WEBVTT
00:00:06.000 --> 00:00:10.320
Hello and welcome back to a very special office hours edition of the What the Tech Podcast Room Boast.
00:00:10.560 --> 00:00:13.679
I'm Paul Dam, Portbost Head of Content and Resident Tech Journalist.
00:00:13.759 --> 00:00:22.719
And this week's episode is part two of a conversation regarding AI and RD tax credits that features both in-house experts Matt Rudishauser and Joshua Martin.
00:00:22.879 --> 00:00:36.880
Check out last week's episode on the feed if you missed part one, and buckle up if you're ready for today's episode, where we dig deeper into how the use of AI is changing RD tax credit claims, and specifically the unique opportunity for innovators to claim provincial credits to fuel their innovation.
00:00:37.119 --> 00:00:42.240
Both myself and Joshua will be in Montreal this week for All In Conference, Canada's largest AI gathering.
00:00:42.399 --> 00:00:47.600
So if you have any questions following this episode and are going to be at the big show, stop by booth B6 to say hi.
00:00:47.759 --> 00:00:50.000
Without further ado, let's get into the interview.
00:00:50.240 --> 00:00:55.280
Okay, so we were talking about CDAE, we were talking about federal shred, and we're also talking about provincial crick.
00:00:55.359 --> 00:01:02.799
Now, Joshua, is there a scenario where a company qualifies for all three, or is it usually just one or two in practice or one here?
00:01:03.039 --> 00:01:04.799
What's kind of the way of the land?
00:01:05.439 --> 00:01:06.719
You know, absolutely.
00:01:06.959 --> 00:01:13.439
So just to clarify, shred uh and crick are applied for in the same application process.
00:01:13.519 --> 00:01:15.920
So crick is just the provincial shred.
00:01:16.239 --> 00:01:23.200
Um, there are differing eligibility criteria in the sense of you can actually include some pre-commercialization activities in Crick.
00:01:23.280 --> 00:01:35.840
So it's more uh uh as you're completing the provincial expenditure forms, you're able to include some additional expenditures, but the submission is still done uh you know in tangent in parallel.
00:01:36.000 --> 00:01:45.040
Uh, and then the the responses are done first federal response, and then following that Hofnie Quebec will omit their decision and the check in consequence.
00:01:45.280 --> 00:01:47.439
You can stack CDAE on top of that as well.
00:01:47.599 --> 00:01:52.000
So, really, there is there are many scenarios where you're stacking and optimizing between different programs.
00:01:52.159 --> 00:01:56.159
It's not at all odd to take advantage of all these different programs.
00:01:56.480 --> 00:02:05.040
One scenario that we see is less common is if you're using uh you know CDA, shred, and multimedia.
00:02:05.200 --> 00:02:14.159
Those three combined are extremely rare because often CDA, the eligibility criteria under CDA and multimedia, you know, don't really go along.
00:02:14.639 --> 00:02:30.800
I think there are some edge cases we've seen, but by and large, you're really gonna pick kind of shred for your uh you know eligible innovation work and then stack on either CDA or multimedia, depending on your business type, business activities, and revenues.
00:02:32.080 --> 00:02:33.280
That makes perfect sense.
00:02:33.360 --> 00:02:35.360
And well noted on the multimedia too.
00:02:35.439 --> 00:02:38.000
I believe the English abbreviation is CTMN.
00:02:38.080 --> 00:02:39.520
Is that correct, Joshua?
00:02:39.919 --> 00:02:41.599
Uh yes, that's the French abbreviation.
00:02:41.919 --> 00:02:43.360
Oh, that's the French abbreviation.
00:02:43.439 --> 00:02:44.400
Okay, that's why you're here.
00:02:44.639 --> 00:02:47.439
Matt, any other notes on that about kind of the stacking strategy?
00:02:47.599 --> 00:02:56.000
I know we covered it last time, but for the provincial programs in Quebec specifically, is it abnormal to take care to take advantage of the three that we named?
00:02:56.159 --> 00:02:58.560
Or to even let's throw in the digital media tax credit.
00:02:58.639 --> 00:03:06.800
How do you kind of see those lay out and maybe even have timelines where it makes more sense for one to be more valuable than the other, volleying back and forth?
00:03:07.199 --> 00:03:12.240
It always makes sense to get as much non-dilutive funding as you can.
00:03:12.479 --> 00:03:27.199
The more sophisticated CFOs out there are definitely the ones who are always looking to optimize their non-dilutive funding, their government tax credit and grant programs.
