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).

Ranking updated on 02 November 2026 at 11:47

Visual top 10 — Le Recul score

  1. 01 ChatGPT Agent 76.50
  2. 02 Microsoft Copilot Cowork 76.50
  3. 03 Gemini Enterprise 73.80
  4. 04 Claude Code 64.80
  5. 05 Codex CLI 64.80
  6. 06 Copilot CLI 64.80
  7. 07 Claude Cowork 62.10
  8. 08 Genspark 61.20
  9. 09 Manus 60.30
  10. 10 Microsoft Copilot 58.50

Le Recul comparator

Who wins on what

Tick at least two models in the ranking table, then come back here.

# 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.

Google 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).

Google 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.

See the full FAQ →

Related rankings