Building Real-Time Financial APIs: Inside Benzinga with Tommy Cotter

GM, GM! Welcome to what's going to be another fantastic episode of DevNTell. So, if you didn't know, DevNTell is a 30-minute podcast held every week allowing founders, hackers, and anyone in between the opportunity to come on the show and showcase what they built. And today, I'm ecstatic to welcome Tommy Cotter, who is the Director of Products at Benzinga. So, if you don't know, Benzinga is a financial media and data technology platform focused on delivering actionable news, market insights, trading tools, and real-time historical financial data for users. So, if you stick around for today's episode, you'll get to meet Tommy, learn all about Benzinga, and how you can get started using it today. All right, let's get into it.

[Music]

Hello, welcome to the show, Tommy. Ecstatic to have you on today, man.

Yeah, thanks for having me on, Narb. Excited to be here.

Yeah, yeah, for sure. And, yeah, I was telling you kind of before we started, Benzinga is kind of entrenched in all basically the major trading platforms out there, even if folks don't realize. But yeah, really excited for our audience to kind of learn more about the platform and the different ways they can plug into that data. But before that, would you just like to give an introduction about yourself?

Yeah, yeah, totally. So, I basically handle all things APIs at Benzinga. And like you said, Benzinga, we are a financial media company, so we write stock market news for retail investors. And aside from the news, we've got a suite of 30 API products that we've built, and we've built some bespoke solutions, we've built some generic data solutions that are redistributed widely. But yeah, that's kind of what I do. I build APIs and help push them out into the world.

Amazing. And yeah, I mean, APIs, they've always been kind of the bread and butter of developers, but I think even now more so, with the demand that's going to be rising from AI agents and whatnot. But just curious, how did you find yourself in the technology space? Like what is kind of the origin story behind Tommy here?

Yeah, I've got a pretty niche and non-traditional route for how I got here, but yeah, I studied mechanical engineering in college. Didn't really know what I wanted to do, and I went to school in Metro Detroit at Oakland University. And it's not the Oakland, California, it's Oakland Metro Detroit, but yeah, so I didn't know what I wanted to do, went there for basketball, studied mechanical engineering, and then I, like most students, went right down the street and got a job in auto. And so I worked for Fiat Chrysler for a couple of years before COVID and during COVID. And then I met the VP of Licensing at Benzinga, and he was looking for some help on the product side, and brought me on. And I'm super gracious to him and Benzinga for that. And the rest is history.

Amazing. And would you say kind of that basketball, the sporting background kind of helped set some of the foundations for your career later on?

Totally. Yeah, I mean, sports is, in my opinion, the best way for young people to learn. You learn to stay organized, diligent, work in a team setting, you kind of have that urgency, that sense of urgency that you operate with in your job and in your life in general. So yeah, I'm really glad that I decided to play basketball in my youth and into my early adulthood.

Yeah, sport, you can't go wrong with sport. There are so many great attributes you can kind of pull from that into whatever you do in your day-to-day. But amazing, man. And yeah, a lot of folks sometimes that come on the show don't really have the most traditional background and end up being amazing people in the tech space. So it seems like you're no different. And I guess for folks who might not be familiar with what a Director of Products might do on a day-to-day basis, could you kind of take them through what your typical day-to-day might look like?

Yeah, so a lot of our product development and our roadmap is kind of dictated by what our clients want. So basically, what I do is I interface with a lot of these big brokerages, big retail platforms, try and figure out what type of demand they're seeing, what type of interest their users want, and then either steer them towards a solution that we might have already built out or kind of brainstorm on how we can fill the gap via a new product. And if we make the decision to go out and build that product, you know, I kind of map out and scope out what's going to be required in order to get it done.

Awesome. Yeah, I bet you must have some very interesting conversations with a lot of different folks. But and I suspect you might have had a lot of inward interest, whether from the retail space or enterprise space, on the AI front. I mean, we can't help it, AI is everywhere. It's consuming everything, from timelines to APIs themselves. Payments is another big developing space. But I guess from your mind, how is Benzinga and I guess yourself kind of thinking about AI and how Benzinga is going to kind of work in that space?

