Tips for building AI agents

Summary

In a recent discussion, Anthropic team members debate the hype surrounding AI agents, emphasizing their definition, utility, and future prospects. They clarified the distinction between workflows and agents, noting that the latter offers more autonomy through iterative learning and decision-making processes. The conversation highlights both overhyped and underhyped aspects of AI agents, exploring their practical applications in coding, search, and business automation. The team also shares personal experiences and offers insights for developers on making AI systems more effective and future-proof.

Highlights

• AI agents allow for autonomous decision-making, unlike traditional workflows. 🎯
• There's more hype around consumer agents than is presently justified; business applications appear more viable. 📈
• Verification in coding agents highlights their potential for high-impact development tasks. 💻
• Successful agent integration in business operations could exponentially increase efficiency. 🚀
• Testing and iterating are essential in developing future-proof agent technologies. 🔄

Key Takeaways

• AI agents are distinct from workflows as they involve autonomous decision-making, which sets them apart. 🚀
• Despite the hype, agents are not always the best solution for every task, especially consumer-focused ones. Instead, simpler systems can suffice. 🤔
• There's a promising future for agents in automating business procedures, potentially revolutionizing scalability. 🌟
• Devs should focus on creating testable processes and be ready for the models to evolve, enhancing capabilities over time. 🔄
• Understanding the model's needs and putting oneself in its place is crucial for effective agent design. 🔍

Overview

Anthropic’s team delves into the distinctions between traditional workflows and AI agents, emphasizing the latter's ability to operate with autonomy. During their discussion, they highlight the current misconceptions about agents being universally applicable, advocating instead for their focused use in appropriate contexts, mainly within business automation rather than consumer services.

With coding agents leading the forefront, they possess robust verification features that many other fields lack, indicating a competitive edge in scaling productive capabilities. Anthropic suggests that agents can handle repetitive tasks efficiently, a potential game-changer for businesses looking to maximize operational output.

Developers are encouraged to adopt a forward-thinking approach, using current technology as a foundation for evolving AI capabilities. The team stresses the importance of understanding the needs of AI models and incorporating innovative problem-solving techniques to foster more robust, scalable solutions.

Chapters

• 00:00 - 00:30: Introduction The introduction chapter discusses the current hype around consumer agents and questions their efficacy. Alex, Erik, and Barry from Anthropic's Claude Relations, research, and Applied AI teams respectively, are featured. They hint at discussions around simplifying processes like vacation booking, contrasting manual with agent-assisted methods, and introduce the topic of building effective agents, asking the fundamental question of what an agent is.
• 00:30 - 01:30: Defining AI Agents The chapter 'Defining AI Agents' discusses the various definitions of AI agents and addresses why developers or individuals involved in AI development should care about these distinctions. Erik mentions that the blog post explores the idea that not everything should be considered an agent. He emphasizes the importance of differentiating between workflows and agents, suggesting that workflows involve multiple linked Large Language Model (LLM) calls, rather than just a single call.
• 01:30 - 03:00: Distinguishing Agents from Workflows The chapter "Distinguishing Agents from Workflows" discusses the concept of agents as autonomous entities that use a language model (LLM) to determine how many steps are needed to complete a task. These agents can continuously loop until they find a resolution, which can be applied to various tasks such as customer support or code iteration. The essential characteristic of an agent, as described, is its ability to autonomously decide its actions and fate.
• 03:00 - 05:00: Evolution of AI Models and Agents This chapter discusses the evolution of AI models and agents, highlighting the shift from predefined paths to more autonomous systems. The conversation covers the distinction between workflows, which are more structured and follow a fixed number of steps, and agents, which are more autonomous. The development of this distinction is attributed to extensive conversations with customers, collaboration with various teams, and experimentation. Barry is asked to elaborate on creating the divide between workflows and agents.
• 05:00 - 07:00: Agentic Prompts vs. Workflow Prompts The evolution of AI models and teams has led to the distinction between agentic and workflow prompts. Teams have moved from using a single language model (LLM) to utilizing multiple LLMs, which can even orchestrate themselves. This has resulted in the recognition of two distinct patterns: workflows pre-orchestrated by code, and agents that operate more autonomously.
• 07:00 - 10:00: Real-life Agent Implementations and Experiences The chapter explores the growing prevalence and capabilities of real-life agent implementations as models and tools improve.
• 10:00 - 13:00: Overhyped and Underhyped Aspects of Agents This chapter discusses the concept of agent workflows, particularly at the level of prompts. It details a step-by-step process where the output of one prompt is fed into the next in a linear fashion, ensuring a fixed sequence of steps is followed. This structured process often involves additional checks through code to validate intermediate results, ensuring each step's output is satisfactory before moving on to the next.
• 13:00 - 17:00: Future of AI Agents The chapter discusses the evolving role of AI agents, highlighting their capability to follow open-ended prompts compared to traditional specific prompts. Traditional prompts are described as linear, taking a user input to transform it into a specific output, such as categorizing a question into predefined categories. In contrast, AI agents work with open-ended prompts, which often allow them to use a variety of tools to respond more dynamically and flexibly to user inputs. This flexibility represents the future of AI, enabling more nuanced and sophisticated agent interactions.

