Services
AI & automation services for growing businesses
Four anchored services — from upfront assessment through production-grade agentic AI — built on n8n, Make.com, Python, LangChain, OpenAI, and Anthropic. Engagements run from a single targeted workflow to multi-month build-and-handover programs.
How we work together
How I help you with AI & automation
From understanding where AI fits, through automation and intelligent integrations, to autonomous agents that work inside your stack.
Common scenarios
What I Solve for businesses
Six recurring situations I see in growing companies — tap a row for how I approach it and what changes.
Expand a scenario for the full approach and concrete outcomes.
Tools I build with
I'm tool-pragmatic: pick what survives production, not what trends on LinkedIn. I'm equally happy meeting clients on the stack they already run — Microsoft, Google, Azure — as I am running self-hosted infrastructure end-to-end. The current default stack:
Where the workflows actually run
n8n
Self-hosted workflow orchestration — my default for SMB automation. Queue mode for scale.
Make.com
Cloud orchestration when self-hosting isn't a fit. Faster to ship, less control.
Power Automate
Microsoft's enterprise workflow platform — the right call when clients are deep in Microsoft 365, SharePoint, Teams, and Dynamics.
Prefect
Python-native orchestration for data and ML pipelines that need observability and retries.
Pipelines, transforms, source of truth
Python
Custom logic, reliability glue, data transforms — anything an off-the-shelf node can't do cleanly.
dbt
Modular SQL transformations with version control, testing, and documentation built in.
dlt
Lightweight Python ELT — load data from any API into your warehouse with schema evolution.
Postgres / Supabase
Source-of-truth storage, vector search, auth, and edge functions in one place.
The reasoning layer
LangChain / LangGraph
Agent orchestration and tool-use patterns when the workflow needs reasoning, not just routing.
OpenAI
GPT models for breadth and tool use. Always evaluated, never assumed.
Anthropic
Claude for reasoning and long-context work. Default choice for high-stakes outputs.
Google Gemini
Strong on multi-modal reasoning, native tool use, and long context. Routed via Vertex AI for clients on Google Cloud.
Nano Banana Pro
Google's premium image generation model — production-grade visual assets with strong prompt adherence. For marketing automation and content workflows.
Where it all runs
Coolify
Self-hosted PaaS for clients who need on-prem or sovereign deployment.
Cloudflare / Cloudflare One
Domain handling, DNS, Zero Trust security, Workers, and edge protection.
Terraform
Infrastructure as code — repeatable, reviewable, version-controlled environments.
Vercel
Hosting for the user-facing layer when speed of iteration matters.
Railway
Backend services and APIs in single-region deploys.
Working with the tools your team already uses
Microsoft 365 / Azure
Deep integration with Microsoft Graph, Azure OpenAI, SharePoint, and Teams — for clients running on Microsoft.
Google Workspace
Drive, Gmail, Calendar, Docs, and Workspace Add-ons — for teams running on Google.
Common questions
What does an AI automation consultant actually do?
Three things: (1) figures out where AI and automation fit in your operation — and just as importantly, where they don't; (2) builds and ships those systems on stacks like n8n, Make, and Python; (3) hands them over with documentation so your team owns the result, not me.
How much does a typical engagement cost?
An AI readiness assessment runs roughly 1–2 weeks and starts at fixed pricing. Workflow automation projects depend on scope; a single targeted workflow is typically 2–4 weeks. Larger programs combining assessment, build, and handover run 2–4 months. I price most work as fixed-scope or time-and-materials with a budget cap. Retainers are available after a first engagement.
What size of business is this for?
The sweet spot is companies with 5–250 employees that have outgrown spreadsheets and consumer SaaS but aren't running a dedicated automation engineering team yet. If you have one specific workflow that's costing you time and you can name the cost in hours or money, that's a good starting point.
Do you work with companies outside Sweden?
Yes. Most engagements run remotely across the EU, with on-site visits in Gothenburg, Stockholm, and surrounding regions when it adds value. English and Swedish.
What's the difference between automation and AI agents?
Automation is deterministic — given input X, do step Y. Useful for handoffs, notifications, sync. AI agents add reasoning — given a goal and a set of tools, decide what to do next. Useful for triage, research, customer interactions where outcomes vary. Most projects use both: deterministic flows with AI nodes where judgment is needed.
Can you work with our existing tools?
Almost always yes. Standard integrations cover CRM (HubSpot, Salesforce, Pipedrive), helpdesk (Zendesk, Intercom, Freshdesk), comms (Slack, Teams, email), storage (Drive, SharePoint, S3), and most modern SaaS APIs. If a tool has an API, it can be wired in.
What happens after the project ends?
You get the code, the documentation, the runbooks, and ownership of any infrastructure. Optional retainer covers maintenance, model updates, and small extensions. No vendor lock-in.
Not sure which service fits?
Start with an AI readiness assessment. Two weeks, fixed price, ends with a prioritized roadmap and a budget range — no commitment to a build phase.