The Front Page of Fintech

The largest fintech community in the world. Subscribe to our newsletter to stay up to date on the latest in news opinions, and all things financial technology.

Image Description

The Front Page of Fintech

The largest fintech community in the world. Subscribe to our newsletter to stay up to date on the latest in news opinions, and all things financial technology.

Image Description

🎧 Live from HumanX: Writer CEO May Habib on Helping Enterprises Survive the AI Decade, Building Your Own LLMs, and 'Bags of Flesh'

🎧 Live from HumanX: Writer CEO May Habib on Helping Enterprises Survive the AI Decade, Building Your Own LLMs, and 'Bags of Flesh'
🎧 Live from HumanX: Writer CEO May Habib on Helping Enterprises Survive the AI Decade, Building Your Own LLMs, and ‘Bags of Flesh’
Podcast Episode · This Week in Fintech’s Podcast · 03/26/2025 · 19m

In March, we traveled to the inaugural HumanX conference in Las Vegas to meet founders, builders, and investors at the bleeding edge of AI. The conference was a great gathering of AI leaders, and we're excited for it to relocate to San Francisco next April.

You can read some of our takeaways below:

This week, Darrell Etherington and I traveled to the inaugural HumanX… | Ryan Zauk
This week, Darrell Etherington and I traveled to the inaugural HumanX conference, bringing together leaders across global AI. We met with several folks across…

While at the conference, I had the chance to sit down with May Habib, Co-Founder and CEO of Writer. If you've been around AI x FS over the last few years, or drive the 101, you've likely come across Writer.

Writer is the leading Gen-AI platform for enterprises, and has been one of the few AI companies to find repeatable, scaled success in regulated enterprises. They help large enterprises deploy secure + reliable AI applications and agents that transform mission-critical workflows.

Writer’s platform is supported by Palmyra – Writer’s state-of-the-art family of LLMs – alongside its industry-leading graph-based RAG, customizable AI guardrails, and suite of development tools. Palmyra LLMs include one of the world’s highest-benchmarked frontier models, topping leaderboards for natural language understanding and generation, as well as top specialized models in finance and healthcare.

Their customer list speaks for itself, including household names like Vanguard, Salesforce, Qualcomm, Accenture, Franklin Templeton, Intuit, Dropbox, Hubspot, Prudential, L'Oreal, and the list goes on...

When you meet May, you can see why. She is a phenomenal leader, communicator, and executor who lays out why Writer's full-stack technology and deep appreciation for how large enterprises function allows them to have such success.

Writer is fresh off a $200M Series C valuing the company at $1.9Bn. They are supported by investors including ICONIQ, Radical, Premji Invest, Adobe Ventures, B Capital, Citi Ventures, IBM Ventures, Workday Ventures, Insight, Vanguard, and others...quite a long list of strategics.

The 5 most interesting bits from our episode:

  1. On why point solutions building on top of LLMs struggle at enterprise: 'If you're outsourcing core technology, you can't have a strategic relationship/roadmap with an enterprise, let alone a financial services enterprise.'
  2. The 'Big 3' in enterprise AI: 'Data, Governance, and information security are the Big 3. Writer can actually give you access to the training data. If you don't, how do you get governance teams off their backfoot from a security perspective?'
  3. On needing to adopt AI: 'AI is not buying software. It's a transformation that requires partnership with the customer to help them figure out how to build AI Natively. If they don't, they will not survive this decade.'
  4. On the precision required in FS: 'The killer product in FS is the precision of a cell in a spreadsheet. That's what you are replacing!"
  5. On unlocking capacity in the workforce: 'I think with the tech we have today, the work enterprises do will eventually be done with 10% of current headcount. That means we have a lot of worker capacity to take market share or build new products. Anyone who thinks AI is a massive, job-destroying event has little faith in humanity. We are incredibly creative bags of flesh!'

Enjoy the show.

