What does Logitelia do that I can't do myself with n8n or Make.com?

Three things: (1) reliable AI judgment steps — we integrate LLM nodes with operator review so the workflow does not silently hallucinate at decision points; (2) production hardening — error handling, retries, dead-letter queues, monitoring, audit logs; (3) maintenance — workflows break when APIs change and need ongoing fixes. We handle all three under a subscription, not a one-off project.

Do you handle EU data residency for these workflows?

Yes. n8n self-hosted runs on EU infrastructure (Hetzner Falkenstein or Helsinki, or AWS Frankfurt). Make.com offers EU data residency on its Enterprise plan; we work with clients to set that up. All workflow secrets stay in the client's secret manager.

Workflow automation is part of the Operations Team subscription, starting at €4,500/month. That covers initial setup, ongoing maintenance, AI judgment integration, monitoring, and operator review. Infrastructure costs (n8n hosting, Make.com subscription) are passed through at cost — typically €40-€400/month depending on volume.

Solutions · Workflow automation

We set up n8n and Make.com for B2B teams.

Configured, hardened, and maintained workflow automations on n8n (self-hosted in the EU) or Make.com — plus AI judgment steps where they make sense, plus a named operator who reviews every workflow before it ships. Part of the Operations Team subscription, from €4,500/month.

What we configure

We do not pitch automation as magic. It is documented integration work, run continuously, with the parts that need judgment handled either by an AI agent (with review) or by a human (with a queue). Here is what we actually build for clients:

n8n (self-hosted, EU)

n8n is open-source workflow automation. We deploy it on EU infrastructure (Hetzner Falkenstein/Helsinki or AWS Frankfurt), set up monitoring, secrets management, and version control on workflows, then build the workflows you need. Common stacks:

CRM sync — HubSpot, Salesforce, Pipedrive, Attio kept in sync with billing (Stripe, Chargebee), product analytics (PostHog, Mixpanel), and your warehouse.
Lead routing — webhook from forms, enrichment (Clearbit, Apollo, hunter.io), scoring, round-robin or territory-based assignment, Slack alert to the rep.
Order desk automation — Shopify/WooCommerce orders → ERP sync → 3PL handoff → customer comms. Exceptions flagged to a human queue.
Billing reconciliation — Stripe events → accounting system entries (Xero, QuickBooks) → variance flag if amounts don't match expectations.
Internal alerts — health checks across your stack, escalation rules, on-call rotations.
AI judgment steps — LLM nodes for classification, summarization, or routing, with operator review on the first 100 runs before unsupervised.

Make.com (managed cloud)

Make.com is a managed visual automation platform. We use it when the client wants no infrastructure to maintain and the workflow logic is mostly visual chaining across mainstream SaaS. Same patterns as above, configured inside Make.com scenarios rather than n8n workflows.

Faster initial setup than n8n (no infrastructure step).
Better for teams without developers who will need to edit workflows themselves later.
Cost scales with operations volume — typically €30-€400/month for B2B teams under 100,000 ops/month.
EU data residency available on Make.com Enterprise plan.

Zapier — we usually migrate off it

If you are already on Zapier and bills are scaling, we typically move you to n8n self-hosted (cheaper) or Make.com (similar UX, lower cost at volume). Honest take in AI agents vs Zapier and Make: Zapier is great for tiny scale or one-off prototyping; for any sustained B2B operation past 5,000 tasks/month, the math falls apart.

When AI agents replace a workflow step

Not every workflow step needs AI. Most of them are deterministic: "if X happens, do Y." n8n and Make.com handle that natively. The interesting question is where AI judgment adds real value vs adds risk.

Examples where AI agents do replace a workflow step (with operator review on the first runs):

Lead classification. Inbound form fills with vague company info — the agent reads the company website, classifies industry/stage/fit, and routes accordingly.
Customer support triage. Email comes in, agent classifies (billing/technical/general), pulls relevant account context, drafts a reply, queues for human review.
Document extraction. Invoices, contracts, expense receipts — agent extracts structured fields, validates against rules, queues exceptions.
Summarization for handoff. Sales call → CRM note + next steps + risk flags. Account research → one-page brief.

Examples where AI agents do not belong in the workflow:

Anything irreversible without human approval — refunds, payment authorization, contract execution.
Anything that requires legal or regulated judgment — compliance decisions, KYC sign-off, medical advice.
Pure deterministic logic that does not need judgment at all — leave it as a standard workflow step. AI here just adds latency and cost.

