/
/
A comprehensive ERP for service-focused SMBs — HR, CRM, projects, timesheets, finance, reporting — with a native AI connector so Claude and ChatGPT can read your company data directly. The first ERP that's actually ready for the AI age.
Visit goran.czGoran ships a native AI connector. Connect your favourite assistant once and ask questions over real HR, CRM, project and finance data — with row-level permissions inherited from your Goran user, full audit trail, and no copying spreadsheets into a chat window.
Log 90 minutes on the G3 project for today's auth-refactor work.
Approve last week's timesheets on the projects I manage.
What's happened on my open opportunities this week — state changes, new comments?
Create an opportunity for ACME Corp at 60% probability with an expected value of 250k.
Build a client report of this month's timesheet activity on the ACME project.
Claude
ChatGPT
+ any AI client (MCP-compatible)Employee records, onboarding, time tracking, leave management — and an AI connector your agent can ask about.
Visual pipelines, lead scoring, opportunity stages — query the same data from Claude when you'd rather just ask.
Invoicing, expenses, budgets, P&L. The numbers are auditable — and now conversational.
Project hierarchies, phases, and per-project timesheets — billable hours, utilisation, and budgets in one place.
Real-time KPI dashboards. Custom reports for what your dashboards don't cover. Generated reports for what you can't be bothered to build.
Authenticated, scoped, audited. Permissions follow your Goran user — your assistant sees only what you'd see in the UI.
Most ERPs give you dashboards. A few give you an export button. Almost none let an AI assistant read your data safely — with the same permission model your humans live under.
Goran's connector fixes that. It exposes the data model
as a set of tools that Claude or ChatGPT can call —
create_my_timesheet_item,
get_business_cases_list — the same way a
developer would call an API, but conversationally, with full
audit logging. Under the hood it speaks the open
Model Context Protocol (MCP) standard, so
any compatible AI client just works.
Every connector request runs as the connecting user. If a salesperson can't see the engineering team's payroll in the UI, their assistant can't see it either. No prompt-injection trick will change that — the boundary is enforced server-side.
Every call is logged with prompt, response, and the data accessed. Hand the log to your CISO, point at it in your SOC 2 evidence room, and move on.
Goran isn't trying to compete with SAP for global manufacturing. It's built for the 50–500-person service company — agencies, consultancies, IT services, professional services — where the data model is people + projects + hours + invoices, and the bottleneck is reporting, not transactions.
Drop-in modules, sane defaults, no six-figure implementation. And now: conversational access to all of it.