Tips for building AI agents Transcription

• 00:00 - 00:30 - I feel like agents for consumers are like fairly over hyped right now. - Okay, here we go. Hot take. - Trying to have a agent fully book a vacation for you, almost just as hard as just going and booking it yourself. - Take one. Mark. - Today we're going behind the scenes on one of our recent blog posts, Building Effective Agents. I'm Alex, I lead Claude Relations here at Anthropic. I'm Erik, I'm on the research team at Anthropic. - I'm Barry, I'm on the Applied AI team. - I'm gonna kick us off here. For viewers just jumping in, what's the quick version of what an agent actually is?
• 00:30 - 01:00 I mean, there's a million definitions of it and why should a developer or somebody that's actually building with AI care about these things? Erik, maybe we can start with you. - Sure. Yeah, so I think something we explored in the blog post is that first of all, a lot of people have been saying everything is an agent, referring to almost anything more than just a single LLM call. One of the things we tried to do in the blog post is really kind of separate this out of like, hey, there's workflows, which is where you have a few LLM calls chained together.
• 01:00 - 01:30 And really what we think an agent is is where you're letting the LLM decide sort of how many times to run. You're having it continuing to loop until it's found a resolution. And that could be, you know, talking to a customer for customer support, that could be iterating on code changes. But something where like you don't know how many steps it's gonna take to complete, that's really sort of what we consider an agent. - Interesting, so in the definition of an agent, we are letting the LLM kind of pick its own fate and decide what it wants to do, what actions to take
• 01:30 - 02:00 instead of us predefining a path for it. Exactly. - It's more autonomous, whereas a workflow you can kind of think of it as like, you know, yeah, a workflow or sort of like it's on rails through a fixed number of steps. - I see. So this distinction, I assume this was the result of many, many conversations with customers and working with different teams and even trying things ourself. Barry, can you speak more to maybe what that looks like as we got to create this divide between a workflow and an agent?
• 02:00 - 02:30 And what sort of patterns surprised you the most as you were going through this? - Sure. Honestly, I think all of this kind of evolved as like model got better and like teams got more sophisticated. We both worked with a large number of customers who are very sophisticated and we kind of went from having a single LLM to having a lot of LLMs and like eventually having LLMs orchestrating themselves. So, you know, one of the reasons why we decide to create this distinction is because we started to see these two distinct patterns where you have workflows that's pre orchestrated by code and then you also have, you know, agent,
• 02:30 - 03:00 which is like a simpler but complex in in other sense, like different shape that we're starting to see. Really, I think like as the models and all of the tools start to get better, you know, agents are becoming more and more prevalent and more and more capable. And that's why we decided hey, this is probably a good time for us to give a formal definition. - So in practice, if you're a developer implementing one of these things, what would that actually look like in your code as you're starting to build this? Like, the differences between,
• 03:00 - 03:30 like maybe we actually go down to like the prompt level here. What does an agent prompt look like or flow and what does a workflow look like? - Yeah, so I think a workflow prompt looks like, you have one prompt, you take the output of it, you feed it into prompt B, take the output of that, feed it into prompt C, and then you're done. Kind of, there's this straight line fixed number of steps. You know exactly what's gonna happen and maybe you have some extra code that sort of checks the intermediate results of these and make sure they're okay.
• 03:30 - 04:00 But you kind of know exactly what's gonna happen in one of these paths. And each of those prompts is sort of a very specific prompt, just sort of taking one input and transforming it into another output. For instance, maybe one of these prompts is taking in the user question and categorizing it into one of five categories so that then the next prompt can be more specific for that. In contrast, an agent prompt will be sort of much more open-ended and usually give the model tools
• 04:00 - 04:30 or multiple things to check and say, Hey, here's the question, and you can do web searches or you can edit these code files or run code and keep doing this until you have the answer. - I see. So there's a few different use cases there. That makes sense as we start to arrive at like these different conclusions. I'm curious as we've now kind of covered at a high level how we're thinking about these, these workflows and agents and talking about the blog post.
