Navigating AI’s Frontier in 2025 - Grace Isford, Lux Capital

Thank you so much, Peter, to Swix, to all of the AI Engineer Summit for having me. I am so thrilled to be here. I'm Grace again, a partner at Lux Capital, and it's a pleasure to really kick off this conference and tackle a pretty tough but exciting task: the state of the AI frontier and how we navigate that in 2025.

So a little bit about Lux as we get started here,right? Lux likes to say, "We believe before others understand." We invest in frontier tech ideas that seem crazy,right? Uh, and we really like to bring sci-fi to sci-fact. In fact, we've been lucky to partner at the earliest stages with some top AI companies.

Hugging Face, I'm sure several folks know it, the GitHub for machine learning, Together AI, the open source AI cloud, Physical Intelligence, that's like a robotics software brain, and Sakana AI, that's a research lab actually in Tokyo, Japan, doing really cool evolutionary nature-inspired algorithms.

They just launched a pretty cool AI CUDA scientist last night, so go check it out. Moving a little bit forward as we think about New York City, if I get my clicker working here.

There we go. Lux is really excited to double down in New York City and AI. Lux was founded in New York City. Our first AI investment was here in 2013, and a majority of the Lux AI portfolio is actually headquartered here or as a major hub, as you can kind of see in the graph behind me.

It's also home to many of you,right? State-of-the-art research and engineering leaders and many Fortune 500 companies, several of whom you're going to hear from over the next few days. We are really bullish on the New York City opportunity, and we're really excited you guys all came to share this opportunity with us.

So when I was creating this presentation, I went back and looked at the last few years of AI,right? And all the way back really to Stable Diffusion, August 2022. And wow, I mean, look at this hockey stick,right? The last two and a half years have been crazy.

The last 18 months have been even more exponential. The progress is getting more aggressive, it's getting more impressive, and really it's getting more spread,right? It's not just OpenAI and Anthropic publishing these models. It's xAI, we just saw the Grok launch this past week.

It's Mistral, it's DeepSeek, it's many, many more. And the models are getting more performant. They're also getting more compute efficient. And as we zoom in to the current state of the world in 2025, it's off to an even wilder start,right?

If you thought the last few years were crazy, 2025 is even wilder. We saw the 500 billion Stargate project announced between the US government, OpenAI, SoftBank, and Oracle. We saw O3, OpenAI's O3,right before the start of the year where they actually exceeded human performance in the ARC AGI challenge.

We saw the DeepSeek mania,right, with DeepSeek's R1 model launching earlier this year, sending, you know, Nvidia shares tumbling down. We also saw DeepSeek go to number one, uh, in the App Store. And then, of course, just last week, we saw the France AI Summit where Macron actually launched a whole new AI initiative with France and Europe back in the game.

So you may be saying, and I think a lot of us are thinking,right, this is the AI agent moment in 2025. And I'd go as far as saying this is the perfect storm for, for AI agents. And frankly, it's easy to see why,right?

Uh, several reasoning models starting with OpenAI's O1, then O3, DeepSeek's R1, Grok's latest reasoning model this past week are outperforming human ability and in some cases even having more capabilities that we've never even seen before. We've seen the rise of test-time compute,right?

That's more compute applied at inference instead of at training. That's increasing this model performance as well. We've seen further engineering and hardware optimizations,right? Whatever you think it costs to actually train that DeepSeek model, you cannot deny it was a feat of engineering and hardware efficiency.

Inference is getting cheaper. Hardware is getting cheaper. The open source/closed source gap is getting closed between DeepSeek and LLaMA models getting more and more performant. And of course, billions of infrastructure powering all this data center and compute. We just talked about the US Stargate.

We talked about Macron and Europe. And also Japan, with SoftBank and Nvidia, has been doubling down on their data center efforts. So all this is setting this exciting groundwork for the eponymous name of our conference, Agents at Work.

And it really does feel like an exciting moment. But in reality, these AI agents aren't really working just yet,right? People are saying it's a perfect storm and I've seen a lot of thunder, I've seen a lot of great momentum, but we haven't seen that lightning strike.

Uh, and everyone I know has different definitions of agents. So for the purposes of this presentation, I'm going to define my definition as an AI agent that is a fully autonomous system where LLMs direct their own actions.

So let's give a little example of what I mean when I say an AI agent isn't working just yet. Here's a seemingly simple query on OpenAI Operator. I'm sure everyone here knows what it is. I asked it to book a flight for me to New York to San Francisco on Monday.

I'm sure it's also a route and something that many people in this room are familiar with. And in reality, it's actually kind of a complex problem,right? I need to leave after 3:00 on Monday, but I want to avoid rush hour traffic.

