AI Agent Era Accelerates: Autonomous Execution Race from Meta to Prentis

에이전트 | Tue Jul 28 2026 00:00:00 GMT+0000 (Coordinated Universal Time) | 6 sources

Meta AI shifted toward an execution-oriented assistant while computer-use agent startups emerged and enterprise agentic AI infrastructure expanded.

Analysis

[Meta AI] launched execution-oriented AI assistant powered by Muse Spark 1.1 ^[1]^[2]

Automated planning and follow-through execution
Calendar-integrated daily briefings
Marketplace search and moodboard generation
Research with real-time course correction

[Meta] integrated Meta AI chatbot into Threads DMs ^[3]

AI conversation support within Threads private message windows
Ability to share posts
images
links
and videos
Global rollout starting Monday
Strategy to prevent defection to third-party AI assistants

[Prentis] negotiated $100M raise at $1B valuation for computer-use agent development ^[4]

Co-founded by Reid Hoffman and Mark Pincus
Claims Hive-32B model outperforms GPT-5.4 and Claude Opus 4.6
Already signed $50M in customer contracts
About 10x cheaper per task than frontier APIs

[Outshift by Cisco] built Internet of Agents through AGNTCY open-source project ^[5]

Support for mutual discovery and identity verification between agents
Open source under the Linux Foundation
Oriented toward distributed artificial superintelligence
Solving collaboration problems in multi-agent systems

[Intel] released infrastructure benchmarks for enterprise agentic AI workloads ^[6]

Extended Terminal-Bench with profiling and telemetry
Capacity planning based on agent density per vCPU
Presented six key metrics
Recommended scale-out-first architecture

Sources