Francesca LaBianca, Factory.AI

Clips
More from The AI Security Summit 2025

Francesca LaBianca

Head of Business Operations

Factory

Inside the rise of agent-driven development and the future of coding agents

The next phase of innovation in enterprise technology is being defined by agent-driven development — a model that merges automation with human insight to reshape the software lifecycle. As companies scale artificial intelligence into core engineering processes, the focus is shifting from productivity metrics to code quality, collaboration and governance.Across industries, teams are exploring how agents can streamline complex development environments while keeping human creativity in the loop. It’s no longer about how many lines of code an AI can generate — it’s about how efficiently teams can guide, review and deploy code with confidence

play_circle_outline Francesca shares her background in investment banking and venture capital.

play_circle_outline Ensuring Quality in AI Coding: The Crucial Role of Code Review and Human Oversight in Agent-Driven Development

Info
Transcript

Francesca LaBianca, Factory.AI

Francesca LaBianca

Head of Business Operations Factory

At The AI Security Summit in San Francisco, theCUBE’s Jackie McGuire sits down with Francesca LaBianca, head of business operations at Factory.AI, to unpack how enterprises are preparing for “agent-driven development.” LaBianca traces her path from investment banking and venture to joining Factory, then explains why solving for the full software development life cycle (SDLC) – not just code generation – is essential. She details how Factory’s agents assist across planning, coding, reviewing, testing and maintenance, and why human-in-the-loop review remains cr... Read more

explore Keep Exploring

What is your professional background and how did you come to be involved with Factory? add

What are the challenges in measuring the effectiveness of AI in coding and how does agent-driven development change the coding process? add

bolt Powered by CUBE AI

Francesca LaBianca, Factory.AI

Jackie McGuire

>> Hello, CUBE community and welcome back to the AI Security Summit in San Francisco, California. You're back with Jackie McGuire. I'm the practice lead principal analyst, and your favorite security host at theCUBE, and I am joined today by Francesca LaBianca. Francesca, welcome.

Francesca LaBianca

>> Thank you. Great to be here.

Jackie McGuire

>> You are the Head of Business Operations at Factory. Yeah?

Francesca LaBianca

>> Correct.

Jackie McGuire

>> All right, so let's start with I want to learn a little bit more about you, so tell me a little bit about your background and what led you to Factory.

Francesca LaBianca

>> Yeah, absolutely. So I started my career in investment banking and the tech media and telecom space. Shortly moved into venture. Really was compelled to work with early stage companies.

Jackie McGuire

>> I was in SVB, so very similar background.

Francesca LaBianca

>> There we go. There we go.

Jackie McGuire

>> Awesome.

Francesca LaBianca

>> And loved it. Spent all my time investing in startups across AI, cyber, software generally, but felt like I wanted to go a bit deeper and always felt like I wanted to join the team almost. I had invested in Factory at the fund I was at, which is how I got to know the founders, Matan and Eno. Really loved this space within Codegen and also really loved their approach in targeting enterprises. So after the series A, actually, Matan called me up and convinced me to join full time. So it's been about a year and almost a half now there leading up business operations, which is a little bit of everything from partnerships, BD, growth, ops, finance, marketing.

Jackie McGuire

>> Wow. That's just small portfolio of responsibility.

Francesca LaBianca

>> We're small but mighty team.

Jackie McGuire

>> Yeah, that's one of the things I love. And same exact thing, I worked at SVB for six or seven years. And same thing, caught the bug and was like, "I have to go do this startup thing." It's one thing to see the money they're raising and be excited by proxy, but it's a whole different thing to actually be in the day to day and kind of fighting that battle.

Francesca LaBianca

>> Totally.

Jackie McGuire

>> All right, so tell me a little bit about what Factory does. I know that you guys really help in the software development life cycle, so maybe kind of a TLDR on what the software development life cycle is for any of our viewers who aren't familiar and then tell us about how Factory helps.

Francesca LaBianca

>> Yeah, absolutely. So we are a platform that enables enterprises to accelerate parts of the software development life cycle with agents. Key to that is it's not just the coding piece, it's really the entire SDLC. Because if you focus on just coding, you lead to bottlenecks, other parts of the SDLC. And coding is just a small subset of what engineers spend their time on in professional environments. And so we help with everything from understanding a code change, planning the code change, coding it itself, reviewing code, testing code, and maintaining code. And that's kind of how we think about the entire SDLC all the way from an idea comes up through maintaining that code in prod.

Jackie McGuire

>> Awesome. Yeah, so one of the statistics that I heard is as people have started to use generative coding assistance. Commits have gone up like 6x, but pull requests have not gone up proportionately and the size of the commits has also gone up exponentially.

Francesca LaBianca

>> Sounds right.