I want to fly United, JetBlue, or American to maximize my chance of an upgrade from economy. I want to keep it under $500 to keep under my work expense policy. I also want an aisle seat that's not too close to the bathroom.

Um, and I want to get there, you know, before midnight. So I put this into OpenAI Operator. And the first thing it did with all this information is go to Kayak, which if anyone has booked a flight before, that's a pretty frustrating experience.

And unfortunately, it did not find a flight. Uh, it wasn't able, um, it couldn't find one. It didn't even seem to look for United or American. Second try, tried again. Uh, this is Skyscanner this time, which is slightly better.

Um, and it did actually find a flight, but it found one that had a lot of traffic. Uh, 5:30 JFK, for those who live in New York, that is a tough traffic time. Um, and ultimately, I also couldn't even pick my seat.

So it didn't really work out based on my prompts, uh, earlier. So what does this all mean,right? Why do these AI agents not work? I think we so often talk about hallucinations and fabrications and AI models kind of going sideways.

We don't talk enough about these tiny cumulative errors that add up,right? Uh, there's a lot of little errors that we see with this whole model, and I'm going to go through a few. It's not an exhaustive list, but it's a sense of some of the things you might run into as you're building these AI agents.

First, decision error. It chooses the wrong fact,right? I may book a flight with AI, but it may book it to San Francisco, Peru instead of San Francisco, California. The model could overthink or exaggerate and do a few other things as well.

Second, implementation error. The wrong access or integration. On the prior slide with my Skyscanner, I actually had to enter CAPTCHA, and that messed up a little bit of the flow. You also could get locked out of a critical access to an important database, and ultimately that AI agent isn't going to work anymore.

Third, heuristic error. The wrong criteria. Unfortunately, the model didn't acknowledge best practice of allowing enough time for JFK. In fact, it didn't even ask where I was coming from, Manhattan, Brooklyn, or beyond. And that could really affect the traffic you're going to get and ultimately if I even make that flight at 5:30 PM.

And fourth, taste error. The wrong personal preferences. For those who know me well, I'm actually a pretty spooked flyer, and I do not like flying Boeing 737 Maxes. If AI books that, you know, I did not put it in the prompt earlier, but if it did book that, I would be very unhappy and I would not get on that plane.

And then there's kind of a fifth, more nebulous error,right? It's a little bit of this perfection paradox,right? We are doing things so magical with AIright now, yet we're getting frustrated when O1 thinks too long or when Operator moves at the speed of a human.

Even if the agent gets itright on the first try, often they're inconsistent and unreliable, leading to really underwhelming our human expectations about the whole experience. Here's another visual of kind of these complex systems where each of these cumulative errors really compound,right?

Two simple agents, one that's 99% accuracy, one that's 95% accuracy to start. Pretty impressive agents at the beginning, but over 50 consecutive steps, you actually realize a pretty big disparity here. There's actually a 50% difference after 50 tasks between the 95 and the 99, and that 99% agents actually down to 60%.

The point here is that something simple like booking a flight is actually really complex in nature when all these tiny cumulative errors add up. And they get even more amplified in a complex multi-agent system with multi-step tasks. So how do you, as all these amazing VPs of AI, these leaders of AI in the room, optimize a complex agent, taking into account all of these really difficult queries to consistently and reliably make theright decision?

The truth is it's hard.

But that hasn't stopped us before, and there is hope. So I thought I would run through some of the best practices that we're seeing building AI agents today and five strategies we can all think about to help mitigate a lot of these cumulative errors.

Let's dive in. First, data curation. How do we make sure an AI agent has the information it needs? Data is messy. It's unstructured. It's in silos. It's everywhere. It's not just web and text data now too. It's design data, it's image data, it's video data, it's audio data.

It's the data in your sensors and your warehouse if you're in the manufacturing world. It's even the agent data that your data, your agent is producing in real time. Think about curating proprietary data, the data the AI agent generates, and ultimately even the data you're using in your model workflow for quality control.

Data is your best asset, and curation is key to making it more effective. Data also isn't static anymore. How do you design an agent data flywheel from day one so every time a user uses the product, it automatically improves in real time and at scale?

A simple example back to our flight example is getting a curated data set of all of Grace's travel preferences, including the 737 Max and all my airline preferences. Or even say we run that agent over time and book many flights, how do we recycle back that content and adapt to my own preferences in real time?

Second, the importance of evals. How do we collect and measure a model's response? How do we choose the correct answer? This has long been important in machine learning and AI and really understanding what'sright versus wrong. You know, it's pretty simple in verifiable domains where there's a clear yes or no answer.