• 04:30 - 05:00 I want to dive even further behind the scenes. Were there any funny stories Barry, of wild things that you saw from customers that were interesting or just kind of far out there in terms of how people are starting to actually use these things in production? - Yeah, this is actually from my own experience like building agents. I joined like about a month before the Sonnet v2 refresh and one of my onboarding tasks was to run OSWorld, which was a computer use benchmark. And for a whole week me and this other engineer
• 05:00 - 05:30 were just staring at agent trajectories that were counterintuitive to us and then well, you know, we weren't sure why the model was making the decision it was, given the instructions that we would give it. So we decided we're gonna act like Claude and, you know, put ourselves in that environment. So we would do this really silly thing where we close our eyes for a whole minute and then we're like blink at the screen for a second and we close our eyes again and just think, well, I have to write Python code to operate in this environment, what would I do? And suddenly it made a lot more sense.
• 05:30 - 06:00 And I feel like a lot of agent design comes down to that. It's like, there's a lot of context and a lot of knowledge that the model maybe does not have and we have to be empathetic to the model and we have to make a lot of that clear in the prompt, in the tool description, and in the environment. - I see, so a tip here for developers is almost to act as if you are looking through the lens of the model itself in terms of like, okay, what would be the most applicable instructions here? How is the model seeing the world, which is very different than how we operate as a human,
• 06:00 - 06:30 I guess, with additional context. Erik, I'm curious if you have any other stories that you've seen. - Yeah, I think actually my, in a very similar vein, I think a lot of people really forget to do this. And I think maybe the funniest things I see is that people will put a lot of effort into creating these really beautiful, detailed prompts and then the tools that they make to give the model are sort of these incredibly bare bones, you know, no documentation,
• 06:30 - 07:00 the parameters are named A and B. And it's kind of like, oh, an engineer wouldn't be able to, you know, work with this. - Right. - You know, work with this as if this was a function they had to use. 'Cause there's no documentation, how can you expect Claude to use this as well? So kind of it's that lack of like putting yourself in the model's shoes. And I think a lot of people when they start trying to use tool use and function calling, they kind of forget that they have to prompt as well and they think about the model just as this,
• 07:00 - 07:30 you know, a more classical programming system. But it is still a model and you need to be prompt engineering in the descriptions of your tools themselves. - Yeah, I've noticed that. It's like people forget that it's all part of the same prompt. Like, it's all getting fed into the same prompt in the context window and writing a good tool description influences other parts of the prompt as well. So that is one aspect to consider. Agents is this kind of all the hype term right now, a lot of people are talking about it
• 07:30 - 08:00 and there's been plenty of articles written and videos made on the subject. What made you guys think that now is the right time to write something ourselves and talk a little bit more about the details of agents? - Sure, yeah. I think one of the, you know, most important things for us just to be able to explain things well. I think that's a big part of our motivation, which is we walk into customer meetings and everything is referred to as a different term even though they share the same shape.
• 08:00 - 08:30 So we thought, you know, it would be really useful if we can just have a set of definition and a set of diagrams and code to explain these things to our customers. And you know, we are getting to the point where the model is capable of doing a lot of the agentic workflows that we're seeing. And that seems like, you know, the right time for us to, you know, have some definitions or just to make these conversations easier. - Yeah, I think for me, I saw that there was a lot of excitement around agents, but also a lot of people really didn't know
• 08:30 - 09:00 what it meant in practice and so they were trying to bring agents to sort of any problem they had, even when much simpler systems would work. And so I saw that as one of the reasons that we should write this as, guide people about how to do agents but also where agents are appropriate. And that you shouldn't go after a fly with a bazooka. - I see, I see. So that was a perfect parlay into my next question here. There's a lot of talk about the potential of agents and every developer out there and every startup and business is trying to think about
