Inside Tropic’s AI Engine: Unlocking Efficiency For Your Procurement
You’ve heard how teams are putting AI to work, now it’s our turn to show you what we’ve built. At Tropic, we’ve always been more about substance than sizzle when it comes to AI. This session gives you a front-row seat to the tools we’ve been quietly building to make procurement faster, smarter, and way less painful.
Welcome everybody that's just joining us, and welcome everybody that's been with us since, one o'clock.
As Justin said, selfishly, this is my favorite session because I lead the product organizations at Tropic, and I'm beyond excited to share with you all all the great things we've built, our posture and point of view on, how we think about AI in finance with, all the stuff that we've built with Tropic, as well as excited to be joined by Tyler Toft from Drata, who's gonna talk about their finance and AI journey and their use of Tropic as part of that AI journey. So excited to jump into it. I'm gonna go ahead and share my screen.
So, Tyler, if you wanna just come off, video for a second, just say hi to everybody, and then we'll take it from there. Hey, Tyler.
Hey, everyone. Looking forward to to this session a lot, actually.
Love it. So the agenda for the next hour, I'll kick it off a little bit, talk about our point of view on AI. We're gonna walk through our five AI agents that we built over the past few months, and then we'll circle back with Tyler, to talk about their Tropic and AI journey. So let's get into it.
And just to talk about my point of view on AI, I'm gonna go off off, sharing my screen for a second. So I spent a lot of time talking to finance and procurement teams, and I keep hearing the same frustrations. Right? Limited visibility into all of their contracts and their spend, renewals blindsiding them at the worst times, negotiations that drag on for months, and just fundamentally spend that's just spiraling out of control while you're trying to stay focused on your strategic priorities.
But beyond that, there's a ton of countless manual and repetitive tasks that just feel endless and just slow you all down. And what's exciting to me is that's exactly why we built Tropic. Right? To relieve those pains and enable your strategic priorities. And as Kunal said in the last session, enable you all to be strategic decision makers and not get dragged down by the repetitive manual tasks. But, fundamentally, to cut through all the noise, we're here to save you time and save you money and enable strategic decision making.
So before we we get into the demo, just to share a few learnings from a recent survey our team conducted with finance leaders on the topic of AI, it's fascinating to me that half of the surveyed organizations are piloting some sort of AI solution.
And by twenty twenty six, everybody's already forecasting three fourths of them are gonna be planning full deployment.
But the the the very interesting insight that came out of that is roughly forty percent of the organizations are reporting zero measurable impact from those current AI investments.
Fascinating. Lots of lots of, hype going into AI, but the reportable impact is not as as we'd want it to be. And I obsess over solving that forty percent problem. And when it comes to AI, my fundamental philosophy has always been, I strongly believe substance over sizzle.
That opinion has been formed over just my twenty years of career of going through various different hype cycles, whether it's the dot com, the cloud, the mobile, the blockchains, the AI now, and lots of lessons learned from experiencing all of those firsthand, being in the boardrooms, implementing those technologies, and all that has shown me that you always have to lean on the side of substance over sizzle.
AI is definitely in a different class of its own. But like the tech of the past, it's now just a very powerful hundred extra tool to solve problems, and it's a pretty seismic one.
It feels like Thor's hammer at times, but hanging a picture into drywall doesn't always warrant Thor's hammer. So I tend to lean on the side of being deliberate and mindful just, as I mentioned earlier, having gone through these hype cycles over and over.
So a few bulleted items just to to frame up the philosophy a little more.
Bad data is gonna result in bad AI. And Kunal said this, and the CFOs in the in the previous session said this as well.
So we spend a ton of time and resources to ensure the data quality and the integrity is high because absent that, the AI is gonna do bad things and steer you down bad pathways.
We've deliberately restrained ourselves, and we've taken a problem centric approach to solving AI because the worst thing that can happen is we put AI in your hands that doesn't perform consistently, is not secure, and is not effective fundamentally to solve your problems.