I mean, AI, we've been thinking about it every day since ChatGPT, and even a little bit before. But I mean, yeah, it's really transformed the way that we can, from the data side, the way that we can build products and ensure accuracy and everything in between. I mean, historically, these big data licensing shops like Refinitiv, S&P Global, and FactSet, they had massive data teams. And they still do. Those big conglomerates still have massive data teams, like tens of thousands of people. And what's interesting and unique is that a small company like Benzinga, I say we're small relatively to some of the other players, but we can go out and compete because if we're able to innovate quickly with an AI product, we can compete against the likes of those conglomerates. And it's really been helpful, and we've got a couple of examples of things that we've built using AI that we wouldn't have been able to do without an army before.

Yeah, it's really amazing, the power of just being able to go with a small team, even maybe like a team of one or two in some cases, and just have like AI go full gamut and build a full product suite. Now there's going to be some pushback on like security and whatnot, which is fair. But the sheer fact that you can move like 10x, 100x faster now just after you get an idea, a shower thought or something, and go build it, have an AI just go build it overnight, is pretty pretty amazing.

And I guess from your perspective, being in the financial space over the years, how do you see generally AI and finance kind of converging today and maybe like five years down the line?

Well, I think one interesting thing is agentic finance. So, I'm a Robinhood user, and they just recently released their MCP, where you can actually place trades using an AI agent. I placed my first trade a few weeks ago when they released it, and I thought that was fascinating. Like, the fact that I'm able to be in Claude and I can hook up the Robinhood MCP and execute a trade directly from Claude. Like, if you would have told me that in 2020, I, one, wouldn't have even known what a chatbot is, and two, would have just thought that's insane that you can do that. So, the simplicity of it, it's just removing complexity from the system and making it easier for people to actually do what they want to do in the end. So...

Yeah, yeah, I caught that. I caught that as well. How much did you load up on the virtual card, if I may ask?

I kept it, you know, I kept it, I think I put in 50 bucks. So I was like, you know what, I'm just going to throw in a little bit of money, see how this works. But I'll be I'll be loading up some more, more than that in the future, definitely.

Cool, cool. Yeah, I know some people who loaded up like more than 100 bucks and just let it go, let it go wild. You're very brave. But yeah, I think this space is just kind of getting started. We're seeing a lot of different protocols kind of coming out around payments themselves and kind of guardrails around what an agent can and can't do with money. So it's a very bleeding-edge sector right now, and things are changing very rapidly every day. But it's like really, really interesting to kind of see it evolve.

But coming back to Benzinga here, since we are a developer-esque show and you are Director of Products on the API side of things, what are some of the more prominent or exciting APIs from your standpoint that you'd love some of the developers watching the show to kind of dig their hands into?

Yeah, so I'll go through a couple of examples. I think one of the interesting ones is our conference call transcripts product. So every quarter, these public companies host these earnings conference calls where the executives will basically sit down with their IR team and the analysts that are covering their company, and they'll go through their financials. The good, the bad, what to expect. And I think it's interesting to hear it from their perspective. But listening to an entire call that's over an hour long is not really feasible for the average person, the average busy person I'd say. And so our conference call transcripts endpoint, it transcribes that audio into text, it summarizes it, and you know, after the call I can basically log into my brokerage account and see that transcript. And so I think that's one fascinating one that can help steer trading over the course of a quarter. And we built that using AI. It's pretty simple to throw an audio file at any decent audio model and get some decent text quality in response. So...

And then is that real-time? So like when the call is kind of going on, can you can you call that endpoint and get a summary of the transcript up to, excuse me, that particular point of time, or does the call have to be concluded and then you get a full summary of the thing?

Yeah, it's real-time. So, the way that works, and I know this is a dev podcast, so I'll go into a little bit of the technical details, but it's an HLS stream. And so how that typically works are these .ts files that are typically like 2 seconds in length for each file. And so we'll take those .ts files, transcribe each one of them, and it'll output a short utterance which we just send through a websocket as a JSON object. And so it'll be like a couple words per object. And so people can use that socket to stream through the transcript. And they can also, you know, stream the real-time audio if they want to. So, it's pretty unique. Well, it's not unique. Other folks have done it, but I think the real-time aspect of it is fascinating.

