Accenture trains 30,000 staff as agent builders

Welcome back to Building AI Agents, your biweekly guide to everything new in the AI agent field!

You may have heard of the popular thought experiment where AIs take their instructions too literally and turn the entire universe into paperclips. We aren’t at that point yet, but one overeager AI agent did turn its CEO’s computer into a paperweight

In today’s issue…

• Accenture and NVIDIA collaborate on new agent strategy

• Popular agent framework AutoGen gets overhauled

• Will agents dominate enterprises by 2030?

• Training doctors on AI-simulated patients

…and more

📰 NEWS

Accenture and NVIDIA partner to train 30,000 AI agent builders

IT consulting giant Accenture is launching Accenture NVIDIA Business Group in collaboration with the chipmaker, which will train 30,000 Accenture staff to help clients adopt agent technology.

Agent framework AutoGen gets overhaul

Microsoft’s popular AI agent framework is being given a complete redesign to incorporate feedback from users, aiming to improve its flexibility and scalability, as well as easing debugging and integration of components from other frameworks.

OpenAI introduces Realtime API

OpenAI debuted its Realtime API, allowing builders to create apps in which users converse with GPT-4o at realtime speed.

If you find Building AI Agents valuable, forward this email to a friend or colleague!

🛠️ USEFUL STUFF

Reinforcement learning vs LLM agents, and their future

An interview with Google DeepMind engineer Frederic Besse on the company’s in-house podcast, in which he discusses the difference between game-playing reinforcement learning agents and those based on LLMs, as well as future synergies between the two.

Amazon Bedrock tutorials on fine-tuning, RAG, and agent-building

A trilogy of tutorial videos by Amazon Web Services (AWS), teaching users to fine-tune models in Amazon Bedrock, connect them with RAG databases, and use them to build a simple agent.

💡 ANALYSIS

Don't sleep on single-agent systems

A contrarian take on the rise of multi-agent systems which makes the case that multi-agent systems are not always worth the additional complexity and cost they bring.

The agent workforce in 2030

A speculative piece on the role of AI agents in the economy in 6 years, arguing that they will be an integral part of virtually all aspects of companies’ operations, from customer service to sales to even high-level decision-making.

The agent economy

An overview of the emerging AI agent market, including top providers within and across verticals.

🧪 RESEARCH

AIPatient: simulating hospital patients with LLM agents

This paper aims to replace traditional patient simulation used to train doctors with LLM-based agents, outperforming prior efforts in the field on established benchmarks.

Improving agent planning using human-agent collaboration

The authors of this paper, including AutoGen creator Chi Wang, introduce Interactive Speculative Planning, a method for enhancing agents’ ability to plan by integrating them with a human interface.

Thanks for reading! Until next time, keep learning and building!

If you have any specific feedback, just reply to this email—we’d love to hear from you

Follow us on X (Twitter), LinkedIn, and Instagram