So all the experiences from the hype cycles have informed this this position, and we've worked over the past few months with high growth finance teams to really understand where are those problems in their workflows and their workdays and identify these five agents we're about to demo to you to really deliver consistently, effectively, securely, and scalably.
So without further ado, I'm gonna jump into the demos.
And the first agent I'm excited to share with you all is our smart request assistant.
And the the overall journey, of our demo is gonna be taken through somewhat of a linear procurement life cycle. So kicking off with requests so submitting requests in an organization can sometimes feel overwhelming and daunting, and this symptomatically shows up with a lot of our organizations as Slacks and back and forth to the procurement team and not knowing what to submit, where to submit. But we've created Ask Tropic in our smart request assistant to conversationally take you through submitting a request. We understand your procurement policies, your workflows, your approvals.
We'll guide you through what exactly needs to happen to get this request to the right place and make sure we're getting out of your way, which ultimately allows any team member to confidently submit requests, faster request creation times, higher compliance rates, and the reduced back and forth with teams having a Slack or message each other to understand who am I supposed to submit this to, where is this supposed to go, what type of request is this, so on and so forth.
So our smart request assistance guides you through the entire experience. You're able to conversationally just request the purchase that you need to make, and it will guide you through overall through a series of questions. And if you're not ready to answer some of these questions, again, you can use your natural language to walk you through those. It'll aggregate all the information we know about said suppliers, said purchase request, and prepare the the submission to go through and ultimately let you submit the request without having to think much about it.
From here, the request has been submitted. It'll navigate through all the workflows and approvals, and you can always come back and use Ask Tropic to follow-up and understand where that request is, what needs to happen next, and be guided from there.
Now that your request in is in, you're negotiating with your vendor, your supplier, you're working through that request internally, we wanna make sure you're securing the neck the best deal, and we will navigate you through our negotiation navigator. Right?
To walk you through this experience, we want to avoid you all entering negotiations blind, and we wanna make sure you're armed with all the information possible to have an effective conversation and build that relationship with that supplier.
So we have built our negotiation navigator that gives you insights into who the supplier is, what are the basic commercial terms, and and things like that that have been aggregated from thirteen billion dollars worth of spend data. You can get deeper insights into the specific products and SKUs that you're looking to purchase. You can actually present the data that you've received from a quote or proposal, and we will guide you through whether this is a great deal or worth negotiating.
We leverage our firsthand experience from all the negotiations that we've done for our customers to bring together negotiation insights and allow you to create a full negotiation playbook that's informed by that years of experience that we've got. You can also go into, generically look at all the negotiation tactics that we've built over time or look at specific custom strategies that will help with this negotiation and that supplier relationship.
The specific tactics that work that are are are best to use, worst to use, we've also given and curated that information and put that in front of you to make sure you are well equipped.
Every negotiation goes in with confidence backed by aggregated market data. We've identified all the savings opportunities and given you what the logical and and specific price range you should land on for your specific use case and your needs. And you can gain that leverage through all of that supplier intelligence and also be able to have strategic valuable relationships with those suppliers armed with all of that intelligence in front of you.
Progressing through the life cycle, you've now gone through your, negotiations with your supplier. You've got your request flowing through. You've got a signed contract in front of you. And from here, we're gonna navigate to our contract intelligence engine. So visibility into your contracts, being able to manage all of the data behind those contracts, having visibility into an example Isaac brought up in the previous session was around, you know, being able to look in all the contracts and understand which ones have auto renewals or certain clauses or terms that you're looking for.
The contract intelligence engine curates all of those experiences.
So once you have the contract in hand, being able to upload that contract as part of your overall request workflow, not have to think twice about the data accuracy and the hygiene. We allow you to upload your contract. We classify that document.
We use AI to minimize any sort of manual data entry as well as any errors as part of the data hygiene, importance that we give. And from here, data hygiene, importance that we give. And from here, our AI scans it, populates the contract, and structures the data out of all of those PDFs and the complex clauses. You could see we also give every field that we extract a confidence rating and allow you to progress through and get that contract ready to basically have all that structured information so you could really understand what are all the contracts, what are all the terms, what are the start dates, end dates, and ensure you're not missing any detail, around that contract.