Like math, like science. Here's actually the Grok 3 benchmarks just up here where you saw they did all verifiable benchmarks in math and sciences. But how do we set up evaluations for non-verifiable systems where there aren't clear yes or no answers?

Like, will Grace like this plane seat based on her preferences? And how do we collect those signals too? We also saw another example of an eval debate over the weekend with Deep Research,right? We have an OpenAI Deep Research product, one from Perplexity, one from Gemini as well.

And there are multiple versions of this same product. The evals here really depend on the eye of the beholder,right? Which one is better for everyday research versus VC market research versus scientific or academic research? We have to keep an eye on collecting those signals.

We need to know and collect human preferences, and we need to build evals in a way that is truly personal. Sometimes the best eval is just trying out the agent yourself and vibes based on your needs with no number or leaderboard telling you what to do.

Third, scaffolding systems. How do we ensure when one error occurs it doesn't have a cascading effect throughout the organization? Ramp, a Lux Portfolio company, has done a great job with this, and I know Rahul is speaking tomorrow, so go check them out.

When Ramp launches a new applied AI feature and it fails, there's infrastructure logic to ensure that doesn't have a cascading effect across the agentic system and also across all of Ramp production infrastructure. We can mitigate scaffolding by building a complex compound system of how all these things work together and sometimes even bringing a human back in the loop.

For reasoning models, this gets even more interesting and important. How do we adapt the scaffold to stronger agents that self-heal and grow? An agent that realizes they're wrong and actually tries to correct their own path, or an agent that's not sure and then needs to break execution to get it back on track.

Back to our travel example again. Could we add a checkpoint for this AI agent to verify the timing for traffic or maybe steer it back in theright direction?

Fourth, user experience or UX is the moat that matters, and that's how our AI agents become better co-pilots. AI apps today are all using the same models. Foundation models are the fastest depreciating asset class on the marketright now.

GPT wrappers are cool. UX really does make a difference for those who reimagine product experiences and really deeply understand the user workflow and really promote that beautiful, elegant human-machine collaboration. Right, here's a few concrete examples. Back to the Deep Research,right?

Asking clarifying questions to make sure it fully got the picture of what I'm trying to accomplish. Like Windsor from Codium, understanding the UX or the psyche of that developer really on a more fundamental level to predict their next step.

Like Harvey in the legal world, integrating seamlessly with the legacy systems to really create real ROI for a practicing lawyer. If you think about all the major AI apps today and categories like coding, like customer support, like sales, these all are using the same models again,right?

And it's truly the UX and the product quality that makes any one company stand out. At Lux, we're really excited about the new AI frontier. Companies who have proprietary data sources and who know the workflow of their user really well, like robotics, like hardware, like defense and manufacturing, like the life sciences.

You know, how do we take a company where they take their proprietary data source, they know the workflow of a biologist or a defense contractor or a chemist, and truly create a magical experience for that end user?

Fifth and finally, how do we build multimodally? You know, we're not just multimodal anymore. We're multimodal. There's new modalities where we can truly reimagine and create a 10x user personalized experience. I am so sick and tired of the chatbot as an interface, and I know there's so many more exciting things we can do with our AI agents to really make them more human,right?

How do we make AI more human? How do we add eyes and ears, nose, a voice? We've seen really incredible improvements in voice over the last year. It's getting pretty scary good. Lux actually has an investment in the smell space called Osmo that's digitizing the sense of smell.

And what about touch,right? How do we instill a more human feeling and sense of embodiment with robotics? I'll go as far to even talk about memories,right? How do we make AI truly personal and know you on a much deeper level than it does today?

Doing all of this reframes what perfection is to a human. And even if that agent is inconsistent, it's unreliable, the visionary nature of the product exceeds all expectations. It's something new. And on the slide behind me, you'll see TLDraw.

That's an amazing Lux Portfolio company. And I think they've done a great job really reimagining the visual canvas,right? Implementing AI through brush strokes. They have a cool thing called TLDraw Computer where you can actually combine a bunch of these cool AI models in tandem and not even know you're working with a large language model in the background.

So really strive to build multimodally. So in summary, we tackled a lot today, but we're at the perfect storm today for AI agents. But unfortunately, that lightning hasn't struck yet, and AI agents are not going to happen overnight.

Cumulative errors add up. We see wrong answers, wrong preferences, wrong criteria, all these wrong human expectations that abound when you're building these systems. Data curation, evals, and scaffolding are all tools you can use to help mitigate a lot of these challenges.

And really, please think bigger. UX, multimodality, innovative product experience that truly set the workflow and the vision apart. And I'm so excited to see what all of you build and really excited to continue this conversation over the next few days.

Thank you guys so much and look forward to talking to you throughout the conference. Thank you.