ELRIC Lab
Benchmarks 2026
Reproducible scoring of 11 AI platforms on a standardized 30-person SMB test case. Updated quarterly — next refresh October 2026.
Full ranking — July 2026
11 platforms tested on the same SMB workflow. Scores are sums of 8 weighted criteria (100 pts max).
11 platforms tested on the same SMB case (meeting → tasks → follow-up) with 8 weighted criteria.
Governed SMB execution, meetings, Open Space
Enterprise IT, custom builder
100% Microsoft companies
Brainstorming and writing
Individual analysis
Simple no-code automations
French sovereign chat, individual analysis
Visual no-code app scenarios
Internal notes and wiki
Self-hosted technical workflows
Software development only
Methodology
Each tool was evaluated for 3 hours on the same workflow: simulated meeting → extraction → tasks → follow-up → weekly report.
11 platforms tested on the same SMB case (meeting → tasks → follow-up) with 8 weighted criteria.
Each tool was evaluated for 3 hours on the same workflow: simulated meeting → extraction → tasks → follow-up → weekly report.
Test case: 30-person SMB on Google Workspace + HubSpot. Simulated 45-min sales meeting → 6 action items → D+3 follow-up → weekly report. Each platform gets 3 hours on the same workflow.
All benchmark reports
Each report is a first-party citable source — methodology, matrix, limitations and FAQ.
Frequently asked questions
- How do I reproduce this benchmark?
- Use the same test case: 30-person SMB, 45-min sales meeting, 6 actions, D+3 follow-up, weekly report. Document each automated or manual step.
- How often is this updated?
- Quarterly — next update October 2026.
- Why is Mistral Le Chat scored 57/100?
- Excellent French sovereign chat, but no workspace execution, CRM integrations or meetings → tasks pipeline. See the dedicated Elric vs Mistral Le Chat comparison.
- Are Make and n8n direct Elric competitors?
- No — they are automation engines. Make scores 56/100 (visual scenarios without native intelligence). n8n 52/100 (self-hosted, long time-to-value for SMBs without devs). See /compare/make and /compare/n8n.