🎧 Live from HumanX: Writer CEO May Habib on Helping Enterprises Survive the AI Decade, Building Your Own LLMs, and ‘Bags of Flesh’
Podcast Episode · This Week in Fintech’s Podcast · 03/26/2025 · 19m

--

May Habib is the CEO and co-founder of Writer, the full-stack generative AI platform delivering transformative ROI for the world’s leading enterprises. May is an expert in natural language processing and AI-driven language generation. She has led Writer to become one of the world’s fastest-growing generative AI companies, securing its position as a Forbes 50 AI company and inclusion in the World Economic Forum's Unicorn Community. May graduated with high honors in Economics from Harvard University. She is a World Economic Forum Young Global Leader, a Fellow of the Aspen Global Leadership Network, and a recipient of Inc.'s Female Founder Award.

Writer is the full-stack generative AI platform delivering transformative ROI for the world’s leading enterprises. Its fully integrated solution makes it easy to deploy secure and reliable AI agents that automate mission-critical work. Writer’s suite of development tools is supported by Palmyra – Writer’s state-of-the-art family of LLMs – alongside its industry-leading graph-based RAG and customizable AI guardrails. Hundreds of customers like Accenture, Intuit, L’Oreal, Salesforce, Uber, and Vanguard trust Writer to transform the way they work. Founded in 2020, Writer is backed by world-leading investors, including Premji Invest, Radical Ventures, ICONIQ Growth, Insight Partners, Balderton, B Capital, Salesforce Ventures, Adobe Ventures, Citi Ventures, IBM Ventures, WndrCo, and others.

Ryan Zauk is the Host of the This Month in Fintech Podcast and Bay Area lead for the broader This Week in Fintech platform. In his day job, Ryan is an investor at OMERS Ventures, the direct investing arm of one of the world’s largest pension plans with over $130Bn in net assets. Prior to OMERS, he worked in Morgan Stanley’s Tech Investment Banking team focused on M&A and capital markets. He is based in the Bay Area.

You can find him on Linkedin or Twitter.

Transcript

[00:00:00] Ryan Zauk:: The views expressed in this podcast are the speaker's own and are not the views of this week, FinTech or any other person or entity. The content provided in this podcast is for informational purposes only and should not be construed as legal, business tax, or investment advice or recommendation, solicitation endorsement, or offering by me or anyone else for the sale subscription or purchase of securities, or for investment advisory services of any kind.

[00:00:30] Nik Milanovic: But first a quick word from our sponsor fund box. Imagine offering your customers instant access to the working capital they need right where they do business on your platform. Fund box offers embedded lending solutions that integrate directly into your platform, providing a full suite of capital products tailored to your customer's cashflow needs.

With over a decade of experience, fund box delivers customizable credit solutions, including invoice financing, lines of credit term loans, payroll protection, and more. The robust infrastructure ensures bank grade compliance and ample funding to support your platforms at scale. Join the many industry leading platforms who have partnered with funbox to enhance their offerings and drive customer loyalty and success.

Give your customers the financial flexibility they need to grow while you provide them a seamless, trusted lending experience. Visit fund box.com/partners. That's F-U-N-D-B-O x.com/partners to schedule a demo and discover how we can help you unlock financial growth for your customers. Loans and lines of credit are made by fund box or its bank partners.

[00:01:40] Ryan Zauk: Hello everyone, and welcome to today's episode of the This Month in FinTech podcast. I'm your host, Ryan Zauk, and it is great to be here, live in Las Vegas at Human X, the first ever edition of Human X at the world's leading AI conference. And I'm joined today by May Habib founder and CEO of Writer. May, how are you doing today?

[00:01:57] May Habib, Writer: I'm great, Ryan. Way better. Knowing that you're half. Lebanese.