• 09:00 - 09:30 how they can build their own version of an agent for their company or product. But you guys are starting to see what actually works in production. So we're gonna play a little game here. I want to know one thing that's overhyped about agents right now and also one thing that's underhyped, just in terms of implementations or actual uses in production or potentials here as well. So Erik, let's start with you first. - I feel like underhyped is things that save people time, even if it's a very small amount of time.
• 09:30 - 10:00 I think a lot of times if you just look at that on the surface, it's like, oh, this is something that takes me a minute, and even if you can fully automate it it's only a minute. Like, what help is that? But really that changes the dynamics of now you can do that thing a hundred times more than you previously would. So I think I'm like most excited about things that if they were easier could be really scaled up. - Yeah, I don't know if this is necessarily related to hype, but I think it's really difficult to calibrate right now, like where agents are really needed. I think there's this intersection that's a sweet spot
• 10:00 - 10:30 for using agent and that's a set of tasks that's valuable and complex, but also maybe the cost of error or cost of monitoring error is relatively low. That set of tasks is not super clear and obvious, unless, you know, we actually look into, the existing processes. I think coding and search are two pretty canonical examples where agents are very useful. Like, take search as example, right? Like, you know, it's a really valuable task.
• 10:30 - 11:00 It's very hard to do deep, iterative search, but you can always trade off some precision for recall and then just get a little bit more documents or a little bit more information than is needed and filter it down. So we've seen a lot of success there with agentic search. - What does a coding agent look like right now? - Coding agents I think are super exciting because they're verifiable, at least partially. You know, code has this great property that you can write tests for it and then you edit the code and either the tests pass or they don't pass. Now, that assumes that you have good unit tests,
• 11:00 - 11:30 which I think, you know, every engineer in the world can say like, we don't. - Yeah. - But at least it's better than a lot of things. You know, there's no equivalent way to do that for many other fields. So this at least gives a coding agent some way that it can get more signal every time it goes through a loop. So, you know, if every time it's running the tests again it's seeing what the error or the output is, that makes me think that, you know, the model can kind of converge on the right answer by getting this feedback. And if you don't have some mechanism
• 11:30 - 12:00 to get feedback as you're iterating, you're not injecting any more signal, you're just gonna have noise. And so there's no reason without something like this that an agent will converge to the right answer. - I see. So what's the biggest blockers then in terms of improving agentic performance on coding at the moment? - Yeah, so I think for coding, you know, we've seen over the last year, like on SWE-bench, results have gone really from very, very low to, I think you know, over 50% now,
• 12:00 - 12:30 which is really incredible. So the models are getting really good at writing code to solve these issues. I feel like I have a slightly controversial take here that I think the next limiting factor is gonna come back to that verification. That, it's great for these cases where we do have perfect unit tests and that's starting to work, but for the real world cases we usually don't have perfect unit tests for them. And so that's what I'm thinking now, finding ways that we can verify
• 12:30 - 13:00 and we can add tests for the things that you really care about so that the model itself can test this and know whether it's right or wrong before it goes back to the human. - I see. Making sure that we can embed some sort of feedback loop into the processes itself. - Exactly. - The right or wrong. Okay. What's the future of agents look like in 2025? Barry, we're gonna start with you. - Yeah, I think that's a really difficult question. This is probably not a practical thing, but one thing I've been really interested in just
• 13:00 - 13:30 how a multi-agent environment would look like. I think I've already shown Erik this, I built an environment where a bunch of Claudes can spin up other Claudes and play Werewolf together. And it is a completely- - What is Werewolf? - Werewolf is a social deduction game where all of the players are trying to figure out what each other's role is. It's very similar to Mafia, it's entirely text-based, which is great for Claude to play in. - I see, so we have Claudes, multiple different Claudes playing different roles within this game, all communicating with each other.