Yeah, I mean, just curious, have you seen any interesting kind of use cases pop up with people using that? Whether it's like creating like at-the-moment kind of trade bots or anything of the sort?

Yeah, I think kind of from a historical standpoint, one unique use case is word clouding these transcripts. And so you can kind of see what's what's hot across an industry. And so if executives start to mention, I don't know, AI demand or, you know, increasing costs across the supply chain, doing that type of word clouding and like semantic clustering is one use case of having all of that raw transcript audio and text.

Awesome. And then for people to be able to access that particular endpoint and I guess the other set of endpoints as well, what's the pricing model and how can people kind of get started with that?

Yeah, so we make it super simple. It obviously depends on the use case. So, we make it super simple to get started though. So, anybody can basically just go over to docs.benzinga.com, hop into the dashboard, spin up a key, start hitting the endpoint. And they don't have to talk to a sales person for 2 weeks. And then when it comes time to when it comes time to actually, you know, use the product in production, then we kind of like to understand a little bit about the use case. But self-service is our model, and and so I can actually go through a quick demo if that makes sense.

Sure, yeah, we'd love that. [Music] Yeah, so I'll pull up my screen here. And while Tommy's pulling that up, the link to the docs and the API will be in the description of the video below, so definitely check that out if it interests you. But yeah, you are good to go.

Yeah, so so this is this is our doc site. And so the way it works is you can go ahead and pull up pull up the docs. You can enter in your API key and authorize the endpoint. And then let's go ahead and we'll fetch 10 calls. So we can fetch 10 calls and they're just printed in this this JSON list. So let me find a call that I think users might be interested in. Let's maybe try, I know Nike was a few days ago, let's try Nike. Here we go. So now I have my call ID from this reference endpoint, and now I'm going to fetch some detail about this call ID. So now basically what this endpoint will print is you've got the call title, the period, the year, a short description, a headline. And then we get into the actual we have two different transcript versions in this array. So we've got the real-time transcript, and I wish this display box was a bit bigger. We need we need we use Mintlify, so Mintlify, if you're watching this, we need a modal to pop up so we can view the full JSON. We love Mintlify though. So yeah, so then you've got after that, then it goes into the segments and that's for the live live transcript. And then let me show you the non-live version. This is the non-live version that's been post-processed, it's been cleaned up by an LLM, it's been formatted for for text, for punctuation and things like that. And so yeah, then it goes into that, and we've got each each speaker, the start time, end time, and the text that they they have for that utterance. We apply the sentiment to that segment as well, so that way you could kind of do some unique things with like highlighting, shading, or something like that. So yeah, it's kind of it's kind of interesting, from a retail perspective, to have access to these transcripts.

Yeah, this is pretty dope. And is there like a cumulative sentiment score at the end as well, or is it just per segment?

Um, I think we we might do the cumulative. If not, that's actually a great idea that we should add. Yeah, one for sure. Yeah, I'll I'll give you a kickback on that.

Awesome. Great stuff. And I guess kind of going towards the rest of the product suite, I know you mentioned Robinhood's MCP. Are you guys, do you guys have an MCP or are planning to release one soon? Is there anything you can say around that?

Yeah, we don't we we kind of have an MCP, but it's not like a true MCP server that you can really hook into hook into Claude in a simple way like like most MCPs are. We get a lot of our our requests from the enterprises, and it just hasn't gotten there yet. So they they really only care about the APIs. But I do want to build an MCP soon. And and yeah, for our use case, with with these datasets, it's mostly just a wrapper around the API. And I know the MCPs can can go into more functionality than just being an API wrapper. Like an MCP could could have a workflow for a specific tool that does that hits four different API endpoints in order to to surface the data for that specific tool. But yeah, we just haven't had a ton of demand for it. But it is in the works, and I think by the end of summer we'll probably have something ready for everyone to to try out.

