AI agents for science

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

Apparently it’s now Common Knowledge that the em dash is a sign of AI writing.

As a a major fan of the em dash myself, I think that’s total nonsense—and I’m going to keep using it. There are way too many useful things you can do with it—this, for instance—for me to give it up to the robots.

In today’s issue…

• How AI agents are changing science

• CrewAI hosts week of events on agents

• Hippocratic raises $141 million for medical agents

• Build agents that act in code

…and more

🔍 SPOTLIGHT

Just as the science of AI agents advances, AI agents are beginning to advance science.

The past year has seen an explosive growth of interest in AI agents, particularly for augmenting—or entirely automating—human labor in enterprise settings. However, their ability to speed up difficult cognitive jobs has led AI agents to be increasingly employed in another setting: scientific research.

One obvious application which dates back to the very beginning of the large language model boom is summarizing the state of research on a given subject, a task which many scientists find to be one of the most burdensome parts of the job as the existing corpus of literature—much of it low-quality—grows exponentially. Almost immediately after models such as GPT-3.5 became available, researchers and hackers (including yours truly, as a former scientist) began using them to build simple software to scrape this vast text space and distill it into useful insights.

However, these research summarizers do not qualify as agents, as they lack any sort of planning, self-reflection, or ability to perform sophisticated actions. With the steady rise of agentic AI, a new generation of LLM-based scientific applications have aimed to automate the entire process of scientific discovery, from hypothesis generation to experimentation to manuscript production. In June of last year, Sakana AI created a stir with their “AI Scientist”, which purported to produce novel research in machine learning—though critics were quick to point out its limitations. Since then, numerous other groups have sought to create agents capable of automating machine learning or data science research, leading some AI safety researchers to monitor these systems’ potential to create self-improving AI.

While machine learning research is a tempting target for science agents due to its ability to be performed entirely in silico, scientists have also begun to apply them to the harder sciences. Many of these cases involve an inversion of peoples’ expectations of AI—instead of a human scientist directing robot workers, an AI agent designs constructs such as nanobodies against SARS-CoV-2, and a human worker creates and tests them in a lab.

With a human in the loop, however, these efforts still fall short of the Holy Grail of AI agent-powered research: a “self-driving lab” in which every step—including benchtop experiments—is executed entirely by computers. As ambitious as this sounds, researchers have begun making strides towards it. In early 2024, a pair of groups respectively debuted ORGANA and ChemCrow, sophisticated LLM agents capable of planning the synthesis of chemical compounds with given properties, then actually running them in a real lab via robotic synthesis platforms. At least one startup, inspired by their work, is seeking to commercialize similar technology.

While enterprise AI agents may seem dull to some, evoking images of spreadsheets and office drudgery, or raising fears of job displacement, scientific agents hold a unambiguously hopeful promise: drastically speeding up the entire enterprise of science, saving millions of lives through medical advances and tackling such global issues as climate change and food insecurity. If this potential is ultimately fulfilled, they could constitute AI agents’ most important contribution to humanity.

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

🤝 WITH THE AI REPORT

There’s a reason 400,000 professionals read this daily.

Join The AI Report, trusted by 400,000+ professionals at Google, Microsoft, and OpenAI. Get daily insights, tools, and strategies to master practical AI skills that drive results.

Sign up now for free and work smarter, not harder.

📰 NEWS

CrewAI to hold “AI Agent Week” tour

Leading agent framework maker CrewAI is hosting a week-long series of events dedicated to the agent field across 3 cities: San Francisco, New York, and another yet to be announced. Crew is also accepting applications for those wishing to host affiliated meetups in other locations.

Cohere launches enterprise agent platform

Enterprise AI company Cohere has launched North, an agent-focused suite for integrating LLMs into business operations, likely intended as a competitor to similar offerings by tech giants such as Microsoft and Salesforce.

Hippocratic AI raises $141 million

The series B round brought the healthcare startup’s valuation to $1.64 billion just 18 months after its founding, providing it with resources to expand its medical AI agent platform.

Google Cloud releases tools for retail agents

At the National Retail Federation’s 2025 trade show, Google Cloud announced a new suite of tools in its Agentspace platform to assist retailers in building AI agents to automate their operations and provide assistance to their customers. The release comes just two days after a similar announcement by Salesforce introducing Agentforce for Retail.

🛠️ USEFUL STUFF

Create AI agents that act in code

Freeact is a lightweight library for building agents that take action by generating code rather than less flexible formats such as JSON. While Freeact and the code it generates are both written in Python, AgentScript is a similar concept in TypeScript writing JavaScript code.

Agentic documentation workflows

An advanced form of agentic RAG by LlamaIndex, allowing users to build sophisticated, multi-step workflows on top of their enterprise documents.

Build AI sales development agents

Agent integration startup Composio introduced SDR-Kit, a set of tools for easily creating AI sales development agents that generate, reach out to, and manage leads autonomously.

💡 ANALYSIS

How AI is transforming enterprises

Box CEO Aaron Levie gives his two cents on the rapid changes that AI is driving in businesses, declaring “this is the most energized I’ve seen enterprises in nearly two decades of being in enterprise software.”

🧪 RESEARCH

A multimodal generalist GUI agent

With agents that can interact with graphical user interfaces (GUIs) all the rage right now, the authors of this paper introduce InfiGUIAgent, which uses a unique two-stage training process to achieve state-of-the-art performance for an open model of its size on several GUI navigation benchmarks.

Agentic search-enhanced large reasoning models

Search-o1 is a framework designed to enhance the capabilities of large reasoning models (LRMs) such as o1 by giving them the ability to retrieve external information using RAG, then analyze it in a separate reasoning chain before bringing it back into the main one.

Efficient tool use with Chain-of-Abstraction reasoning

Chain-of-Abstraction is a novel technique proposed by Meta researchers in this paper, which allows an LLM to generate an abstract reasoning chain linking different concepts together, then fill them in using tool calls, outperforming standard chain-of-thought and tool-augmented LLM reasoning.

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