00:03:27.599 --> 00:03:33.599
So you'll see people go after all this money that is out there that is free.
00:03:33.840 --> 00:03:42.479
The one thing that I will say is it's not so much a differentiation in timeline as to which of these programs you claim.
00:03:42.560 --> 00:03:52.159
And this is part of the reason Joshua was saying you don't see much overlap between the multimedia tax credit and CDAE.
00:03:52.560 --> 00:03:57.039
It's more the type of work that you see being done.
00:03:57.199 --> 00:04:16.720
So the nice thing about Shred is it is industry agnostic, whereas CDAEA is very much AI focused, and the name gives it away, but multimedia tax credit is multimedia focused.
00:04:16.959 --> 00:04:23.279
So that's more the dividing line than uh timeline or those pieces.
00:04:23.600 --> 00:04:37.839
And yes, as Joshua mentioned, there are some points in time, or no, there are some pieces where CDAE, IA, and multimedia can overlap, but there are they are not very common.
00:04:40.079 --> 00:04:40.720
Heard.
00:04:41.040 --> 00:04:51.279
All right, man, I think you kind of alluded to this in the answer that you just gave here, but to put a fine line under it, if a finance leader takes away one thing about stacking these programs, what should it be?
00:04:51.600 --> 00:04:54.160
There's the obvious, it's free money.
00:04:54.319 --> 00:04:59.920
Okay, it's not a hundred percent free money, but for your company, it pretty much is free money.
00:05:00.240 --> 00:05:11.519
The one thing that I would say on top of that is it can get very nuanced, all the different details.
00:05:11.680 --> 00:05:21.920
There's a lot of minor little tweaks that can really move the needle one way or another as to whether you're going to be able to claim something or not.
00:05:22.160 --> 00:05:31.279
Crossing your keys, stotting your eyes really matters here, and so it is good to be on top of all that ahead of the game.
00:05:31.600 --> 00:05:57.759
Doing it, trying to cram things in retroactively, build stuff backwards is a far more time-consuming exercise, and you end up seeing more things that could possibly slip through the cracks, which has a direct impact not just on how much you can claim, but how fast you get the turnaround, because we all know that cash flow matters.
00:05:59.839 --> 00:06:08.319
If I were to add on Matt's point, yeah, free money's great, you know, it's a it's an entitlement.
00:06:09.279 --> 00:06:14.639
My kind of core message would be about thinking strategically or when it comes to tax credits.
00:06:14.800 --> 00:06:18.160
Too few companies wait until the end of the fiscal year.
00:06:18.319 --> 00:06:24.959
They have a really incomplete process which leads to unclaimed entitlement amounts.
00:06:25.279 --> 00:06:35.439
So thinking proactively, mapping out your strategy, understanding which documentation is needed per different credits is the real message here.
00:06:35.759 --> 00:06:40.720
Use software, you know, everyone's everyone, AI is the word of the day.
00:06:40.879 --> 00:06:48.160
Uh, there are tools that can make your life easier to ensure you're claiming these different tax programs.
00:06:48.399 --> 00:06:49.839
It's time to have that conversation.
00:06:49.920 --> 00:06:53.199
It's time to improve your tax credit strategy.
00:06:54.399 --> 00:06:55.120
I love that.
00:06:55.199 --> 00:07:12.319
And I think both of the points that you made lead to the next section because, like Matt was saying about the documentation and making sure that it's not a retroactive activity, something you're doing proactively or prospectively throughout the year, and making sure that you have all that information documented and in place to make the claim.
00:07:12.720 --> 00:07:18.240
And to Josh's point, also about being strategic about it from the start so that you know what you're going to be claiming.
00:07:18.319 --> 00:07:25.279
I mean, or at least what you're going to be tracking, so that you can have a solid claim and a bulletproof claim if there is such a thing.
00:07:25.439 --> 00:07:41.120
So that brings me to audit readiness and kind of common pitfalls as they relate to building the claim, but then after you've already filed, and let's say that one of the agencies that is going to be administering the actual funding comes knocking and they want to know, wait, hold on, something doesn't smell right here.
00:07:41.279 --> 00:07:43.600
So, Matt, gonna go with you first.
00:07:43.759 --> 00:07:54.959
When a stack shred CDAE claim gets reviewed, understanding that these might not be claimed through the same avenue, what does a strong audit position really look like above and beyond what I just rambled through there?
