PROCESSES LIVING IN EXCEL AND EMAIL
Automate the processes that are stealing hours from you every week
Invoices, reports and data carried by hand between systems cost a company hours every week. We automate them with n8n and our own code — starting with an audit that shows what is worth automating and what is not.
A process audit with an ROI estimate
Lost / week
20–40 h
what a typical company of 30–300 people loses on operations that can be automated
WHEN IT'S WORTH IT - 4 SIGNALS
When automation makes sense - and when it's better to stay with manual operation
Signal 01
Repetitive manual tasks steal 10+ hours a week
Invoicing, copying data between the CRM and the warehouse and manual reports take the team several days a month.
PLN 38ka year saved on a single process
Signal 02
Human errors cost concrete money
Invoice mistakes, missed deadlines and data that does not match between systems cost money. Automation checks the data on the way in.
up to PLN 10kless in penalties per avoided error
Signal 03
Scaling up operations requires 2× people at 2× volume
The company grows and the back office grows with it, because someone handles every order by hand. Automation lets the same team handle more.
up to PLN 300ka year: 1–3 fewer positions
Signal 04
Integrations between systems by people, not by APIs
Systems in the company do not talk to each other, so people carry the data and reports run late. An API integration keeps it current.
6–12 monthstypical payback time
// 4 TYPES OF AUTOMATION - CHOOSE YOUR CONTEXT
What specifically are we automating
Back office processes
Invoices, reports, billing - back office
Generate invoices from sales system to accounting, recurring reports (daily sales, monthly management), import bank statements, payment matching, deadline reminders.
Typical integrations: Subiekt GT/Nexo, Comarch ERP, Optima, Fakturownia, iFirma, bank APIs (mBank, ING, Santander).
CRM integrations ↔ other systems
CRM ↔ marketing ↔ sales ↔ customer service
Synchronization of customer data between CRM (HubSpot, Pipedrive, Salesforce) and marketing tools (Mailchimp, Brevo, Mailgun), helpdesk (Intercom, Freshdesk), telephony (CallPage, Aircall). Sales funnels, lead scoring, automatic notifications.
Typical integrations: HubSpot, Pipedrive, Salesforce, Mailchimp, Brevo, Slack, Discord, Linear, Jira.
Warehouse and logistics processes
Warehouse, orders, couriers
An order from the shop goes straight to the warehouse, the courier label is created without retyping the address, and the customer gets the tracking number. Stock levels update after every shipment.
Typical integrations: online shop, Allegro, BaseLinker, courier companies.
Reporting and analytics
Management reports, analytical panels, alerts
Consolidation of data from several sources (sales, marketing, operations, finance) into a single management panel. Alerts on deviations (sales decreased by 20%, customer acquisition cost increased by 30%). Cyclical reports for management.
Typical integrations: Google Analytics 4, Meta Ads, Google Ads, Stripe, the Statistics Poland (GUS) API, custom tools, Metabase, Looker Studio.
CASE IN POINT
Tender radar: 71 notices, 6 worth reading
Every hour a bot pulls new notices from three public tender registers, AI scores each one against the company profile, and only the relevant ones reach the team — with no one involved. In one run, 6 of the 71 notices fetched were left.
MAKE / ZAPIER / DEDICATED CODE - DECISION
When Make/Zapier is enough, and when you need dedicated code
Make and Zapier are great tools - for 60-70% of simple automations they are faster and cheaper than dedicated code. Our rule: we don't sell dedicated code when an off-the-shelf tool will do.
However, the boundary is sharp. When the process requires custom logic, large scale or data control - Make/Zapier starts to hurt with cost, limitations or failure rate. Then dedicated code (n8n self hosted + Node.js) is cheaper in 12-24 months.
Decision table
Criterion | Make / Zapier | Dedicated code (dvnt) |
|---|---|---|
Number of operations per month | <10 000 | >10,000 (Make cost increases linearly) |
Complexity of the process | Linear (A → B → C) | Branches, loops, conditions, custom logic |
Maximum duration of one run | Make: 40 minutes per scenario run on paid plans (5 minutes on Free) | We set the limit ourselves in the server configuration (n8n imposes none by default) |
Data sensitivity | Public or low-sensitivity data | Financial data, medical data, GDPR (data in the EU, control over infrastructure) |
Monthly cost at scale | PLN 50-500 (small volumes) | 50-200 USD hosting + maintenance |
Cost at 10× volume | PLN 500-5,000 (scales linearly) | USD 50-200 (scales economically) |
Code access/ownership | None - Make has control | Your repository (our code and workflows); n8n under the Sustainable Use License |
Time to implement the first automation | 1-3 days | 2-4 weeks (audit + design + implementation) |
Modification time after process change | 1-2 hours | 4-16 hours (with testing and implementation) |
Reliability when an external API fails | Retries and notifications within the platform, depending on settings and plan (Make: automatic retry of incomplete runs; Zapier: error notifications on paid plans) | Retry with exponential delay, alerts, change log |
dvnt recommendation
Make/Work:
- Single automation: email → CRM, form → sheet, notification
- Small volume <5,000 operations/month.
- Non-technical employee to maintain configuration
Dedicated code (n8n self hosted + Node.js):
- System of 5+ related processes
- Volume >10,000 operations/month.