What you get from a subscription, not a project

The most common mistake teams make with n8n or Make.com is treating setup as a one-time project. APIs change. SaaS tools deprecate endpoints. Workflows break silently and you find out in the worst possible way — a customer reports it.

Subscription delivery covers the parts that one-off projects miss:

Continuous monitoring. Workflow failure alerts. Health checks. Dead-letter queues with operator-assigned ownership.
Maintenance. When HubSpot deprecates an API or Stripe changes webhook signatures, we update the workflows before anything breaks.
Iteration. Workflows that worked in month one rarely match the business in month six. We adapt as your processes evolve.
Documentation. Every workflow has a written description of what it does, what it depends on, and how to debug it. Your team can take over any time without us being a bus factor.
Operator review on AI steps. Periodic audit of LLM-driven decisions to catch drift or hallucination patterns.

How it integrates with the rest of Logitelia

Workflow automation is one capability of the Operations Team, but it usually does not run alone. The interesting setups are cross-team:

Ops + Growth. Lead routing workflows feed into lifecycle email automation (see our lifecycle setup pattern).
Ops + Finance. Billing reconciliation workflows feed into monthly close (monthly close in 3 days).
Ops + Dev. When workflows outgrow no-code, the Dev Team builds custom services that n8n calls into. We do not lock you in.

Pricing

Workflow automation is part of the Operations Team subscription, not billed separately. Tiers:

€4,500/month (Starter). One team, narrow scope. Typical setup: 3-5 core workflows on n8n or Make.com, monthly iteration. Good pilot.
€8,900/month (Scale). Two teams. Common combo: Operations (workflows) + Growth (lifecycle email, content). 10-15 workflows under management, weekly iteration.
From €16,000/month (Enterprise). Three or more teams. Full operations + finance + dev coverage. 25+ workflows including custom code on top of n8n. Dedicated operator team.

Infrastructure costs are passed through at cost. Typical n8n self-hosted runs €15-€80/month on Hetzner. Make.com runs €30-€400/month depending on operations volume.

What we do not do

One-off workflow projects. If you want a single Make scenario built and handed over, we are not the right vendor. Try a freelance Make consultant. We make sense when you want ongoing operation, not a deliverable.
Build your own platform. We use existing tools (n8n, Make.com, Temporal, Zapier where it fits). We do not build custom workflow engines from scratch unless you are at a scale where that math actually works (rare).
Replace your engineering team. When a workflow needs more than n8n can do — high concurrency, complex state machines, low-latency requirements — we hand it over to the Dev Team or to your engineers. No-code has limits.
Touch regulated systems we are not certified for. Workflows that touch licensed medical, legal advisory, or audited financial output do not get automated by us. Back-office around those, yes.

FAQ

Do you self-host n8n or use n8n Cloud?

Both. Self-hosted for clients with strict data residency requirements (Hetzner or AWS in the EU). n8n Cloud for teams who prefer not to manage infrastructure. Secrets stay in your vault either way.

When do you recommend Make.com over n8n?

Make.com wins when workflow logic is mostly visual chaining across mainstream SaaS, the team includes non-developers, and you want managed infrastructure. n8n wins for self-hosted, custom JavaScript steps, larger volume, and engineering-capable teams.

What does Logitelia do that I cannot do myself with n8n or Make.com?

Three things: (1) reliable AI judgment steps with operator review; (2) production hardening — retries, dead-letter queues, monitoring, audit logs; (3) ongoing maintenance as APIs change. Subscription, not project.

Can you migrate workflows from Zapier or Make.com to n8n?

Yes. Typical reason: cost. Migrations take 1-3 weeks depending on workflow count. Common outcome: Zapier €600/month → n8n €40/month, with feature parity.

Do you handle EU data residency?

Yes. n8n self-hosted on EU infrastructure (Hetzner Falkenstein/Helsinki, AWS Frankfurt). Make.com Enterprise plan supports EU residency. All credentials stay in your secret manager.

What does it cost?

Part of the Operations Team subscription from €4,500/month. Infrastructure passed through at cost (typically €40-€400/month depending on volume).

Want to see your stack mapped to a workflow plan?

Book a 30-minute intro call. We will sketch the 3-5 workflows that would pay for themselves first, with honest estimates of where AI judgment adds value vs adds risk.

Book intro call