Review of Materia – The Accounting Technology Lab Podcast – August 2024

–

Or use the below podcast player to listen:

–

Transcript (Note: There may be typos due to automated transcription errors.)

Brian F. Tankersley, CPA.CITP, CGMA  00:00

Welcome to the accounting Technology Lab sponsored by CPA practice advisor. With your hosts, Randy Johnston and Brian Tankersley,

Randy Johnston  00:10

good day. Welcome to the accounting Technology Lab. I’m Randy Johnston, with my co host, Brian Tankersley, my we’ve seen a lot of new technology as this year’s unfolded. Today we wanted to speak to you about materia, which has a rather odd website. Try material.ai now this particular product really fits more in the audit and assurance space. And the CEO, Kevin marlini and the CTO Lucas Adams were both at Facebook before they came up with the idea for materia. Now one of the things that you should know about materia is we think it represents the trend that we see in products where AI is going to be incorporated into the platform. It’s going to become part of the platform. So many are worried about, are they going to get left behind on AI? And our position has been, you can use the generic platforms, the chat gpts and the copilot 365 that we’ve spoken with you in other technology lab, but we think for most accountants, AI will become part of the tools, just like you don’t think about secure socket layer being part of your web browsing experience, it all got built in. So Brian on materia, you know, again, this is an AI platform for accounting and insurance operations. What do you think is important for our listeners to hear about?

Brian F. Tankersley, CPA.CITP, CGMA  01:40

Well, I mean, I’ve looked through their information and so forth. But the the idea here is that it’s really a tool to analyze and summarize documents to, you know, again, and to to summarize and create, create memos that talk about the guidance and so kind of summarize the related rules. You can go through and look at engagements, and you can again, go through and kind of standardize workflows with it. The thing, the thing about it here is that when I’m when I’m looking at this, I think there is a lot of value if you have a lot of documents, okay, so people with significant lease leases or debt or other things like that. Could, you could certainly use this to go through and and figure things out about, about, you know, discover things out of the related documents. You know, the I actually went through and read they, they actually have a privacy page. And I was actually quite impressed with what they said on their privacy page, because they said your data is not used to do you know, they have a SOC two, type two. They they say categorically that they do not your data is yours. We do not train any models or content provided to materia. And again, they’ve got they, they’re saying all the right things with respect to it on privacy here. And I think that’s important, because, like you and I have discussed, this is kind of the AI world right now, is kind of the Wild West. And you know, the free versions of all the products, almost universally use your data to train models into it actually uses your data to train models according to their privacy policy and QuickBooks Online and their their website. And so I think it’s, I’m just very encouraged to see somebody have a categorically say, we do not train models with your data.

Randy Johnston  03:44

Yeah. And, you know, Brian, when you and I have spoken about this, these private AI models, private to the firm, is actually a pretty important step forward in AI, you know, further materia has said, you know, we’re US centric. We’re going to try to work with your firm’s guidance and policies. We’re going to ingest your client’s document to help you with your day to day work. They’re going to read your firm’s checklist, whether it’s reviewing board minutes or debt agreements and so forth, and it’ll actually look for your engagement data to make sure that everything inside the work papers are, you know, connected now, over the fairly recent past, I believe It was on June 20, materia raised about 6.3 million in funding. And that’s important to know too, because that particular round was done by spark capital, haystack ventures, temps Thompson, Reuters ventures, exponential founders capital and the Allen Institute for AI. Now they didn’t disclose the very. Amounts, although I think I did you know, learn of some of the amounts along the way, but I thought the Thomson Reuters investment was important because it gives you an idea that this might get incorporated with the Thomson Reuters checkpoint engage AI, it might be incorporated more heavily into advanced flow, but the founders are trying to keep this thing relatively independent at this point in time.