• 13:30 - 14:00 - Yeah, exactly. And then you, you see a lot of interesting interaction in there that you just haven't seen before and that's something I'm really excited about as you know, very similar to how we went from single LLM to multi-LLM, I think by the end of the year we could potentially see us going from agent to multi-agent. And there are some I think interesting research questions to figure out in that domain. - In terms of how the agents interact with each other, what is this kind of emergent behavior look like in that front as you coordinate between agents doing different things.
• 14:00 - 14:30 - Exactly. And just whether this is actually gonna be useful or better than a single agent with access to a lot more resources. - Do we see any multi-agent approaches right now that are actually working in production? - I feel like in production we haven't even seen a lot of successful single agents. - Okay, interesting. - But, you know, this is kind of a potential extension of successful agents with the, I guess improved capabilities of the next couple of generations of models.
• 14:30 - 15:00 Yeah, so this is not advice that everyone should go explore voting agent environment. It's just, I think, you know, to understand the model's behavior, this provides us with a better way to understand model behaviors. - I see. Okay, Erik, what's the future of agents in 2025? - Yeah, I feel like in 2025 we're gonna see a lot of business adoption of agents, starting to automate a lot of repetitive tasks and really scale up a lot of things that people wanted to do more of before but were too expensive. You could now have 10X or 100X how much you do with these things.
• 15:00 - 15:30 I'm imagining things, you know, every single pull request in triggers a coding agent to come and update all of your documentation. Things like that would be cost-prohibitive to do before. But once you think of agents as sort of almost free, you can start doing these, you know, adding these bells and whistles everywhere. I think maybe something that's not gonna happen yet, going back to like what's overhyped? I feel like agents for consumers are fairly overhyped right now. - Okay, here we go. Hot take. - 'Cause I think that, we talked about verifiability.
• 15:30 - 16:00 I think that for a lot of consumer tasks it's almost as much work to sort of fully specify your preferences and what the task is as to just do it yourself, and it's very expensive to verify. So trying to have an agent fully book a vacation for you, describing exactly what you want your vacation to be and your preferences is almost just as hard as just going and booking it yourself. And it's very high risk. You don't want the agent to go actually go book a plane flight without you first accepting it.
• 16:00 - 16:30 - Is there a matter of maybe context that we're missing here too from the models being able to infer this information about somebody without having to explicitly go ask and learn the preference over time? - Yeah, so I think that these things will get there, but first you need to build up this context so that the model already knows your preferences in these things and I think that takes time. - I see. - And we'll need some stepping stones to get to bigger tasks like planning a whole vacation. - I see. Okay. Very interesting. Last question, any advice that you'd give to a developer
• 16:30 - 17:00 that's exploring this right now in terms of starting to build this or just thinking about it from a general future-proofing perspective that you can give? - I feel like my best advice is, make sure that you have a way to measure your results. 'Cause I've seen a lot of people will go and sort of build in a vacuum without any way to get feedback about whether their building is working or not. And you can end up building a lot sort of without realizing that either it's not working or maybe something much simpler would've actually done just as good a job.
• 17:00 - 17:30 - Yeah, I think very similarly, you know, starting as simple as possible and having that measurable result as you're, you know, building more complexity into it. One thing I've been really impressed by is just, I work with some really resourceful startups and they can do everything within one LLM call and the orchestration around the code, which will persist even as the model gets better, is kind of their niche. And I always get very happy when I see one of those because I think they can reap the benefit
• 17:30 - 18:00 of future capability improvements. And yeah, I think realistically, you know, we don't know what use case will be great for agents and the landscape's gonna shift, but it's probably a good time to start building up some of that muscle to think in the agent land, just to understand that capability a little bit better. - Yeah, I think I wanna double-click on something you said of, being excited for the models to get better. I think that if you look at your startup or your product and think, oh man, if the models get smarter,
• 18:00 - 18:30 all of our moat's gonna disappear. That means you're building the wrong thing. Instead you should be building something so that as the models get smarter, your product gets better and better. - Right. That's great advice. Erik, Barry, thank you guys. This is Building effective agents. Thank you. - Thanks.