00:07:55.519 --> 00:08:08.639
Yeah, so it is there's a few things to look for, and it's probably better to break this down by the different programs because each program has different nuances.
00:08:08.879 --> 00:08:25.759
On the shred level, there's there is the documentation, but even beyond the documentation, what you want to do is make it easy for the CRA to agree with you and hard for them to disagree with you.
00:08:26.240 --> 00:08:38.720
So what you're trying to basically do is set up your documentation in a way that really outlines that whole narrative that you're going for.
00:08:39.039 --> 00:08:49.759
And the more you can have everything come together as a cohesive picture, the more convincing that picture is to the CRA.
00:08:50.159 --> 00:08:56.799
And the easiest way to do that is not to try and duct tape things together after the fact.
00:08:57.120 --> 00:09:01.679
It's as you build, as you go, this is the reality.
00:09:01.919 --> 00:09:07.759
You're just looking at it, or we're helping you look at it through a shred lens.
00:09:08.080 --> 00:09:16.559
That is a large part of what our team and our platform does, is apply that shred lens to the data you're already correct collecting.
00:09:16.799 --> 00:09:19.440
So that's the dread piece.
00:09:19.919 --> 00:09:41.200
Where I would look at with the CDAE IA piece, that is far more on the financial documentation side, and showing that you have the documentation that actually matches their checklist and all the different pieces that they want to have in place.
00:09:41.919 --> 00:09:48.000
One of the and there's actually and the same type of thing goes for Craig.
00:09:48.639 --> 00:09:55.919
The one other point that I want to make is there's a few different risks with audit.
00:09:56.159 --> 00:10:07.120
The the obvious risk with audit is that the they come in and just cut the whole thing or reduce your claim or whatever.
00:10:07.519 --> 00:10:18.639
But even if they come in and they say yes, everything's good, that still has two negative effects on your company, on the ROI of these claims.
00:10:18.960 --> 00:10:30.639
Because one, you're now spending up to 40, 80 hours, like you can just be you can these audits can be time sinks.
00:10:31.840 --> 00:10:38.080
And then the other thing an audit does is it really can delay when you get your check back.
00:10:38.720 --> 00:10:48.559
And depending on how you've set up your finances, that might mean that you need to get bridge financing in place, which is an additional cost.
00:10:49.200 --> 00:11:00.080
And so knowing the audit risks, knowing that, hey, you're covered in case there is an audit, is something that is valuable.
00:11:01.759 --> 00:11:02.639
I love that.
00:11:02.879 --> 00:11:16.080
So to recap a little bit there, knowing the risks is one part of it, being educated on what could potentially happen to you, because again, free money is only valuable if you receive the free money and on a timeline that you can actually spend it.
00:11:16.240 --> 00:11:20.720
So you want to avoid scenarios where you're gonna get bogged down and then not have a check come out.
00:11:20.879 --> 00:11:27.919
But to that other point about the time commitment too, it's about knowing the risks, and it's also about having risk coverage, more or less.
00:11:28.159 --> 00:11:33.759
So, again, not to say that both is a traditionally insurance for your RD claim.
00:11:33.919 --> 00:11:41.039
We do just make sure that we have everything you need to back up that claim because we've been there before and we know how to navigate these realities.
00:11:41.279 --> 00:11:51.039
Now, you also talked a little bit about the nuances between the Shred program, which is a lot more about kind of that technical qualification, making sure activities themselves meet that criteria.
00:11:51.279 --> 00:11:57.440
Whereas CDIE, you touched on, is a little bit more about making sure on the finance side everything makes sense and everything adds up.
00:11:57.600 --> 00:12:01.600
Josh, I want to turn to you a little bit to pull on that thread and expand on it a little bit more.
00:12:01.759 --> 00:12:07.519
Is there anything Quebec-specific that trips companies up during an IQ or Revenue Quebec review?
00:12:09.679 --> 00:12:18.720
So it's very much, I think the key differentiator between Shred and Revenue Quebec reviews is the frequency.
00:12:18.879 --> 00:12:26.159
You know, Shred is one every, you know, let's say four to five years is kind of the general rule of thumb.
00:12:26.240 --> 00:12:30.000
They're gonna validate your technology, your expenditures, and make sure that it fits.
00:12:30.159 --> 00:12:34.480
And there's on a gray when it comes to a shred uh a shred uh audit.
00:12:34.559 --> 00:12:37.120
You know, how has the technology evolved?