Excellent, excellent, looking forward to that. And yeah, just curious, from your from your lens, are you guys starting to think about X402 and how how that side of the world might be able to consume your API data?

I'm not familiar with X402. What is it?

Ah, yes, it's basically a new primitive allowing AI agents to be able to to purchase things on demand. So, you can think of it as, if I have an AI agent, and I don't necessarily need a subscription, a monthly subscription to let's say the transcript endpoint, but I need it on demand, I can, any any merchant that has X402 enabled can purchase a request or anything really, on demand like that, whether you you price it at whatever. But it basically lets AI agents kind of pay pay for things with with crypto rails. So, again, it's it's pretty new, I'd say maybe like half a year, year's worth into it. But yeah, I think that would be an interesting use case to to apply around this data.

And it's and it's for crypto specifically?

The predominant driver is is for crypto payments, as Coinbase is the one that's developed it in conjunction with some other other folks. I think, if I'm not mistaken, the Linux Foundation now has now has the the open source code for it. Don't quote me on that, I'm not sure. But yeah, it's one of the more popular protocols at least in the agent payment space. So, I guess you get two for free, two for free.

Yeah, and you know, like um, and uh, I'm familiar with the agentic payment protocol, which I think which Stripe and OpenAI were working on. I think this must have been maybe 6 months ago. And then I think they changed it to machine payment protocol. But yeah, it's kind of it's kind of demand-driven, right, with these protocols. Like, going out and inventing a new protocol doesn't really make a ton of sense until it's really needed. And so I think kind of working through the details and, you know, from a demand standpoint, figuring out what do we actually need in place, and then having some of these big players go out and like open source a codebase and get this get this protocol like adopted, I think that's that's a good approach. And so, yeah, we'll we'll have to take a look at that.

Excellent, yeah, yeah, definitely. And yeah, a lot of these things are bunch of like alphabet soup of different protocols and people trying to kind of establish standards for other folks to use, but it's like very, very fragile and evolving like every every week, every month, so. Yeah, it's exciting times, for sure. And I guess as we're kind of closing towards time, is there anything that you'd want to share on the roadmap for Benzinga, coming up in 2026? Any any alpha you might be able to share with the audience?

Yeah, um, yeah, I'll I'll just share for anyone interested. I know, from a retail perspective, company financials and being a true fundamentals investor is kind of a safe bet. And and and, you know, not that that isn't advice or anything, but I think we're going to be focusing on fundamentals, company fundamentals in the second half of this year. And so making that that SEC information that comes directly from the 10-Qs making that easy to consume in a standardized format is something that's on the roadmap and something to look forward to.

Exciting, exciting, looking forward to that. And I guess for the aspiring founders and builders watching today, is there any words of advice you'd want to give them, given your extensive experience in the tech space?

Yeah, I guess my advice is um, to be agile and to be innovative quickly in today's day and age. Like, so much is changing so fast, and if you think that you're going to be able to to meaningfully project what 12 months from now will look look like, I think that anybody who thinks that is wrong. And so just to, just to move quickly and put ideas out there, see if they stick, see if there's any type of, you know, early product-market fit, and if it does, go all in on it. So that would be my advice.

Well said, well said. Totally agree. And as we're coming to time here, finally, what's the best way for folks to kind of keep up to date with all things Benzinga, yourself, and how can they get involved?

Yeah, um, anyone can follow us on LinkedIn, it's Benzinga APIs. And if you have any feedback or any questions or anything technically related, you can always hit me up on LinkedIn as well. It's Tommy Cotter on LinkedIn.

Beauty. And with that, Tommy, thank you so much for taking the time out of your busy day to come chat with us. It was amazing learning more about Benzinga, as I'm sure our audience also felt the same way. And yeah, really exciting things ahead, I'm sure.

Yeah, thanks for having me. Take care.

Yeah, you too. All right, gang. With that, I wish everybody a very happy Thursday, happy long weekend for our friends in the States, Happy 4th of July. And we will catch you back here for another great episode of DevNTell next week. Until then, have a good one, folks. Cheers.