Freelancers and small business owners usually end up stitching together an invoicing app, a spreadsheet for expenses, and a separate tool (or an accountant) for tax filing — and those three things rarely agree with each other. Frihet puts all of it in one place: invoices, expense photos, and the books that need to balance at the end of the month. Frihet is an AI-native ERP covering invoicing, OCR-based expense scanning, multi-currency support (170+ currencies), double-entry accounting across 96 countries, and automated tax compliance including Spain's VeriFactu and TicketBAI regimes — a strong signal the product is built especially for Spanish-speaking markets. It adds a CRM with predictive churn scoring, a REST API and TypeScript SDK, an open-source MCP server for AI-assistant integration, and 100+ integrations (Stripe, Shopify, WooCommerce, Zapier, n8n). Pricing is freemium: free up to 999 invoices/month, Pro at €9/month (tax models, CRM, recurring invoices) and Premium at €29/month (advanced accounting, e-invoicing, AI copilot with 100 messages/month).
| # | Tool | Best for | Price |
|---|---|---|---|
| 1 | Développeurs et organisations construisant des agents IA en production nécessitant conformité, audit et gouvernance des données dès la conception | — | |
| 2 | Développeurs d'applications IA sensibles à la latence d'inférence (temps réel, interactif) | — | |
| 3 | Joueurs | — | |
| 4 | Radioamateurs | — | |
| 5 | Passionnés de musique | — | |
| 6 | Musiciens | Développeurs | — | |
| 7 | Professionnels | — | |
| 8 | Développeurs | — | |
| 9 | Entreprises | Startups | — | |
| 10 | Musiciens | — | |
| 11 | Freelances | Petites entreprises | — | |
| 12 | Petites entreprises | — |
A free, open-source memory system for AI agents that remembers things the same reproducible way every time — instead of the usual approach where an agent's memory changes slightly each time it searches, which makes its behavior hard to trust or audit.
An AI inference API built specifically for speed — over 1,000 image, video, audio and language models served through one endpoint with sub-1-second latency and no cold starts.