00:12:37.360 --> 00:12:44.159
Is your works does that still meet that kind of high threshold of uncertainty that's really required for a shred claim?
00:12:44.320 --> 00:12:54.879
And then our mandate as your consultant is to really uh you know fight and show through documentation and current practices how you still meet that that threshold.
00:12:55.200 --> 00:13:00.720
Revenue Quebec reviews when it comes to CDA or CDA C D A E IA are totally different.
00:13:01.120 --> 00:13:04.240
It's a much more narrow credit.
00:13:04.559 --> 00:13:11.759
You either meet the the you know required uh uh you know activities and employee count, or you don't.
00:13:12.000 --> 00:13:14.879
The nuance is often in the documentation.
00:13:15.200 --> 00:13:25.919
So do you have proof that those you know six eligible employees actually did spend a certain amount of their time doing the required activity types?
00:13:26.240 --> 00:13:38.080
Do you have proof that your revenues actually were derived from these business codes that are really detailed on the Revenue Quebec's website and FST Small Quebec's website as well?
00:13:38.399 --> 00:13:51.759
So it's more about clarifying and showing using documentation to show how you fit into the the required boxes that uh revenue Quebec and FST Small Quebec lay out.
00:13:51.919 --> 00:14:03.039
Whereas for Shred, it's often, I don't want to say it's more theoretical, but there there's almost like uh an intellectual component to it where it's you know, how's technology evolved?
00:14:03.200 --> 00:14:07.279
How does this meet and surpass the ways technology has evolved?
00:14:07.440 --> 00:14:08.480
Why is it unique?
00:14:08.639 --> 00:14:22.799
You know, it's a lot more there's a lot more gray to play with, whereas a lot of the times we can predict when a claim is going to have trouble when it comes to CDAIA because of the you know clarity around the credit.
00:14:22.960 --> 00:14:34.639
Obviously, this year with the introduction of CDA IA, there's more uncertainty in the program, but nonetheless, documentation kind of allows us to ensure that we're claiming every dollar that companies are entitled to.
00:14:35.279 --> 00:14:47.919
Just to add to that, even though the program has significantly changed, the mindset of Investissement Quebec and Revenue Quebec has not.
00:14:48.399 --> 00:14:59.360
So the same mindset for reviews that they used with CDE CDAE, they're gonna use for CDEAE AI or IA.
00:15:00.960 --> 00:15:09.600
Cat, yeah, and I think another point that I'm just gonna pull on from my content perch is when you're watching the news, you hear a lot about agencies themselves using AI.
00:15:09.759 --> 00:15:23.919
I don't think that changes necessarily, and this is just anecdotal coming from me, the burden that you need to put into creating a compelling narrative with your RD claim, with the shred claim specifically, because it is a little bit of storytelling, like they were saying here.
00:15:24.000 --> 00:15:29.440
Um, you need to make sure that you're saying what has actually changed in a way that, in human words, are convincing.
00:15:29.679 --> 00:15:36.799
So even if they're deploying their own tools, it's not taking the human or the narrative or the storytelling component out of the equation at all.
00:15:36.960 --> 00:15:50.799
It might be speeding up the process in the same way that our own team uses AI or the AI-enabled tools to speed up our collection and make sure that we can claim things and make sure that we have the data that we need closer at hand to build your claim.
00:15:50.879 --> 00:15:54.320
But that doesn't mean that a human is coming out of the equation full stop.
00:15:54.559 --> 00:15:55.440
Go ahead, Matt.
00:15:55.759 --> 00:16:06.240
Just one story to underline that point is so a company that we know of, they used AI to put together their thread claim.
00:16:06.480 --> 00:16:07.919
It went to audit.
00:16:08.159 --> 00:16:14.080
The auditor used AI to underline why it wasn't tread eligible.
00:16:14.320 --> 00:16:23.840
Now, when they came to us, so they were able to get Claude to say this is thread, and then the CRA got Claude to say no, this isn't tread.
00:16:24.000 --> 00:16:29.200
And so when they came to us, we were able to go back and say, you know what?
00:16:29.440 --> 00:16:30.559
That's Claude.
00:16:31.039 --> 00:16:41.120
Here's the actual human reasoning, the human intelligence, the HI that finally managed to get that to go through Audit.
00:16:42.799 --> 00:16:43.840
I like HI.
00:16:44.080 --> 00:16:47.039
I'm gonna make that happen, at least down here in the States.
