#1 Edition: This is the week big tech goes all-in on AI Agents

[FYI: I sent this newsletter already yesterday, but some of the links were broken. Thanks to everyone who flagged it and shared feedback! I've now republished it — and I can assure you all links are working properly.]

Hey, it’s Andreas.

Spring is in the air — and so is a seismic shift in AI.

This week alone, we saw some of the biggest moves yet from big tech, all pointing in one direction: It’s not AI-first anymore — it’s AI-agent-first.

Perfect timing to launch Human in the Loop — your weekly edge on AI agents, real tools, and what’s actually working. If you care about building, scaling, or staying ahead in the agent economy — this is for you.

Why AI Agents?

AI agents are the next big frontier. And right now, very few are watching it closely. This is your early window to find your niche, level up fast, and help shape what comes next. At IBM, there’s barely a client conversation these days that doesn’t involve AI agents.

And the market knows it too: The global AI agent market is projected to grow from $5.3 billion in 2023 to $216.8 billion by 2035 — with a compound annual growth rate of over 40% (!!!).

And AI agents won’t be a feature on the side. They represent a new operating layer — and the next wave of intelligent systems. Remember what NVIDIA CEO Jensen Huang said at the beginning of the year at CES 2025? That “Agentic AI is the next frontier” — because for the first time, intelligence becomes action. There are clear signs that this new frontier — the era of agentic AI? It officially began this week.

Every major tech company — from Microsoft to Google to SAP — is now building toward a world where AI agents become the dominant force in how work gets done.

But the playground is still forming and everybody is still experimenting, and that’s exactly why now is the moment to build an edge.

That’s what Human in the Loop is here for.

In today’s edition, we will cover…

• Why every major tech company is going all-in on agents

• Tool ppotlight Mem0 — memory for agents

• AI-Agent news worth knowing

• and much more…

Let’s dive into it!

_{First time reading? Sign up here.}

🔥 Top Story
Why every big tech company is now all-in on AI-Agents

We saw three flagship conferences this week, but all with a very similar narrative across.

Microsoft (Build), Google (I/O), and SAP (Sapphire) all laid out their AI roadmaps — and across every keynote one theme dominated: AI agents are no longer optional! They’re becoming foundational infrastructure for enterprise AI.

And this wasn’t a one-off. Just two weeks ago, IBM made a similar move at THINK — launching agent builders, orchestration tools, and deploying Watsonx agents across real enterprise use cases.

Let’s break down this week’s biggest announcements: ⬇️

🔧 Tool Spotlight
Mem0 — Memory for agent workflows

One of the biggest gaps in AI agents? They don’t remember anything.

Claude plans. Cursor codes. Windsurf executes. But none of them talk — let alone share context.
Mem0 changes that. It’s an open-source memory layer for AI agents, built on the Model Context Protocol (MCP). It lets tools read and write to shared memory — locally, securely, and with long-term, structured recall.

Surprised more people aren’t talking about it. This solves a core blocker in agent workflows.
→ See Mem0 on GitHub

🧰 Stack Update
New drops: Tools, frameworks, and agent-building Infrastructure

🌀 Codex by OpenAI (Research Preview) A cloud-based engineering agent that writes features, fixes bugs, runs tests, and proposes PRs — now in research preview for ChatGPT Pro, Team, and Enterprise users.
→ Powered by codex-1, a version of OpenAI’s o3 model tuned specifically for software tasks.

🌀 Jules by Google Now in public beta, Jules is a coding agent that reads your codebase, writes tests, fixes bugs, and works asynchronously — complete with GitHub integration and even audio changelogs.
→ A strong push from Google to own agentic software development workflows.

🌀 Open Computer Agent by Hugging Face A free, cloud-hosted AI agent running on a virtual Linux machine — capable of browsing, launching apps, and running real computer tasks.
→ Open-source answer to Operator — slower, but shows where OSS agents are headed.

🌀 Strands by AWS An open-source SDK for building and deploying model-native AI agents — already powering Amazon Q, Glue, and others internally.
→ Multi-agent-ready, production-tested, and built for orchestration on AWS services.

🌀 SWE-1 by Windsurf A new family of LLMs purpose-built for engineering and DevOps. Flow-aware, production-tested, and optimized for human–agent collaboration in IDEs and terminals.
→ Engineering workflows are the next battleground — and Windsurf is going deep.

🎓 Learning & Upskilling
Sharpen your edge— top free courses this week

🟦 Google
Launched a free Agentic AI Startup School - starting June 10th. It’s a 16-day global training program with live sessions twice a week, covering everything from the basics of agentic AI to hands-on building.

🟦 Coursera
Reinforcement Fine-Tuning with GRPO covers the foundations of RL and introduces the Group Relative Policy Optimization algorithm — a powerful method for boosting reasoning in LLMs.

🟦 Hugging Face:
Released 9 free courses on AI agents, LLMs, vision models, and more. Great starting point for hands-on learning from the open-source frontier.

🟦 Coursera x Anthropic
New course on Model Context Protocol (MCP) — learn how to build an MCP server that delivers tools, prompts, and resources into real agent workflows.

🌱 Mind Fuel
Strategic reads, enterprise POVs and research

🔹 “AI or Die” — hands down the most interesting thing I’ve read this week. It’s a wake-up call to everybody still treating AI as a feature, not a foundation.

🔹 IBM 
Just released a new report on Agentic AI in Financial Services and opportunities, risks, and responsible Implementation.

🔹 PwC
Dropped one of the most grounded surveys on enterprise AI agent adoption to date. TLDR: 88% of execs are increasing budgets, and 66% already see productivity gains.

🔹 Artificial Analysis
Released their quarterly snapshot of the AI landscape.

🔹 Sequoia Capital – AI Ascent Event
Sequoia Capital hosted its annual AI Ascent event — very interesting perspective on what the future might look like.

🔹 Rekki’s CTO Borislav Nikolov on Becoming AI-Native
The big insight here: being AI-native is a mindset shift, not a feature list. And those who get it early will win the next wave. Good read!