- Sensitive data (GDPR, financial data).
- Custom business logic
- Integration with ERP and accounting systems that Make/Zapier does not support (local or legacy ones)
Some automation conversations end with a recommendation: “start with Make and come back to us in 6–12 months, when you outgrow it”. We lose a lead; the client comes back as a regular.
// HOW WE IMPLEMENT - 3 STAGES
From process audit to production
Stage 01
Process audit with ROI valuation
We list the processes done by hand, estimate their cost and decide what to automate first, what in Make or Zapier, and what not at all.
Stage 02
Implementation after priorities
We roll out the automations one by one, starting with the one that pays back fastest. Each comes with monitoring and documentation.
Stage 03
Monitoring and iteration (on a continuous basis)
We keep the automations running and adjust them when systems or processes change. Maintenance can be handed over to your team.
Total time from first call to working first automation: 4-6 weeks.
Technologies: n8n (5+ years of production experience, active support, source code publicly available under a fair-code license) + Node.js (10+ years) + PostgreSQL (change log, process status). Zero version 0.x tools or no proven track record.
Let's start with a process audit
FAQ
Questions we get before we start automation
How do we calculate the return on investment in automation?
Simple formula: monthly process time × employee rate × 12 months = annual savings.
Example: a manual invoice takes 5 min × 200 invoices/month. = ~17 h/month. × PLN 80/h × 12 = PLN 16,320 per year.
The automation valuation of PLN 15,000-30,000 pays for itself in 11-22 months. After 3 years: PLN 32,640 net savings.
The process audit (Stage 1) counts this formula for all processes in the company and shows which to automate first. Sometimes it turns out that process #5 on the list saves more than processes 1-4 combined.
n8n vs. Make vs. Zapier vs. dedicated code - which to choose?
The decision depends on the volume, complexity and sensitivity of the data. Bottom line:
- Zapier - The simplest, broadest catalog (more than 5,000 integrations), expensive at scale
- Make (formerly Integromat) - More advanced than Zapier, cheaper, complexity limitations
- n8n - Source code publicly available (fair-code license), can be hosted independently (data in EU, control over infrastructure), requires technical person
- Dedicated code (n8n + Node.js) - unlimited, cheapest on the scale, highest entry threshold
The full decision table is above on this page. The process audit (Stage 1) recommends a split: which processes go to Make/Zapier and which to dedicated code.
What if the systems you automate change?
Each automation has 2-3 points of contact with external systems (data source, destination, optional notification). When one system is changed, the modification usually takes 4-16 working hours.
Three scenarios:
- Minor API change (new field, format change) - 2-4 h, cost PLN 200-800
- Migration to a new system (e.g., from Subiekt to Comarch) - 8-40 hours depending on complexity, cost PLN 1,600-8,000.
- Full process re-architecture - Individual pricing, usually 20-50% of the cost of the original automation
In a permanent care contract (PLN 450-1,500/month), minor modifications are included in the price. Larger modifications billed on an ad hoc basis.
Who maintains the automation after implementation?
Three options (the same as in other dvnt projects):
- Own team - We deploy your developer in ~1 week, documentation allows you to take over the project. Best if you have an IT department or an experienced administrator.
- Ongoing dvnt care - PLN 450-1,500/month depending on the number of processes and SLA. Includes monitoring, minor modifications, disaster recovery. Best for companies without an in-house IT department.
- Hybrid - dvnt maintains critical processes, your team manages simpler ones. Typically for companies of 100+ people with partial IT skills.
Is automation replacing employees?
The short answer: in 95% of cases, no. Automation frees employees from monotonous tasks to do more valuable work.
Specifically: an accounting employee who used to spend 2-3 days a month manually issuing invoices, after automation has time for financial control, contacting customers about backlogs, optimizing processes. Same position, more value for the company.
Exception: in some companies, automation allows to handle 2× volume with the same team instead of hiring new people. This does not"deceleration", only"non-employment" additional people. The decision is made by the customer, not us.
What does regulatory compliance (GDPR, accounting audit) look like?
The automation we host ourselves gives you full control over your data: our own infrastructure in the EU, GDPR compliance, ability to audit code and processes.
For compliance we build in:
- Change log - Each operation saved with timestamp, user, before/after data. Export to PDF for the auditor.
- Data anonymization - when the process does not require full personal data, we use hashed identifiers.
- Data retention - Automatic cleaning of old records after GDPR retention deadlines.
- Encryption at rest and in transit - TLS, AES-256 for sensitive fields.
RESOURCES · BEFORE WE TALK
Before we talk
ARTICLE
Business process automation — examples and where to start
Business process automation: how it differs from RPA and AI, Polish data, KSeF e-invoicing, our hands-off funnel, examples by department and the first step.
ARTICLE
What is n8n — and when does it make sense for a business
What is n8n? An automation tool with AI nodes, in the cloud or on your own server.
ARTICLE
Low code and no code — what they are and when they replace programming
Low code and no code explained: who a citizen developer is, what a low code platform suits, its price limits and what you can take with you when you leave.
// Next step
Describe your project. We reply within 24 h.
A few sentences are enough: what should work, what it connects to and by when. No brief, no commitment.
What happens next
- We read your message and reply within one working day.
- We set up a short call about the scope, with no sales pitch.
- You get proposed stages and a price range.