AI ranking
Proprietary / cloud agentic systems
Proprietary and cloud agentic systems from Le Recul's editorial catalogue (Claude Cowork, ChatGPT Agent, Microsoft Copilot, Manus, Genspark…). Le Recul Agentic score (0–90 method).
Visual top 10 — Le Recul score
- 01 ChatGPT Agent 76.50
- 02 Microsoft Copilot Cowork 76.50
- 03 Gemini Enterprise 73.80
- 04 Claude Code 64.80
- 05 Codex CLI 64.80
- 06 Copilot CLI 64.80
- 07 Claude Cowork 62.10
- 08 Genspark 61.20
- 09 Manus 60.30
- 10 Microsoft Copilot 58.50
Compare models. Tick up to 3 models in the table below, then run the comparison: who wins on which use, by how much, and how solid the measurement is.
| # | Model | Vendor | Score | Assessment | |
|---|---|---|---|---|---|
| 01 | OpenAI agent built into ChatGPT, able to browse the web and handle files. | OpenAI | 76.50 / 100 | Le Recul | |
| 02 | Cowork variant of Microsoft Copilot for multi-agent collaboration. | Microsoft | 76.50 / 100 | Le Recul | |
| 03 | Google enterprise offering built on Gemini, with customisable agents. | 73.80 / 100 | Le Recul | ||
| 04 | Anthropic command-line coding agent, built on Claude. | Anthropic | 64.80 / 100 | Le Recul | |
| 05 | OpenAI command line for Codex/GPT, built into the OpenAI ecosystem. | OpenAI | 64.80 / 100 | Le Recul | |
| 06 | GitHub Copilot command line, built into GitHub. | Github / microsoft | 64.80 / 100 | Le Recul | |
| 07 | Anthropic collaborative workspace built on Claude — email, docs, research, analysis. | Anthropic | 62.10 / 100 | Le Recul | |
| 08 | Web research and synthesis agent, an alternative to Perplexity. | Genspark ai | 61.20 / 100 | Le Recul | |
| 09 | Autonomous multi-task agent positioned as a general-purpose alternative to cloud assistants. | Manus ai | 60.30 / 100 | Le Recul | |
| 10 | Microsoft agentic suite built into Microsoft 365, Outlook and Teams. | Microsoft | 58.50 / 100 | Le Recul | |
| 11 | Google agentic studio built into Workspace (Gmail, Docs, Sheets). | 58.50 / 100 | Le Recul | ||
| 12 | Droid Factory.ai agentic platform for software development. | Factory.ai | 54.90 / 100 | Le Recul |
Indicative score, weighted from public sources. Transparent methodology. See the AI ranking page for the full methodology and the other categories.
Proprietary and cloud agentic systems
This view gathers the proprietary and hosted agentic systems from Le Recul's editorial catalogue: commercial agents, run at their vendor's, to which you delegate a complete task rather than a single answer.
Their promise is appealing, and that is precisely why caution is required. A cloud agent reaches tools, browses, writes files, sometimes sends messages — all of it outside your machine. The question is therefore not only « is it capable » but « what can it do, with which permissions, and what happens when it gets something wrong ».
This view is an editorial selection, not an automatic ranking: agentic systems do not yet lend themselves to public measurement comparable to that of models. It serves as a map of the landscape, not as a scored league table.
What this ranking is for
This view helps you place the cloud agents available and understand what really separates them: scope of action, supervision required, reversibility of the actions taken.
It is a reminder that delegating to a hosted agent is also a confidentiality decision: the data processed and the access granted leave your environment.
How to read the ranking
- Real scope of action: tools reachable, actions allowed.
- Level of supervision required for serious use.
- Reversibility: what can be undone if the agent gets it wrong.
- Handling of data and of the access granted.
- Maturity of the system and stability observed publicly.
FAQ — Proprietary / cloud agentic systems
Why does this view have no numerical score?
Because there is not yet a comparable, reproducible public measurement for complete agentic systems. An invented score would be less useful than an acknowledged selection.
What is the difference with the AI agents ranking?
The AI agents ranking covers agentic systems as a whole. This view isolates the proprietary and hosted ones, as opposed to local or open-source systems.
Is a cloud agent risky?
It is not dangerous by nature, but it acts outside your environment with the access you grant it. The risk is managed through permissions and supervision, not through trust.
Should you prefer a local agent?
It depends on the sensitivity of the data and the tasks. Local protects confidentiality but demands more configuration and vigilance from you.
How often does this page change?
It follows the editorial catalogue, updated as tests and the market evolve. The date is shown at the top of the page.