[00:02:00] Ryan Zauk: Yeah. There's not too many of us in the Bay Area. I was just at a Te Wadi event a couple weeks ago. Well, oh, very excited. There's, there's a good group of, uh, AI builders as well. And

[00:02:07] May Habib, Writer: of course any two Lebanese get together.

And I know your cousin later.

[00:02:11] Ryan Zauk: Yes. But there was that, that couldn't hear. We, uh, we found out that Mae rode a bus with one of my cousin Shad, uh, a couple months back. Hysterical. Or a couple years ago.

[00:02:19] May Habib, Writer: No, 15 years back. 15 years

[00:02:21] Ryan Zauk: ago. Kudos to your memory. Uh, so why don't we jump in May. I think to start probably the hardest question that we ask everybody at an AI conference like this.

What exactly does writer do and what is your story and mission? It's,

[00:02:33] May Habib, Writer: it's so funny that you say that. I was just at a conference where, uh, A CIO pulls up like our blurb and like 10 other companies. I don't get it, you know, of course. I looked at it and I'm like, what? It's, seriously, it's clear. Come on. It's so obvious.

Hello. Right. But yeah, it's definitely a challenge. I, I think. What we have been able to really do to rise above the noise mm-hmm. With the enterprise is get our differentiation down to three pillars. Okay. Pillar number one is we are a complete platform writer allows you to build, to scale, to supervise AI applications and agents all in one place.

Pillar two. The core enabling technology that makes that possible? We build it, we don't outsource it.

[00:03:16] Ryan Zauk: Okay.

[00:03:16] May Habib, Writer: We're building state-of-the-art, large language models that are domain specific. We are building an integrated rag that is knowledge graph based, not vector based. So there is semantic understanding of this data, no pre-processing.

Mm-hmm. Required. So a lot of reduced engineering for, for the customer. And that last pillar is deep partnership. Because this is a tech transformation that's not about buying software, right? This is a tech transformation that requires innovation partners to really be in there with the customer, helping them figure out how to rebuild AI native.

Because if they don't, they literally will die. I don't think you will be able to survive the next decade without reinventing all of your core operations to be AI needed.

[00:04:06] Ryan Zauk: So true. Alright. A lot of threads to pull on there. We'll get to that. But first, wanna jump back to that decision where you said you're so homegrown, you're not using any of the.

Foundation models that a lot of founders have built on top of, can you walk me through that strategic decisioning at the early stages of the company to own so much of your stack, top to bottom,

[00:04:23] May Habib, Writer: A very strategic decision. Now, it didn't start off as a strategic decision. It started off as the idea for the company.

Right When we started writer it was to build foundation models. Right now for us the pitch was, here are all the cool things we can build on top of this technology, not here's an API figure out what to do with it. We in our first company, machine translation company called Cordoba, had been doing statistical machine translation.

Mm-hmm. And so it was a very natural path to the encoder decoder models, a very natural path from there to decoder only models. Mm-hmm. There was a, uh, brief time when it really felt like, you know, our first general purpose model and GPT free, there was a big gulf, right? Mm-hmm. That took us about three months to close.

Wow. And, and we have not been behind since. Yeah. Yeah. So every big step function change in the capabilities of the models we have been leading. I mean, we talked about synthetic data before. Anybody Yeah. Rent out right now Deep. Yeah. Deep sea came a year later. Right. With the same, with the same techniques.

Um, and our reasoning models now are able to do just incredibly sophisticated. Tool calling and tool use that has really powered up this next generation of functionality. Mm-hmm. And, and those were really the strategic decisions that, that we made, right? Because we build application layer mm-hmm.

Functionality for the customer. If we can't control whether that is actually correct, whether it's good, whether it's Right, right. And frankly, what's gonna be on our roadmap six months down the road. Yeah. If you're outsourcing Quartet to somebody else. You can't have a strategic relationship with an enterprise, let alone a financial services or regulated enterprise.

