Hi all,

Hope you've had a good week so far. Here's the final roundup of links - agent news has been coming nonstop even in the last five days.

Also: AGT NYC is taking a vacation. Have a great summer, see you in August!

Cheers,
Ivan

Invite a friend to join AGT NYC at agtnyc.com.

Events

• Jun 11 - Rust Demo Night w/ PostHog and One - hosted by Jacob Rissman (AGT member)
• Jun 6 - Agentic & Applied AI for the Enterprise Conference, Atlanta
• Jun 10 - IBM TechXchange Hands-On Workshop: How to Provide Agents an AI-Ready Data Platform
• Aug 10 - Agentic AI Summit, Berkeley
• Oct 5 - Agentic AI, Virginia
• Oct 28 - AGENTICS 2026, France

Find more events on the AGT NYC Luma calendar.

News

• Apple approves Poke as the first AI agent on its Messages for Business platform
• Morgan Stanley will soon open its trillion-dollar wealth management funnel to AI agents

One type of agent expansion that hasn't been seen as much earlier is opening up existing markets and processes to second-party and third-party agents. This is likely more efficient than doing all of the research directly.

Launches

• Meta introduces Business Agent (and personal agent earlier)
• NVIDIA launches RTX Spark, a "superchip" for personal agents and Vera, the CPU for Agents
• Project Solara: a new platform for agent-first devices
• Introducing Microsoft Scout: Your always-on personal agent
• Introducing Managed Agents in the Gemini API
• IBM watsonx Orchestrate introduces an agentic control plane
• Workday launches Agent Passport

Big Tech has thoroughly capitalized on the agent trend, to the point where even niches aren't safe. That said, it's not clear which products will stick.

Deals

• Gradient Labs ($26M) - financial specialist agents
• SageOx ($15M) - shared agent context
• CodeWords ($9M) - prompt to agents
• Foundation ($6M) - agent authorization platform
• Elastics ($2M) - prediction market agents

These smaller fundraises round out last month's batch with interesting ideas. Market and trading agents seem like a promising category, though likely not that easy to turn into a proper commercial offering.

Projects

• Open Agents - agent reference app
• Astrid - agent operating system
• Link CLI - Stripe for agents
• Cua Driver - agent computer use
• Homecrew - agent skill manager
• Awesome Harness Engineering - useful harness links

A really positive trend here is startups open sourcing their battle-tested and production-grade tools and patterns. It makes building so much easier.

Other

• The Agent Labs Thesis by swyx
• Agentic Evaluations at Scale, For Everybody — Nicholas Kang & Michael Aaron, Google DeepMind [20:02]
• Who’s Adopting AI Agents - and What They’re Actually Doing With Them in HBS Working Knowledge
• MOSS: Self-Evolution through Source-Level Rewriting in Autonomous Agent Systems
• Your Checklist to Launch a Production-Ready AI Agent by Arthur
• AI Risk Quadrant for Agent Security
• Adapting the Interface, Not the Model: Runtime Harness Adaptation for Deterministic LLM Agents
• TLMs: Tiny LLMs and Agents on Edge Devices with LiteRT-LM — Cormac Brick, Google [1:20:57]
• Emergence World: A Laboratory for Evaluating Long-horizon Agent Autonomy

A broader roundup here than usual, with some fascinating info. I'm particularly interested in the combination of small models, long horizons, self-evolution, and harness adaptation. That could be the recipe for scalable production agents.

Comments, suggestions? Reply to this email, let me know what you think!