Jackie McGuire

>> So we're creating thousands of additional lines of code. And humans are most likely not doing a great job reviewing all of that before it's merged into prod, right? So how does Factory help in that case?

Francesca LaBianca

>> Yeah, that's a great question and we see this a lot. I think one of the initial ways that people try to measure AI and coding and its impacts was with the lines of code, but it's not a great metric because you can create infinite lines of code, but is it ready to be merged and all the other things? Does it create other downstream problems? As you mentioned, review becomes a really critical part of this new process. And really how we think about it at Factory is you're moving from single human in an IDE writing every line of code to now being able to send a task to an agent, let it run to end to end. And that requires a fundamentally new way of working. And we call this agent-driven development and it's really you spec out a task to an agent much like you would to an engineer on your team being very thoughtful about how you do that, delegating that task, watching the agent run, seeing everything that it's doing as it pulls in context, calls tools, and then reviewing it. So the role of review becomes super important. We work with a lot of companies on our side too to create one on PR automated reviews from agents that can pick up a lot of the low-hanging fruit, but the human is still very needed in the loop of review and that role becomes only more emphasized with the greater amount of agents you incorporate.

Jackie McGuire

>> And so I've talked to a handful of venture investors over the last several months and they kind of talk about this dual mandate that the enterprise is trying to handle right now. The first is integrate AI into the enterprise, buy AI tools and find ways to help make the enterprise more productive with existing AI tools. But the other is to make the enterprise AI ready or as you refer to it as agent ready. So what does that actually mean to you and how are you seeing some of your more successful customers do that?

Francesca LaBianca

>> Yeah, great question. We talk about this a lot with our large enterprise customers, especially folks like Ernst & Young, McKinsey who are helping orgs not just usher in AI coding tools, but really think about how they transform their org to match this new wave. How can you best set up your org so that you can have agents running and they have these very clear guardrails so that their output is better? And we think about this as agent readiness and this incorporates a number of things, but the initial way companies start is very small, starting with an AGENTS.md file, which is an open source standard now amongst many companies, including OpenAI and others, Factory, that sets up rules for your agents as they run. But this extends to how your tests are set up, how your linters are set up, how your dev containers are set up, all these things that are primitives and upstream of letting an agent work in your org. And that's not just Factory specific, that's any coding agents you use within your org lead to higher quality output, not necessarily just more lines of code created, but really more like PRs merged with higher code quality.

Jackie McGuire

>> Do you think this is a bit of an inflection point? So anybody who's watching this full set of interviews has heard me say this a few times, but they always say if you ask five engineers how to solve a problem, they'll get seven or eight answers, right? And that's flown in the enterprise for a long time because engineers were magical beings and we need to be able to do our work however we want to. Do you think this is a little bit of an inflection point to start to establish more consistency and standards across how we solve problems and how we build things?

Francesca LaBianca

>> Yeah, super interesting question. I think the creativity piece will still remain, because like I mentioned, there's kind of this role at the beginning of the process, which is speccing it out. So that is where the human is very in the loop. There's creativity involved. You're telling the agent how you're thinking about a problem. It will think on new ways on how to solve that as well, and it's non-deterministic, and so that may not always be the same case, but you do still have these guardrails for how it's stylized, how it'll fit within your broader SCLC. So I do see how, in some cases, yeah, there might be much similar to how you may feel like if you use ChatGPT for writing, it all sounds the same. It will definitely standardize some of that, but there will still be that creativity, especially in the beginning with spec and at the end with reviewing the code.

Jackie McGuire

>> So what's your advice? Let's say I'm in your role at a company who's trying to make itself agent ready. What do you think are the one or two low-hanging pieces of fruit that most companies should start to tackle first in terms of getting ready for bringing agents into the enterprise other than calling you?

Francesca LaBianca

>> Yeah, first call, Factory. Secondly, I would say really setting up your tests and linters to have full coverage to make sure that they are incorporating all the right rules there is the first thing you can do and that's the first line of defense. That's no different than pre-AI coding, and that's very critical for your engineering org. But if you have those guardrails up to start, then it's super helpful along with your AGENTS.md file that agents can very clearly follow. Those are kind of the first three steps we help cover. And then there are ones as we get deeper into process.

Jackie McGuire

>> And then looking forward, we ask people this question a lot. So let's say we sit down next year at the AI Security Summit part two, what do you hope to be able to say that factories accomplish? So what's on your roadmap for the next kind of 6 to 12 months? And I know everything changes every 30 days now and the world as we know it is very uncertain, but ideally, we'll be sitting here next year. What do you want to be able to tell us?