Reduces manual work. We automatically extract all of that data. Data hygiene is super important to us, so we don't exclusively rely on AI to do the work. We also have a team of human data operations professionals that stay in the loop to ensure we have ensure high data quality.
So, ultimately, that delivers the benefits of slashing time spent on actually entering the contract details. You never miss critical renewal dates or any sort of important opt out dates and contract obligations.
We instantly structure as much information as possible and make sure you have all the visibility into the various elements related to contract management.
And then you can always use AskTropic to query all of that information in a conversational way and ultimately freeing up legal finance procurement and office of the CFO teams to really get access to the information they need and freeing up any sort of manual, repetitive data entry tasks as well.
Now you've got the contract uploaded. You've got, the request still flowing through.
And from here, we're gonna preview our compliance Copilot.
Policy confusion sometimes create approval bottlenecks, and they delay these critical purchases. Manual compliance checks, a lot of our data shows where, most of the procurement purchase life cycle slows down is in those critical steps of legal reviews and security reviews.
So risk and risk reviews are super important, but our aim with the compliance copilot is to make them as easy as possible.
And we wanna make sure these compliance issues aren't discovered too late in the purchasing life cycle, and basically slowing down your request process.
So our compliance Copilot works inside the a specific request or a contract.
It will is navigated through our Ask AI, conversational experience.
Once you say check compliance, it instantly scans the request or the contract for all the documents associated with it. It's gonna run through and validate and assess that contractor document against the procurement policies and preferences that you've cataloged.
And you still retain control. We're not changing anything related to the contract, but we are flagging any sort of potential compliance issues before they become problems.
And it help it serves as your sort of intelligent guide to navigate all those complex procurement preferences and policies and making sure we're, moving that purchase life cycle along to be as efficient and effective as possible.
Dramatically reducing those review times, moving faster with confidence, eliminating surprise compliance issues, and empowering you all to to run these self serve compliance checks so you can also be educated on where things potentially might get slowed down or bogged up.
And last but not least, one of the most cited, problems in the previous sessions and a lot of the conversations I have with finance leaders is month end close and making sure you're getting invoiced what you're contracted with. So excited to feature our AI invoice match agent.
Account payables team spend a ton of time as part of month end close to ensure invoices are received in a timely manner. They're actually matched up against not just the payment terms or or the contracted spend, but also the payment terms that have have been agreed to and negotiated.
Obviously, this creates a lot of, manual work. But beyond the manual work and the time, there's a lot of policy and risk that also is introduced if you can't reconcile the right invoices against the the right contracted spend. And then inconsistent supplier data creates confusion and processing delays. So all in all, sometimes a valid invoice that is, delayed payment can also invoice matches is aimed to reduce all of that administrative burden.
So to walk through it, you just received an invoice from a supplier.
This works alongside your existing AP, solution. So if it's bill dot com, ramp, brex, you basically forward the invoice to a dedicated inbox for your Tropic instance. So in this case, invoices dot tropic app dot I o.
Once that invoice is forwarded to that specific email address, we run that against, the contracted spend. We will identify the supplier. We will then reconcile the invoice against not only the amounts but also the payment terms, and we'll flag any sort of issues that we've spotted against what's in the contract.
You can navigate here to dig further into the invoice and see exactly where the discrepancies are.
And once you look at the invoice within the Tropic instance, we will specifically flag where our AI has high confidence against the invoice and the reconciliation process, but also where the specific issue lies.
This is aimed at freeing up accounting and finance teams, preventing those costly errors and overpayments, defending against those fraudulent invoices, processing invoices faster, and and not delaying any of those payment cycles, and redirecting all the finance and AP resources to focus on the strategic efforts as opposed to, again, the administrative tasks.