00:16:47.120 --> 00:16:56.720
And also from my content perch, what you just described with the two Clauds talking to each other is my worst nightmare because again, that's my whole industry going out the window right there.
00:16:56.799 --> 00:17:01.120
But I'm not gonna digress because that's why we're three guys talking to each other here.
00:17:01.279 --> 00:17:04.960
We're the human intelligence, not behind the entire boast product.
00:17:05.039 --> 00:17:08.319
I'm certainly not, but that you need to really make that difference.
00:17:08.400 --> 00:17:13.680
And again, tell a compelling story because we're ultimately serving humans at the end of the day, not AI.
00:17:13.839 --> 00:17:19.440
So yeah, I'll digress on that, but I think it's an important point that I want everyone here to hear now.
00:17:20.000 --> 00:17:22.480
All right, so guys, we're moving into the closing questions.
00:17:22.640 --> 00:17:31.039
So we're gonna be covering a lot of ground here that we already covered, but again, points that I think bear repeating and that are gonna be good for this audience to really take home with them.
00:17:31.200 --> 00:17:33.200
So, closing question for you both.
00:17:33.359 --> 00:17:35.359
Joshua, I'm gonna point this at you to begin with.
00:17:35.519 --> 00:17:44.000
But if a CFL or CTO watching this is building AI into their product right now and has never looked into any of these programs, what's the first step?
00:17:45.759 --> 00:17:48.400
I know it can seem overwhelming at first, right?
00:17:48.640 --> 00:17:57.759
You have Freddy, multimedia, plethora of grants to also complement these tax credits.
00:17:58.160 --> 00:18:06.160
A good starting point is talking to other founders and hearing what their experiences have been, what they recommend.
00:18:06.400 --> 00:18:16.079
Obviously, consultants are going to be pushing you in one way or another, but I think it's good to start by hearing what experiences other founders have had or other CTOs.
00:18:16.400 --> 00:18:20.960
From there, the likely solution is understanding, okay, can I do this in-house?
00:18:21.039 --> 00:18:22.960
Is it worth using a consultant?
00:18:23.279 --> 00:18:32.720
For more companies than not, using a consultant is the right approach, just simply for the reason that you know we all have to do this, to-do lists that are that are too long.
00:18:32.799 --> 00:18:36.400
So adding something to it means you likely won't end up getting it done.
00:18:36.559 --> 00:18:44.720
Uh, and then I would start by finding a consultant that's able to really pre-qualify your work and start with their estimate for you.
00:18:44.880 --> 00:18:50.559
That estimation process allows you as an operator to understand, okay, great, is this worth my time?
00:18:50.640 --> 00:18:57.200
If I'm gonna commit 20 hours of resources, which is often the case, what's gonna be my ROI at the end of the day?
00:18:57.279 --> 00:19:00.480
If it's only for$15,000, does it make sense to claim?
00:19:00.640 --> 00:19:01.519
Likely not.
00:19:01.759 --> 00:19:05.599
But often with tread CDAE, the claim size is much larger.
00:19:05.759 --> 00:19:15.119
So really making sure you find someone that can help you with that estimate, pre-qualify you, is something I really recommend uh, you know, prior to any you know claim process.
00:19:16.319 --> 00:19:16.799
Love it.
00:19:16.960 --> 00:19:18.559
Okay, Matt, same question to you.
00:19:18.720 --> 00:19:23.119
What is the first step that CTOs or CFOs in the room should be taking?
00:19:23.599 --> 00:19:35.039
I wish I could just take what Joshua said and repeat it, but maybe the first step that I would do is go ask questions.
00:19:35.759 --> 00:19:45.039
Go there is in addition to the C uh founders, which is a very valuable resource, there are a lot of resources on the web.
00:19:45.200 --> 00:19:54.799
And I know I've said there's limitations to AI, and that is the case, but AI is also a good starting point to get a general sense of the program.
00:19:55.680 --> 00:20:02.400
The other thing I will add to that, though, is this will only take you so far.
00:20:03.440 --> 00:20:08.960
Likely what will come up out of you asking these questions is more questions.
00:20:09.599 --> 00:20:19.839
In fact, if you're asking the right questions, it's almost guaranteed to lead you to more questions, such as, is this project eligible?
00:20:20.079 --> 00:20:21.920
What part of this project is eligible?
00:20:22.160 --> 00:20:26.400
How do I set myself up for success in claiming all of this?
