Enterprise-grade from day one
Live from the first day, and portable later to any system. No vendor lock-in to Jourier: take the code, the schema and the models and run them yourself.
Time Tracking
Connect Harvest
Dashboards, a full API and MCP access over your Harvest data. Hosted in Finland. Self-hostable, or buy the platform outright and run it yourself.
Portable technical infrastructure to any cloud.
A production-ready foundation designed to scale. Plug in pre-built modules for common workflows, or build your own on top.
Hosted in Finland
Data and AI run in a Finnish data centre. UpCloud in Helsinki, and Mistral trained with CSC on the LUMI supercomputer in Kajaani.
Enterprise modular architecture
Scales to any level, from a new startup to a large enterprise. Build your own modules, or change the ones that ship.
The stack behind the unicorns
Talos Linux, Kubernetes and Cilium. GitOps deployments, a separate database per customer, automated security controls and self-healing.
Harvest API
One API over Harvest and everything else
Most systems expose one API shaped around their own storage. Useful, but you build against that shape, handle its pagination, and repeat the exercise for every other system you own.
Jourier reads Harvest and serves the result as modelled JSON, on one endpoint, alongside every other system you connect. It is joined, typed and reconciled to what the dashboards show.
- Entries, projects, rates and approvals as first-class objects
- The same numbers the dashboards use, from one definition
- Read-only against Harvest; nothing is logged or approved
# project burn against quoted hours GET /v1/time/burn?status=active { "source": "harvest", "as_of": "2026-08-16T06:12:00Z", "rows": [ { "project": "Atlas rollout", "quoted_h": 320, "logged_h": 291, "pct_done": 74 } ] }
Harvest MCP
Ask Claude about your own hours
Model Context Protocol is how an AI assistant calls a real tool instead of guessing. Jourier exposes your connected systems as MCP tools, so Claude, Cursor or any MCP client can query Harvest directly, and answer from the time record rather than from a plausible sentence about it.
Every answer resolves to rows you can open. The model retrieves; it does not compute the number, which is the only reason the answer is worth having.
- Works with Claude Desktop, Cursor, and any MCP-compatible client
- Scoped to what that user is allowed to see
- Self-hosted models available, so your time data never leaves the EU
You Which active projects will overrun the quote at the current rate? Claude Two will. One is 74% through the hours and 52% through the scope.
Where the data lands
Harvest data in the tool you already use
Each has its own setup page: what syncs, how often, and what the schema looks like on arrival.
Harvest is one door in
Behind it is a full platform. EU-hosted in Finland, built on open components, portable by construction. None of it depends on staying a Jourier customer.
Sovereignty
Finnish hosting, Finnish AI
Your data stays in Finland, on EU infrastructure, under EU law. The AI that reads it is self-hosted, so your ledger is never a US vendor's training material.
The platform ArchitectureSelf-hostable Kubernetes
Standard Kubernetes, open components. Take the manifests and run it on your own cluster whenever you decide to. No proprietary runtime holding you in.
The platform AtlasCompany intelligence
Every registered Finnish company: financials, ownership, lifecycle and public procurement. Find buyers who look like your best customers, and tenders you can bid on.
Company Atlas MonitoringWatch what moves outside
Continuous external monitoring on independent scopes: supply chain, competitors, and how visible you are to AI assistants. It tells you; you don't go looking.
Monitoring
Connecting Harvest
Paste credentials, nothing to install
Access is granted per customer, so connecting is a matter of pasting keys. Nothing is installed and no port is opened.
What you'll need
- Account ID
- Secret
- Personal access token
- Secret
- Access
- Read only
Harvest and Jourier, answered
Can I get Harvest data into a dashboard?
Yes. Jourier reads Harvest and serves utilisation, billable ratio and project burn as dashboards that are already built. You connect Harvest and they fill with your own entries, so there is no modelling project first.
Does Jourier write anything back to Harvest?
No. The connection is read-only against Harvest. Nothing is written back.
Where is Harvest data stored when I use Jourier?
In Finland, on EU infrastructure, under EU law. The AI that reads it is self-hosted, so your data is never a US vendor’s training material.
How long does connecting Harvest take?
The connector is live. Connecting is a matter of pasting credentials: nothing is installed and no port is opened, and data flows the same day.
Can I use Harvest data from Claude or another AI tool?
Yes. Jourier exposes MCP tools, so Claude, Cursor or any MCP client can query Harvest directly and answer from the time record rather than from a plausible sentence about it. Access is scoped to what that user is allowed to see.
Can I run it on my own infrastructure?
Yes. It is standard Kubernetes built on open components. You can take the manifests and run it on your own cluster whenever you decide to, and there is no proprietary runtime holding you in.
Does this work with other systems as well as Harvest?
Yes. Jourier serves Harvest alongside every other system you connect, on one endpoint, joined, typed and reconciled to what the dashboards show.
Get started
Let's discuss what Jourier can do for your company.
Book a meeting
Try the demo
Instant access to Jourier Lens. Business email only.