Brian F. Tankersley, CPA.CITP, CGMA  05:37

Yeah, I think the, I think, generally, as as we’re kind of looking at this, there’s, there’s a lot of there are a lot of interesting things this is doing right now. This seems to be a tool for big size and big, big firms and mid sized firms that are trying to, again, use AI to process information. One of the interesting things that they also have in there about is, is, where is their their sub processors, okay, and this is another thing that, and I’m trying to find it on the page here, but this is another thing where, where they are. They are very forthright, saying that they use, they use, open a eyes, so in their trust portal, actually, they list their sub processors, and their sub processors include, include Claude. They so the Claude engine, merge, cohere, Amazon Web Services, anthropic, which, of course, does Claude data dog that in there? And again, they’re using AWS for their hosting, and they’re in the US East region, okay, so, so they’re, they’re doing a lot of things, a lot of things, right? And they’re working with multiple models, which I think is pretty important,

Randy Johnston  06:54

yeah, you know, Brian, I think that’ll be a great topic for another technology lab, because, you know you and I have spoken about sub processors. You did work on that in AI confidential for k2 this year. And I’ve actually concluded, Brian, we should produce a naughty and nice list for sub processors, because we are worried about the data when things go to a sub processor. But, you know, again, the sub processors here looked pretty good. You know, reputable companies, is the way I would the

Brian F. Tankersley, CPA.CITP, CGMA  07:30

only one that gives me any pause is cohere, you know, because they’re, I think they’re, they have a lot of links to the US intelligence community,

Randy Johnston  07:40

yeah, and so the whole sub process or conversation friends, if you’re not aware of it, we will make you aware in a future technology lab. But Brian was correct. The materia company has paid for a SOC two type two audit, and they do integrate with Microsoft 365, and a variety of internal data and document stores and beyond that. The implementation here is quick. They do offer pilots so you can evaluate material if you’d like to do that. They also are very conservative with the data, as Brian has pointed out, they try to make your data, your client’s data, accessible to the AI platforms to answer questions and do workflows and so forth, but the relevant part of your data is the only thing that’s sent to The AI models, and so the data is never used to train any models. Now, in prior Technology Labs, we’ve actually spoken about open, AI’s, chat, GPT, which the $20 model is used to train models, but the team’s model, the $25 model, does not use your data to train models, and that’s a shift in open AI’s policy, you know, here in the past year or so. But material is used by all levels of your auditing and accounting team, intended to be used from staff and managers all the way up to partners to try to get answers faster for your clients.

Brian F. Tankersley, CPA.CITP, CGMA  09:26

And I think, I think just generally, that’s the major story we have here. You know, the what we’re seeing again is that the a lot of, the lot of the really, real innovation kind of the moonshot type projects are taking place with with startups like materia, as opposed to taking place from within the big firms and the big the big publishers themselves. There are notable exceptions, and some that we’ve talked about it we’re going to talk about in subsequent Technology Labs. Uh, you know the Thompson co counsel feature, the Wolters Kluwer answer, Connect tool, are two notable, notable exceptions to that. But in general, in general, you’re going to want to watch the you’re going to want to watch the space, because AI is going to hit and and it’s everybody’s talking about it now, but it’s not going to be too long before some folks are using it to great effect. So I would encourage you to do some experimentation and take some demos and check out some of these emerging tools to make your firm more effective, more efficient and effective.

Randy Johnston  10:38

Very good. Well, you know Brian, I think that’s a good summary. The assistant inside material that does the embedded assist, embedded citations, the Microsoft 365 integrations and the document workspace are all part of how this product works. I’ll just part with the additional, I’ll say comment that the integrations are to SharePoint and OneDrive, as well as Box and Dropbox and Google Drive, so they’ve integrated to a variety of other systems out there. So again, it’s worse. We think, if you are in the audit business in the US to at least look what material might do for your productivity. We look forward to speaking with you again on a future technology lab. Good day.

Brian F. Tankersley, CPA.CITP, CGMA  11:34

Thank you for sharing your time with us. We’ll be back next Saturday with a new episode of the technology lab from CPA practice advisor. Have a great week.

= END =