00:20:26.799 --> 00:20:57.440
And at that point, depending on how many of those questions you have, I think is really the time to talk to someone who has a lot of experience in the industry, who's able to kind of talk with you, help you navigate those questions, take a look at your work, and see yes, here's how much likely this can come out to and give you kind of that clarity and transparency on the process.
00:20:59.200 --> 00:21:00.000
I love it.
00:21:00.160 --> 00:21:06.480
All right, so to bring it home, this one's for Joshua specifically, because we're gonna actually broaden it out to talk about Quebec.
00:21:06.640 --> 00:21:10.720
And so, I guess in one sentence, but you can go on for a few more than one sentence.
00:21:10.880 --> 00:21:14.000
Why does this matter for Quebec's innovation economy specifically?
00:21:14.160 --> 00:21:22.160
And I'm talking about the changes to the CDAE program, even having their own provincial programs for digital media and the Crick program that we were talking about.
00:21:22.319 --> 00:21:26.319
Just why is Quebec really pushing hard on all these funding fronts?
00:21:26.400 --> 00:21:28.960
And what does it mean for innovation in the province?
00:21:29.920 --> 00:21:31.279
I think the answer is pretty simple.
00:21:31.440 --> 00:21:34.640
There's no more transformative technology than AI.
00:21:36.079 --> 00:21:47.200
Economies that are able to integrate, adopt, develop AI solutions will see higher prosperity.
00:21:47.359 --> 00:21:58.960
So Quebec is has bet historically, you know, with Yashua Bengio, with Mila, with all the researchers we have in Quebec at UCann, at McGill, Concordia.
00:21:59.119 --> 00:22:02.559
Um, we have a rich history in artificial intelligence.
00:22:02.799 --> 00:22:06.720
One thing Canada is less good at is the commercialization aspect.
00:22:06.960 --> 00:22:08.960
Too many founders head south of the border.
00:22:09.200 --> 00:22:19.599
I hope the strength and the strengthening of these programs rebalances the capital imbalance that we have uh with our neighbors to the south.
00:22:19.759 --> 00:22:27.839
Uh, and so incentivizes builders to stay here and scale here because that's really uh essential to Quebec and Canada's prosperity.
00:22:28.079 --> 00:22:32.160
The government's trying to do something which uh is good to see.
00:22:32.640 --> 00:22:33.680
Will it be enough?
00:22:33.839 --> 00:22:43.039
Uh I don't have a you know uh the the you know as a seeing ball, but I I hope it sure is because it's great to work with builders as they scale.
00:22:43.200 --> 00:22:59.119
Uh and we play a small role in that, but helping them gain additional capital so they can grow faster is certainly something essential, especially in such a capital-intensive business as AI, where we're seeing token costs, GPU costs, uh skyrocket.
00:23:00.480 --> 00:23:06.160
You need to make sure you're you're well funded to compete with the our our once again our our good neighbors south.
00:23:08.799 --> 00:23:14.720
Yeah, to to maybe close this out and add just one final point, and again, this is purely anecdotal.
00:23:14.880 --> 00:23:20.160
I think it provides a little more stability than you can get to the neighbors to the south today, at least.
00:23:20.319 --> 00:23:38.079
Or while things are figuring themselves out on a global platform, Canada is not wasting time to make sure that there's funding in place for businesses that are doing innovative work to get comfortable here on this soil and make sure that they're gonna get the resources they need to see their dreams take flight without them taking flight.
00:23:38.319 --> 00:23:40.640
So gonna leave it at that.
00:00:06.000 --> 00:00:10.320
Hello and welcome back to a very special office hours edition of the What the Tech Podcast Room Boast.
00:00:10.560 --> 00:00:13.679
I'm Paul Dam, Portbost Head of Content and Resident Tech Journalist.
00:00:13.759 --> 00:00:22.719
And this week's episode is part two of a conversation regarding AI and RD tax credits that features both in-house experts Matt Rudishauser and Joshua Martin.
00:00:22.879 --> 00:00:36.880
Check out last week's episode on the feed if you missed part one, and buckle up if you're ready for today's episode, where we dig deeper into how the use of AI is changing RD tax credit claims, and specifically the unique opportunity for innovators to claim provincial credits to fuel their innovation.
00:00:37.119 --> 00:00:42.240
Both myself and Joshua will be in Montreal this week for All In Conference, Canada's largest AI gathering.