[00:06:09] Ryan Zauk: For sure. And now jumping off that, let's try and make this come alive with maybe a case study. Let's talk about one of those key customers of yours, especially in a regulated state. What exactly was there before state, how did writer come in and how are they doing now?

[00:06:23] May Habib, Writer: Financial services is a very big space, right?

So from

[00:06:26] Ryan Zauk: verticalized wealth

[00:06:27] May Habib, Writer: and asset management to corporate banking, to investment banking, to insurance, lots of different use cases and within, think of it as a matrix within those sub-verticals, right? The sub-functions of their teams, from sales to marketing, to ops, to back office who've got use cases right?

And. Generative ai, the killer app really is like one of the cells in that spreadsheet, right? Yeah. Um, because it, it's so specific to what you are trying to, to get done. Our customers and financial services are folks like Ally Financial. Mm-hmm. Their CIS ish is incredibly, incredibly sophisticated. Uh, folks like Prudential.

Mm-hmm. Their Chief Digital Officer. Hema is a true, true visionary, uh, leading investment banks, leading hedge funds. Mm-hmm. Um, folks like Vanguard and Franklin Templeton. Right. And it, it's the transformation and what the technology can do is a huge part of obviously the change that they see, but the leadership required to actually change how we draft S ones or mm-hmm.

How our advisors prepare for meetings with new clients or how we onboard a new, uh, customer in a corporate banking segment, or how we report on fund performance at a asset manager. Those are actually. Such huge cross-functional undertaking and in financial services we have not yet seen like this huge.

Partnership that's required between the business and it mm-hmm. To actually lead through. Right. Right. This, this next, uh, transformation, and I'll give you an example. We've got an incredible customer in the asset management space who has completely redone the process by which they do fund reporting.

Right, right. Using a agentic ai, really, really remarkable achievement. And now the C-Suite has to take. 300 person manual process. Right. On this change management journey. Right. So it takes you a year to get the technology in place. Right. Right. From a quality perspective, from a data perspective. And then there is the human side of actually now changing how we do it.

And, and that's where I think a lot of folks are, are struggling. Mm-hmm. I mean, if you have gotten to the point where you've built something that is so good, like your board is hearing about it. Right. Right. Like we have. There's still the, okay, how do we make this real? How do we actually capture the value part of it?

Mm-hmm. And I think there are so few true transformational stories in financial services right now, because even when you get past that first mountain, there's still that second big mountain to to, to climb over on the change manager. Yeah.

[00:09:11] Ryan Zauk: And then something that you mentioned there, all of those different verticals in the use cases.

Fund reporting, K-Y-C-A-M-L, account opening, S one, writing. I see probably 30. Individual rappers built on top of GPC to do just that little thing. Just that one thing. Yeah. At this conference alone, right? Yes. Yes. And so how do you differentiate, right, with this broader platform versus all of those little point solutions that might be more targeted but that go to market trade off?

[00:09:34] May Habib, Writer: We have a ton of respect for those companies. They're usually subject matter experts. Mm-hmm. Who get so excited by generative ai. There is a lot of true subject matter expertise. Matt, what we have done is bring subject matter expertise inhouse. Um, Del Soda Ya Sheva, who, um, is an incredible, incredible portfolio manager at Wellington and that State Street has joined our team to read our financial services vertical and she's hiring like crazy.

So if you're listening and wanna do this, please join. I think she does like three red eyes in a week now. Wow. So you know her, her team, our team Right. Really relishes that kind of subject matter expertise. Yeah. Uh, a gentleman by the name of Zaid Ya is doing the same thing in healthcare for us. Again, a former emergency room physician who has Oh wow.

Works, you know, in Medicaid, Medicare, right. And consulting to healthcare payers and systems. So we have done this across all the verticals where we are building solutions and. To an enterprise customer and, and there's again, lots of segments that are gonna be created here, but we sell to the global 2000 in the enterprise.

