/2026)

Example workflows

Three demonstrations, step by step

Each example shows the trigger, the automated steps, the human checkpoint and what we would measure. They are demonstrations with sample data, not client case studies.

Demonstration, not a client case study

Enquiry to CRM record and booked appointment

A new enquiry becomes a clean HubSpot record, an approved reply and a booked meeting, with the account owner approving before anything is sent.

TriggerNew website form submission or inbound email in the sales inbox
HumanThe account owner sees the classified enquiry, the enriched record and the draft reply…
Demonstration, not a client case study

Invoice and document intake with human approval

Supplier invoices from a shared mailbox are extracted, matched and coded, approved in Teams where needed and passed to the accounting system with an audit trail.

TriggerNew email with a PDF attachment in the accounts-payable mailbox
HumanThe approver receives a Teams card with the extracted fields, the suggested coding, the…
Demonstration, not a client case study

Support ticket classification and draft response

Tickets are classified and prioritised, order data is looked up in Shopify, a sourced draft reply is prepared and an agent approves or escalates.

TriggerNew ticket in Zendesk or Freshdesk
HumanThe agent sees the classification, the matched order, the draft and the articles it…

All example workflows →

FAQ

Questions people ask before they buy

What exactly counts as one automation?

We price in credits, not vague “automations”. A simple workflow (one trigger, up to three routine actions, one or two apps) is one credit. A standard workflow with branching, an approval step and one controlled AI step is two. Advanced workflows with custom APIs, knowledge retrieval, sensitive data or high volume are four to eight or more. The final classification is confirmed in discovery, before you pay for a build.

How much does AI automation cost?

Our fixed packages start at €1,250 for a single pilot workflow and €6,900 for a team-scale build with twelve credits. Larger multi-department projects start at €16,500; enterprise work begins with a paid discovery. Prices exclude VAT and third-party usage. See the pricing page for exactly what is included and what changes the price.

Which tools can you connect?

Anything with an API or an established connector: Microsoft 365, Google Workspace, HubSpot, Pipedrive, Salesforce, Zendesk, Shopify, Stripe, lexoffice, sevDesk, DATEV interfaces, Slack, Teams, Notion, Airtable and hundreds more via n8n, Make, Zapier or Power Automate. Desktop-only or legacy software needs discovery first.

Who owns the workflows?

You do. Wherever practical we build inside accounts you own and pay for, hand over full documentation and exports, and never hold your credentials as the only copy. If you stop working with us, everything keeps running.

What happens when the AI is uncertain or wrong?

AI steps run behind confidence thresholds. Uncertain or sensitive cases are routed to a person for approval instead of being executed automatically. Every AI output is logged so you can review, correct and improve it. There is always a fallback path and a kill switch.

Is software or API usage included?

No. Automation platform fees, AI model usage, telephony and hosting are paid by you directly in your own accounts, so you can always see what it costs. We estimate running costs in the proposal and set cost caps during launch.

How long does implementation take?

A single pilot workflow takes 10–15 business days. Solo packages take two to three weeks, Team packages three to six, Business projects six to twelve. Enterprise rollouts are planned in stages and usually run ten weeks or longer.

Can this work with sensitive data?

Often, with the right design: data minimisation, least-privilege access, EU-hosted or private model options, approval gates and an explicit review of the data-protection and AI Act implications for that use case. Sensitive workflows are scoped through the paid audit rather than a fixed package, and we tell you when specialist legal review is needed.

Will this replace employees?

Our goal is to remove repetitive work, not people. Public research shows AI reduces workload far more often than headcount. In practice, teams reinvest the recovered hours in customers, quality and growth, and grow without hiring for every extra unit of volume. We do not build workflows that make autonomous hiring, firing or performance decisions.

What maintenance is required?

APIs change, models get updated, and businesses change their processes. Every package includes a post-launch support window. After that, optional care plans cover monitoring, alerts, regression testing after model updates and a monthly allowance of changes—from €290 per month.

Do you work internationally?

We work in German and English and currently focus on businesses in Germany, Austria, Switzerland and the United Kingdom. Remote delivery works anywhere; for other countries we check data-protection and contracting requirements first.

What is the difference between an AI agent and a workflow?

A workflow follows defined steps and rules; it is predictable and easy to test. An agent decides its own steps to reach a goal, which is flexible but harder to control. Most business processes are best served by a deterministic workflow with a few well-bounded AI steps. We use agents only where they are clearly better—and always with limits and oversight.

Next step

Start with one workflow worth fixing

Answer nine short questions and get an honest read on whether your process is a good automation candidate — and which package fits. No sales call required.