00:00:42.399 --> 00:00:47.600
So if you have any questions following this episode and are going to be at the big show, stop by booth B6 to say hi.
00:00:47.759 --> 00:00:50.000
Without further ado, let's get into the interview.
00:00:50.240 --> 00:00:55.280
Okay, so we were talking about CDAE, we were talking about federal shred, and we're also talking about provincial crick.
00:00:55.359 --> 00:01:02.799
Now, Joshua, is there a scenario where a company qualifies for all three, or is it usually just one or two in practice or one here?
00:01:03.039 --> 00:01:04.799
What's kind of the way of the land?
00:01:05.439 --> 00:01:06.719
You know, absolutely.
00:01:06.959 --> 00:01:13.439
So just to clarify, shred uh and crick are applied for in the same application process.
00:01:13.519 --> 00:01:15.920
So crick is just the provincial shred.
00:01:16.239 --> 00:01:23.200
Um, there are differing eligibility criteria in the sense of you can actually include some pre-commercialization activities in Crick.
00:01:23.280 --> 00:01:35.840
So it's more uh uh as you're completing the provincial expenditure forms, you're able to include some additional expenditures, but the submission is still done uh you know in tangent in parallel.
00:01:36.000 --> 00:01:45.040
Uh, and then the the responses are done first federal response, and then following that Hofnie Quebec will omit their decision and the check in consequence.
00:01:45.280 --> 00:01:47.439
You can stack CDAE on top of that as well.
00:01:47.599 --> 00:01:52.000
So, really, there is there are many scenarios where you're stacking and optimizing between different programs.
00:01:52.159 --> 00:01:56.159
It's not at all odd to take advantage of all these different programs.
00:01:56.480 --> 00:02:05.040
One scenario that we see is less common is if you're using uh you know CDA, shred, and multimedia.
00:02:05.200 --> 00:02:14.159
Those three combined are extremely rare because often CDA, the eligibility criteria under CDA and multimedia, you know, don't really go along.
00:02:14.639 --> 00:02:30.800
I think there are some edge cases we've seen, but by and large, you're really gonna pick kind of shred for your uh you know eligible innovation work and then stack on either CDA or multimedia, depending on your business type, business activities, and revenues.
00:02:32.080 --> 00:02:33.280
That makes perfect sense.
00:02:33.360 --> 00:02:35.360
And well noted on the multimedia too.
00:02:35.439 --> 00:02:38.000
I believe the English abbreviation is CTMN.
00:02:38.080 --> 00:02:39.520
Is that correct, Joshua?
00:02:39.919 --> 00:02:41.599
Uh yes, that's the French abbreviation.
00:02:41.919 --> 00:02:43.360
Oh, that's the French abbreviation.
00:02:43.439 --> 00:02:44.400
Okay, that's why you're here.
00:02:44.639 --> 00:02:47.439
Matt, any other notes on that about kind of the stacking strategy?
00:02:47.599 --> 00:02:56.000
I know we covered it last time, but for the provincial programs in Quebec specifically, is it abnormal to take care to take advantage of the three that we named?
00:02:56.159 --> 00:02:58.560
Or to even let's throw in the digital media tax credit.
00:02:58.639 --> 00:03:06.800
How do you kind of see those lay out and maybe even have timelines where it makes more sense for one to be more valuable than the other, volleying back and forth?
00:03:07.199 --> 00:03:12.240
It always makes sense to get as much non-dilutive funding as you can.
00:03:12.479 --> 00:03:27.199
The more sophisticated CFOs out there are definitely the ones who are always looking to optimize their non-dilutive funding, their government tax credit and grant programs.
00:03:27.599 --> 00:03:33.599
So you'll see people go after all this money that is out there that is free.
00:03:33.840 --> 00:03:42.479
The one thing that I will say is it's not so much a differentiation in timeline as to which of these programs you claim.
00:03:42.560 --> 00:03:52.159
And this is part of the reason Joshua was saying you don't see much overlap between the multimedia tax credit and CDAE.
00:03:52.560 --> 00:03:57.039
It's more the type of work that you see being done.
00:03:57.199 --> 00:04:16.720
So the nice thing about Shred is it is industry agnostic, whereas CDAEA is very much AI focused, and the name gives it away, but multimedia tax credit is multimedia focused.
00:04:16.959 --> 00:04:23.279
So that's more the dividing line than uh timeline or those pieces.