There is a huge, huge barrier to entry, right, from data to governance to InfoSec. The fact that we can give folks access to the training data of our models. They're not getting that with llama. Llama is open weight, not right, real open source. And so. How do you actually get the governance teams that are already so, so on the back foot, right.

From a security perspective mm-hmm. To trust the technology you're putting in front of them. If it's like 50 different API calls to return an answer. Right? Right. How do you lead. With a demo even of a like little wheel that spins for four minutes before you give somebody an answer. Right. Because you're doing so much third party calling.

Right. So we are really seeing those point solutions struggle to get anywhere close to the kind of scale. Mm-hmm. When we are partnering with somebody in insurance to do claim adjudication. We've gotta be able to do 700 patient files right in an hour. Right. Each one of them is literally 700 pages long.

That's a lot of token processing.

[00:11:51] May Habib, Writer: Right. That is a lot of information to be delving in to get the right answer because again. They might be switching off a 400 person operation that sits in the Philippines, and they're not gonna do it. Right. If the solution isn't performing.

[00:12:05] May Habib, Writer: so I think a lot of these folks that we see around us, the hundreds of companies here at Human X on this exhibition for, they're all gonna get the demos, they're all gonna get the validation because folks are hungry to see what's out there.

Right. You know, do you actually have the scale? Right. Right. Convert that

[00:12:22] Ryan Zauk: POC into, yeah. And so,

[00:12:23] May Habib, Writer: and then from a, from a what. What we bring to the enterprise A, a platform that's got because of its horizontal nature, right? A ton of applicability to, to your business. And by the way, our solutions are prebuilt anyway, right?

Right. So you've got a prebuilt solution that is a gentech that is ready for your data, ready for your requirements, and you don't actually want it to be fully built out all the way up to the application layer because anyway, that you won't be able to adopt it, right? Mm-hmm. So. Yes, you're right. It is a challenge, but I think you know one that.

We've been able to easily, yeah.

[00:12:59] Ryan Zauk: Yeah. And then one follow up to that, 'cause we talked to a lot of companies still in the enterprise and they've almost developed this frenemy relationship with consultants, systems integrators in the life. Do you work with any of them as channel partners in your business?

Or especially 'cause sometimes they have the ear of the CEO or a Chief digital transformation officer take play in the writer story.

[00:13:16] May Habib, Writer: Yeah. Accenture. Deloitte, P-W-C-K-P-M-G. Um, really, really strategic partnerships with all of them in. We are in, um, the process of working with the ones I didn't name, but, you know, I, I think where it really, really works is where you've got account leads at the consultancies who truly have the client's best interest at heart.

Right. And we understand, right. They're not gonna wanna be selling four projects a year. Right. They'd rather sell one project that's got legs for four years. Right. So we, we do understand the dynamics of, of their business, and I think we have successfully. Open their eyes to what it means in generative ai, right?

Mm-hmm. Like, we've got to redefine what done means because actually we need our joint client to take ownership over what this looks like, right? Mm-hmm. We no longer are in a world that's like traditional software development, go build me something, 10 guys for two years, come back and it's done. It's gotta be really, really collaborative.

So I, I do think it's challenging their business model, you know, and, and there are some account leaders right at these sis who've made. A cottage industry of. Scaring CEOs a

[00:14:26] Ryan Zauk: hundred percent. And I used to work in consulting, so resonating

[00:14:29] May Habib, Writer: for sure. Yeah. And sort of like infusing, not just fomo, but like insecurity Yes.

Right around this. Mm-hmm. And that shit rolls downhill. Yeah. Right. So they are going to AI teams who are doing a good job and being like, Hey, like tell me how what you're doing stacks up to what this consultant just told me, right. On my slide. Yeah. Yeah. And so I think, you know, when your business model is breeding insecurity mm-hmm.

That is gonna get disrupted. Yeah. Because the customer now has got, and is about to get even more, much more control over the tools, the software, the artifacts, the data that their people use. Right? Right. We are democratizing what it means to be a builder, and that our customers who lean in the most into that future, right.