With that, I'm really excited to bring Tyler into the conversation. So five agents solving real problems, leaning on the side of deliberate intentional use of AI, focusing on the five over fifty agents that actually will get the job done. So, Tyler, welcome to the conversation.
Excited to bring you in. And, yeah, I wanna turn to you to just introduce yourself to the audience as well as talk about your Tropic and AI journey at Drata.
Yeah. Thank thanks for having me. I'm super super excited to talk about, the product, how we're using it. So I've been at Drata three and a half years.
I run, the FP and a team as well as procurement, and we've been using Tropic a little over three years. So we've kinda seen it evolve in this last kind of, push for AI has been super exciting for us. And, at at Drata, we're AI is kind of in our DNA, not only in our product, but also in our use cases and the vendors that we choose. So in a way, we can leverage AI to make our streamline processes, kind of double check our work.
We're super excited about it. So the two two areas that we've been leveraging a ton lately is both on our contract intelligence and AI invoice match. I think, historically, it would take a lot of time. So we would especially on contract intelligence, every Friday we sign contracts.
I'd have to block off thirty minutes to an hour after we send all the contracts to load everything into Tropic, scan through the documents, grab the right details, load it in. We're now at five, ten minutes each Friday. I can upload all the documents. It scanned it.
I do a quick once over, and we're done.
So that's been super helpful for us to kinda streamline, especially on my end, huge time savings every Friday morning as we go inside all those. And then we move over to the AI invoice match.
Previously, we would have, we would review our bills in our our AP platform. Then we'd go double check tropics. So there's a lot of back and forth, like, making sure our things are right, human error.
This gives us a lot of confidence in seeing it right up front. And then a couple weeks ago, we actually had an a vendor invoice us that was thirty thousand dollars higher than, than the contracted spend. A AI invoice match flagged it. We're able to talk to the contract owner.
He contacted the vendor, and turns out they actually sent an erroneous invoice, and they needed to correct it. And it saved us a a thirty thousand dollar error. So, this broad attempts are through the roof lately, and so this kind of is an extra layer of protection that, we're able to seamlessly add. And it's a pretty low lift being able to turn those features on.
And, super excited with, like, what we're able to do with AI. I think we're kinda just scratching the surface, especially from a procurement standpoint of of where AI is headed.
That's amazing. So time savings from just the contract extraction on a Friday. You can now have your Friday afternoon back as well as the the thirty thousand dollar invoice saved as real ROI, that you've just called out. Yeah.
I'm curious. As we've been rolling out these AI capabilities. You know, we've obviously rolled out Ask Tropic as part of the original hype around, you know, generative AI and being able to provide those natural language experiences. But over time, you know, we've sort of adhere to that substance over sizzle phenomenon.
As we've rolled out these different agents and these experiences, what's been you your and your team sort of adoption, openness, or how have you guys approach that whole, hey. Tropic put something out that's AI oriented. Is there apprehension? Is there excitement? What's the adoption curve look like?
I think our approach is encouragement just to start using it. I think any new technology's a little bit different or you're not sure, you're maybe apprehensive at first to start. But our kind of approach is let's start using it. At first, maybe it's not gonna feel totally right or it's gonna be different, but after a couple iterations or as new releases come out pretty quickly, you're able to start using it. So that's kind of like our overall our overall approach is, like, just dive in, start using it, figure out how you can leverage it. Even if it's for one specific use case, that's still a win in my books. And then you kind of learn from there how to evolve it and and apply it to more areas.
Got it. You know, it's interesting to ask this question of of someone coming from Drata, but a lot of conversation around AI has been just, security, data privacy. Where does this information go? How do you guys use it? What's, Drata's take been on it, and how have you been, you know, sort of coaching your team and guiding them through Yeah. Open it to AI.
Yeah. There there's certainly a push of, AI full speed ahead, but also with caution and making sure we have security reviews. So we've kind of we've kind of, pushed the overall company and team into, let's explore all these AI tools. But when we get to a point where maybe sensitive data is gonna be going through it, we need to make sure the security team is looped in as able to do a review.
