Enterprise AI Agent Ecosystem Expands with Accelerating Standardization

에이전트 | Wed Jun 17 2026 00:00:00 GMT+0000 (Coordinated Universal Time) | 8 sources

OpenAI, Anthropic, Hugging Face and others are driving enterprise agent adoption through training programs, partnerships, and open protocols.

Analysis

[OpenAI Academy] launched new Agents and Workflows training courses ^[1]

Three courses: AI Foundations
Applied AI Foundations
and Agents and Workflows
Partnerships with BCG
Accenture
and BBVA
Teaches building of repeatable agent-based workflows

[Anthropic] formed alliances with TCS and DXC for Claude agents in regulated industries ^[3]^[4]

TCS will deploy Claude to 50
000 employees across 56 countries
DXC OASIS adopted Claude as the default model for agent workflows
Industry-specific agent solutions for claims processing and lending advisory

[Anthropic Seoul] opened a Korean office and expanded agent partnerships with NAVER and Samsung SDS ^[2]

NAVER deployed Claude Code to its entire engineering organization
Samsung SDS runs agentic workflows with Claude Cowork and Claude Code
Adoption by LG CNS and Hanwha Solutions

[Hugging Face] released the Agentic Resource Discovery open specification ^[5]

Co-developed with Microsoft
Google
GoDaddy
and others
Discovers MCP tools
A2A agents
and Skills at runtime
Includes ai-catalog.json manifest and POST /search REST API

[OpenEnv] launched as an open-source environment standard for agent RL training ^[6]

Steering committee includes Meta-PyTorch
Nvidia
Microsoft
and Hugging Face
Gymnasium-style reset/step/state API
Interoperability layer for publishing
distributing
and consuming environments

[H Company Holo3.1] released a locally runnable computer-use agent model ^[7]

Performance on AndroidWorld improved from 67% to 79.3%
Provides FP8
Q4 GGUF
and NVFP4 quantized checkpoints
Various sizes from 0.8B to 35B-A3B with native function-calling support

[IBM Research] highlighted the importance of agent logic in enterprise agents ^[8]

Software primitives such as knowledge graphs and program analysis
Reduces hallucinations and token usage by shrinking the LLM context space
Applied to Cobol comprehension
test generation
and compliance modernization

Sources