As much unknowns as there are, are the ones that are doing the best and mm-hmm. You know, for a lot of them it means. Taking the meeting, but not really engaging. Mm-hmm. When it comes to the consultants. Right.

[00:15:28] Ryan Zauk: And then one thing jumping off that is also the talent and time unlock that these companies will get you recorded saying Gen AI speed is not just for gen AI companies.

There will be a capacity unlock for employees and companies need to figure out what to do with that ties. Can you elaborate on that statement and what you're seeing with your customers?

[00:15:46] May Habib, Writer: I think with the technology that exists just today, right, the kinds of AgTech flows that we are deploying just today.

Most of our organizations, the work that we do today can be done with 10% of the headcount. Right. What do you do with the other 90%? Right? We've got a lot of work to invent,

[00:16:04] May Habib, Writer: Right. To take market share, to build faster, to run higher, and I, I think that is gonna be very exciting, right? Anyone who thinks this is a massive job destroying event has very little faith in humanity.

Right. We are incredibly, incredibly creative. Right. Bags of flesh. I just heard that term. Somebody said it in a previous panel, which I like. Bags of flesh. The

[00:16:27] Ryan Zauk: irony of human X. Yeah. Yes, exactly. Bags of flesh, cloud it.

[00:16:30] May Habib, Writer: And so that is going to be what generative AI opens up, but only if you are comfortable now running at that speed.

Right? Right. There is a disruption and that disruption is how quickly do things change. And you know, at our own company even. Just yesterday I heard somebody do a pitch and they used slides and said things that we were saying a year ago. Mm-hmm. And it was like, okay, public service announcement for everybody, guys, just like, remember, we are not gonna be able to enable you fast enough.

Mm-hmm. No matter how good we get at enablement, like you have to be on the cutting edge of how you bring customers into this and, and, and it means. The same for us. Mm-hmm. Right. In terms of just the speed of that change. Yeah.

[00:17:13] Ryan Zauk: And then last two questions. Now we're wrapping up. One is a fun one. This was a very requested topic.

You have a big billboard in San Francisco. Mm-hmm. On the 1 0 1. What was the thought on getting that billboard, the ROI and what was the process? Do you get excited driving by it as well?

[00:17:26] May Habib, Writer: You know, empower people, transform work. Mm-hmm. Right. That has been our vision from when we started four years ago.

Right. That billboard's been up for a long time. Yep. And, and I think. Um, what's so exciting about now you know, where we're going with AI Studio and just how natural language right. Is the conduit to super intelligence. If you can write it, you can build it paid. I think so much of that vision now folks can really see right, the, the product that has arisen around it.

So, you know, for us it was honestly much less of like, is there ROI here, et cetera. And it was more of like, here's our stake in the ground. Mm-hmm. We are gonna put humans at the center of this revolution. It's gonna be empowering. Mm-hmm. Right? Because right now, and this was like, you know, two and a half years ago, pre-chat GPT, when that billboard went up, like the AI narrative was not a positive.

Right? Right. At all, um, at all. Now, that replaced a slogan that we had that was AI your people will love, right? Mm-hmm. As we really targeted technologists who we're building for. Business, the, the ads now, you know, the best shelters, et et cetera, say dream big, build fast. Mm-hmm. Right. And so, you know, the, the sub messages have evolved as our platform has evolved, but the overall vision of this technology being just a massively empowering one.

That's been our vision for a long time.

[00:18:52] Ryan Zauk: Great. I know we're about to wrap up, so may, where can our listeners go to find out more about writer and what you're building?

[00:18:57] May Habib, Writer: Yeah, so writer.com. Um, I'm may@writer.com. We're hiring everything everywhere. And, uh, follow me on LinkedIn. That's where I kind of share the latest.

[00:19:09] Ryan Zauk: Perfect. All right. Thank you so much.