And that's where we Tropic gets involved in. And so we've set up our our workflows in a really streamlined way that, something that previously maybe took a couple weeks to get reviews all done. The security team is able to kinda quickly, within a week or under a week in most instances, be able to review a tool and give the employee or the team a thumbs up on using the tool further. But, yeah, that that's a a big topic that, we're constantly discussing and making sure we're not slowing people down, but also not exposing the company to too much risk.
That's amazing.
I know when, you and I chatted earlier this week, you know, we sort of aligned on our posture around we're definitely in the middle of the hype cycle. We need to be careful about, you know, AI and and falling in for the hype. But curious if you could expound on that and just your your point of view and how you've been sort of, you know, either, you know, leaning into that or sort of coaching your teams around, you know, what to do with the hype cycle.
Yeah. Everyone wants to be the first to claim they have AI in their product. And I've been on a few demos where if everyone claims they have it, and then it's like, let's actually see it. Then that's usually where kind of things start to fall apart. So anywhere we can see a demo or things online online where the AI is actually in action in a specific use case immediately builds trust with me as a buyer and gives, like, me comfort in, like, knowing, oh, this is exactly how I would even apply it. And usually indicates to me that there's there the path and the, the road ahead for whatever vendor really has a grasp on AI and how they plan on incorporating it versus just kinda like the, yeah, we have it, but we're not totally sure how to incorporate it yet.
Yep. Very cool.
Zooming back out and, you know, covering a little bit more of, obviously, you know, all the AI agents we built, they index on saving time and saving money. As you're sort of looking at your finance function and your finance and AI journey, what are some of the the prioritized benefits, impacts, ROI things that you're looking at as you're sort of evaluating the various different AI tools or even just as you look at the AI agents we just walked through, how do you evaluate them on the value that's provided?
I think, immediately is time savings. That's, time is limited. So anytime we can add time savings, that's always a huge value add. I really enjoy the chat features when you can quickly find something instead of navigate or search through different files. Like, I need to find this contract or something. You can quickly find it, which goes back to time savings.
So that's, like, the immediate biggest value. Now one area we didn't touch on is the, highlighted savings as part of the kind of the benchmarking data where in the background, AI is matching up some of your existing contracts and flagging, hey. We think you might have additional savings here. So there's, like, real ROI in some of those, instances where you can actually put a dollar value on it. But to me right now, time is is certainly the biggest ROI.
That's great. Yeah. And for the benefit of the audience, you know, stay tuned for a lot more. We do a lot of enrichment, today, but also planning to do a lot of enrichment agents against the contract data.
So Tyler's calling out our proactive price benchmark, which if you look at your contracts inside Tropic rather than you having to go pull your price benchmark, we actually extract all the information out there and actually give you a read on if there's a savings opportunity here or not. If so, is it high, medium, low? Are you getting typical pricing? Things like that.
Stay tuned for a lot of other agents such as alternate suppliers. Are there other suppliers you should be looking at? Do you have, you know, various different suppliers that provide the same functionality that maybe there's a consolidation play? We have other agents today that run that will identify shadow suppliers and shadow spend. So as you connect various different integrations, we will light up. You know, we see a supplier in NetSuite that we don't see a contracted spend for or vice versa and things like that. So lots of, exciting things coming ahead as we continue our AI agent journey.
But, I'll pause there for now. But thank you so much, Tyler. Really appreciate, you talking us through all the things you've been doing at at Drata and how you've been getting the value out of, Tropic. But yeah. Any questions?
Thanks again for having me.
Yeah. Let's take a look if we have any questions.
Nothing specifically came up.
Few of them came up that were already answered.
Sorry. There is one in the chat that came in from Krista, actually. Is there any progress on, like, foreign currency matching, likely with, AI invoice match?
Yeah. It's a great question.
On the backlog to consider lots of things to to make sure we get right with contract extraction and making sure we can actually do the AI invoice match against the right currency. So on the backlog, something we're looking into, but stay tuned for more.

Our Speakers

Srinivas Somayajula


