AgentenTrust 2.0
36 normalized controls for privacy, security, human control and auditability — each with scope, verification date and source.
View trust evidence →The independent intelligence platform for AI agents
AI agents understand, find and compare.
AgentenCode analyzes and documents AI agents, platforms and agentic systems with primary sources, verified history and a clear focus on privacy, the EU AI Act and practical deployment.
Independent & transparentprimary sources & verificationPrivacy & GDPREU AI Act focus
130AI agents
analyzed & documented
analyzed & documented
802Trust signals
with source references
with source references
250Changes
in verified history
in verified history
EUPrivacy & AI Act
considered systematically
considered systematically
★Focus on Europeand responsible AI
NEW · AI ECOSYSTEMS
Compare OpenAI, Claude, Gemini and other AI ecosystems.
One shared agent directory, seven ecosystems and source-first comparisons that distinguish agents, platforms and models.
Ask AgentenCode
One question. One evidence-backed answer.
Ask about agents, privacy, governance, MCP, human approval or changes. AgentenCode shows evidence, unknowns, scope, verification dates and primary sources.
Explore
AI agents by category.
</>
Development
Code generation, software engineering and DevOps.
→▥Enterprise
Processes, knowledge work and business applications.
→◎Research
Analysis, research, learning and decisions.
→⚙IT & Operations
Orchestration, monitoring, security and systems.
→▤Service & data
Customer service, CRM, analytics and data work.
→◇Knowledge & governance
Privacy, EU AI Act, security and methodology.
→Trust, privacy & EU AI Act
Governance becomes visible — not just claimed.
AgentenCode separates confirmed evidence from unknowns and shows which claims about privacy, data residency, human control, auditability and EU relevance are actually documented.
◉
Privacy / GDPR
Data processing, retention, DPAs, subprocessors and training on customer data.
Open trust evidence →✦EU AI Act
Regulatory context and control mapping without blanket compliance claims.
Open EU layer →◎Data residency
Documented locations, EU hosting and the scope of provider statements.
Review evidence →⌁Governance & auditability
Human approval, audit logs, history and traceable changes.
Open history →Agent profiles
Key AI agents at a glance.
Quick cards for orientation and detailed profiles for decisions: capabilities, use cases, limits, source status, FAQs and changelogs.
✓ verifiedResearch
ChatGPT Deep Research
OpenAIMulti-step research in ChatGPT using the web, files, controllable sources and cited reports.
Shopping researchProduct comparisonTravel planning
Verified Sep 30, 2026 · profile 2.0Open agent profile →
✓ verifiedResearch
Gemini Deep Research
GoogleAgentic research in Gemini with Google Search and optional Google data sources.
Travel planningShopping researchLearning
Verified 01. Oct 2026 · profile 2.0Open agent profile →
✓ verifiedCoding
Devin
CognitionSoftware engineering agent for delegable development tasks such as changes, migrations and maintenance.
Backlog tasksMigrationsBug fixes
Verified Sep 30, 2026 · profile 2.0Open agent profile →
✓ verifiedCoding
Claude Code
AnthropicAgentic coding assistant for terminal and development environments that can read and modify code and run commands.
Analyze codebasesImplementationDebugging
Verified Sep 30, 2026 · profile 2.0Open agent profile →
✓ verifiedEnterprise
Salesforce Agentforce
SalesforceSalesforce platform for building and operating agents that use CRM data, business logic and actions.
Customer serviceSalesCRM automation
Verified Sep 30, 2026 · profile 2.0Open agent profile →
✓ verifiedCustomer service
Intercom Fin
IntercomIntercom AI agent for customer service that answers questions and can perform defined actions or handoffs.
Customer serviceFAQ automationSupport triage
Verified Sep 30, 2026 · profile 2.0Open agent profile →
Agent systems
From individual agents to the system layer.
AgentenCode documents not only products but also sourced components, protocols, delegation, handoffs and governance as a versioned layer in AgentenGraph.
Agent systems
Separate agents, frameworks, components, systems and institutions and document their relationships source-first.
Open system layer →
Agent approval
Assess concrete system states across identity, permissions, delegation, approval, audit, rollback, memory and hosting.
View assessment model →
Artificial Symbiotic Intelligence
The current DeepMind Institute essay on agent societies, orchestration and institutions — with a clear separation between thesis and product fact.
Read analysis →
News
Important AI-agent changes.
Not a news firehose: only editorially confirmed product changes with a primary source, verification date and a direct link to the agent profile.
Security & agent execution · October 8, 2026
GitHub Copilot local sandboxing reaches general availability
GitHub Copilot adds generally available local sandboxes for CLI, app and VS Code Agent Host sessions, with controlled filesystem and network access.
Read update →Coding & local models · October 8, 2026GitHub Copilot CLI discovers local Ollama models
Copilot CLI 1.0.94-0 discovers eligible Ollama models through /model; installation, offline mode and telemetry settings remain separate.
Read update →Coding & model availability · October 8, 2026Claude Haiku 5.5 arrives in GitHub Copilot
Claude Haiku 5.5 is rolling out in GitHub Copilot across IDEs, CLI and its cloud coding agent, subject to plans, admin policies and usage-based pricing.
Read update →From agent to system
The next layer: composition, relationships and governance.
An agent is increasingly not an isolated box. Models, tools, memory, protocols, subagents and human approvals form a system whose structure must itself be documented.
For organizations
Is your process really a use case for an AI agent?
The free Agent Check separates automation, copilot, hybrid and agent patterns and also evaluates technical readiness, risk, autonomy and business case.
✓ Free✓ No registration✓ Data stays in your browser
Check your process →Example assessment
Agentic hybrid process
Agent opportunityhigh
Readinessmedium
Control & riskhigher
Knowledge center
Understand AI agents before deploying them.
Definitions, technology, security and practical context — the editorial foundation of the index.
For individuals
Practical guidance without technical overload.
Understand agents for research, learning, purchasing decisions and everyday tasks.
Explore individual use cases →For organizations
From use case to controlled rollout.
Select agents, limit autonomy, assess risks and prepare a measurable pilot.
Explore enterprise use →Source-first methodology
A reference must be traceable.
Documented product properties and editorial interpretation are separated. If reliable confirmation is missing, AgentenCode does not guess.
Read methodology →01primary sources first
Official documentation and provider sources take priority.
02Unknown rather than guessed
Unconfirmed features are not presented as facts.
03Verification date & changelog
Changes remain traceable.
04No paid rankings
Commercial relationships do not change the assessment.
AgentenCode intelligence layer
Not only “What can the agent do?” — but “What is documented, in what scope, and what changed?”
AgentenCode combines agent profiles with 802 governance signals, 79 explicit relations and 250 verified history events. Field-level trust history continues from October 5, 2026.
Ask AgentenCode
Structured questions map to evidence, relations or stored history events. Unknown remains visible.
Ask the evidence layer →History query
Search verified changes by agent, provider, type and time period. No retroactively invented states.
Search history →