Francesca LaBianca

>> Yeah, our mission at a high level is to bring autonomy to software engineering. And in a year's time, obviously, model capabilities will develop, agentic capabilities will develop, our products we will continue to develop. I think the greater amount of time we can save engineers to free up their time to work on the creative aspects of software engineering systems, design, building things up, coding is even the core piece that they love and not have to do the mundane things like review, and test, and documentation. That is really what we set out to achieve. And the more that we can do that at scale for enterprises is what we would love to achieve in the next year.

Jackie McGuire

>> Yeah, coming from data science, I would be more than happy to let an agent take care of the last 5% of tuning my models where it's just beating my face against a Jupyter notebook trying to make them better.

Francesca LaBianca

>> Absolutely.

Jackie McGuire

>> So also annotating my code because I'm usually terrible at that, so I would love to have an agent that actually tells people what I thought I was doing when I was writing something. So to take a look at it from the investor side, since you have great experience there, I think a lot of companies right now, because potential enterprise customers are not necessarily as well versed in AI, what guidance can you give leaders at companies looking to bring in vendors on how to tell the difference between an AI company riding a hype wave and the companies who are out there really making a difference? So as an investor, when you were trying to find companies to invest in, how do you separate the codes in the way?

Francesca LaBianca

>> Yeah, good question. I think it's all in the product feedback. Developers vote with their feet. Are they using the tool? Do they like it? Do they not? That is really the truth of it all. And I think that is how investors can parse through the noise and the hype. If you talk to someone, how much time has this saved you? Not just lines of code created, but how much has your on-call time dropped? Has your test coverage increased? These types of things. I think it all really just comes back to user feedback at the end of the day, as early as you can get that as an investor is what I would say.

Jackie McGuire

>> Yeah, awesome. So I know we're both women in tech, so I always like to get the opinion. I'm of the opinion that now is ... I mean there's arguably never been a better time to get into AI and security, and I think from the perspective of being able to use a coding assistant, I think it's dramatically lowering the barriers of entry for women, people who might be transitioning from other fields. So do you have guidance for young women who are kind of looking at this and they're a little daunted by the terminology and all of that in terms of ... I just had to kind of dive in when I was young, but what guidance do you have for women who are looking at this and saying, "How do I take advantage of this insane opportunity and start educating myself?"

Francesca LaBianca

>> Yeah, I love this question. For folks who come from a non-technical background, which is part of your question, that's myself. I have absolutely loved being a part of this whole wave. I think there's more opportunity than ever to try new tools. There's so much education out there, people are trying so many different things, posting about it on X, YouTube, the ways in which they use it, and there's never been a better time to just jump in, try things, see what works, see what doesn't. Think about how you can automate the things you're doing. We on our BizOps team, every process that we have, the first question is, can we use Droids, which are our agents, to automate this? And that creates a completely new mentality in how you work. And I think that's been something that's a lower barrier to entry. And for women jumping into AI and tech in general, there's amazing communities of women in AI. I'm part of one, actually. It's called Sapien. Shout out to the group, but it's like 200 women operators across the ICE startups and I think you'll find a lot of community in it. And I think that, that is just starting to even increase more and compound and really fun to be a part of.

Jackie McGuire

>> Did you find that your non-technical background was actually kind of a superpower when you got into this field? While it took me a while to ramp, coming from finance and transitioning into data science and learning how to write Python and all of that, that was tough. But being able to take some of the background I had from finance and apply frameworks that were completely unrelated to security and data science, things that I knew from other industries, did you find that there were some transferable things from your non-technical background?

Francesca LaBianca

>> Certainly. Certainly. And I think our team, one of our core values at Factory is embrace perspectives. And when I first joined Factory, we were a team of 10, all engineers to date, and it was interesting to bring the perspective of finance. How are we thinking about margin? How are we thinking about compute? How are we thinking about partnerships with folks like Snyk and others? You have a different type of knowledge set. And now that as we build out the tool, I'm always sending feedback in our engineering CLI channel because I use Droid for all types of non-technical use cases and non-technical use cases. They're now technical. Whether that's writing JDs to automating our RevOps stack, all these types of things. And I think there's certainly superpower in that because now we can also service a broader set of users and use cases as well.

Jackie McGuire

>> Yeah, so shout out to all the young women out there who are coming from non-technical backgrounds.

Francesca LaBianca

>> Yes.

Jackie McGuire

>> There are probably organizations who really, really need those skills and will let you learn how to ramp up in the technical roles as well.

Francesca LaBianca

>> Absolutely. We're hiring on the BizOps team of Factory, so reach out.

Jackie McGuire

>> Awesome. Francesca, thank you so much for being here today. We really appreciate it.

Francesca LaBianca

>> Thank you so much for having me. Great to chat.

Jackie McGuire

>> Thanks. From the AI Security Summit in San Francisco, California, this is Jackie McGuire for theCUBE, your source for tech news and analysis.