00:04:23.600 --> 00:04:37.839
And yes, as Joshua mentioned, there are some points in time, or no, there are some pieces where CDAE, IA, and multimedia can overlap, but there are they are not very common.
00:04:40.079 --> 00:04:40.720
Heard.
00:04:41.040 --> 00:04:51.279
All right, man, I think you kind of alluded to this in the answer that you just gave here, but to put a fine line under it, if a finance leader takes away one thing about stacking these programs, what should it be?
00:04:51.600 --> 00:04:54.160
There's the obvious, it's free money.
00:04:54.319 --> 00:04:59.920
Okay, it's not a hundred percent free money, but for your company, it pretty much is free money.
00:05:00.240 --> 00:05:11.519
The one thing that I would say on top of that is it can get very nuanced, all the different details.
00:05:11.680 --> 00:05:21.920
There's a lot of minor little tweaks that can really move the needle one way or another as to whether you're going to be able to claim something or not.
00:05:22.160 --> 00:05:31.279
Crossing your keys, stotting your eyes really matters here, and so it is good to be on top of all that ahead of the game.
00:05:31.600 --> 00:05:57.759
Doing it, trying to cram things in retroactively, build stuff backwards is a far more time-consuming exercise, and you end up seeing more things that could possibly slip through the cracks, which has a direct impact not just on how much you can claim, but how fast you get the turnaround, because we all know that cash flow matters.
00:05:59.839 --> 00:06:08.319
If I were to add on Matt's point, yeah, free money's great, you know, it's a it's an entitlement.
00:06:09.279 --> 00:06:14.639
My kind of core message would be about thinking strategically or when it comes to tax credits.
00:06:14.800 --> 00:06:18.160
Too few companies wait until the end of the fiscal year.
00:06:18.319 --> 00:06:24.959
They have a really incomplete process which leads to unclaimed entitlement amounts.
00:06:25.279 --> 00:06:35.439
So thinking proactively, mapping out your strategy, understanding which documentation is needed per different credits is the real message here.
00:06:35.759 --> 00:06:40.720
Use software, you know, everyone's everyone, AI is the word of the day.
00:06:40.879 --> 00:06:48.160
Uh, there are tools that can make your life easier to ensure you're claiming these different tax programs.
00:06:48.399 --> 00:06:49.839
It's time to have that conversation.
00:06:49.920 --> 00:06:53.199
It's time to improve your tax credit strategy.
00:06:54.399 --> 00:06:55.120
I love that.
00:06:55.199 --> 00:07:12.319
And I think both of the points that you made lead to the next section because, like Matt was saying about the documentation and making sure that it's not a retroactive activity, something you're doing proactively or prospectively throughout the year, and making sure that you have all that information documented and in place to make the claim.
00:07:12.720 --> 00:07:18.240
And to Josh's point, also about being strategic about it from the start so that you know what you're going to be claiming.
00:07:18.319 --> 00:07:25.279
I mean, or at least what you're going to be tracking, so that you can have a solid claim and a bulletproof claim if there is such a thing.
00:07:25.439 --> 00:07:41.120
So that brings me to audit readiness and kind of common pitfalls as they relate to building the claim, but then after you've already filed, and let's say that one of the agencies that is going to be administering the actual funding comes knocking and they want to know, wait, hold on, something doesn't smell right here.
00:07:41.279 --> 00:07:43.600
So, Matt, gonna go with you first.
00:07:43.759 --> 00:07:54.959
When a stack shred CDAE claim gets reviewed, understanding that these might not be claimed through the same avenue, what does a strong audit position really look like above and beyond what I just rambled through there?
00:07:55.519 --> 00:08:08.639
Yeah, so it is there's a few things to look for, and it's probably better to break this down by the different programs because each program has different nuances.
00:08:08.879 --> 00:08:25.759
On the shred level, there's there is the documentation, but even beyond the documentation, what you want to do is make it easy for the CRA to agree with you and hard for them to disagree with you.
00:08:26.240 --> 00:08:38.720
So what you're trying to basically do is set up your documentation in a way that really outlines that whole narrative that you're going for.
00:08:39.039 --> 00:08:49.759
And the more you can have everything come together as a cohesive picture, the more convincing that picture is to the CRA.
00:08:50.159 --> 00:08:56.799
And the easiest way to do that is not to try and duct tape things together after the fact.
00:08:57.120 --> 00:09:01.679
It's as you build, as you go, this is the reality.
00:09:01.919 --> 00:09:07.759
You're just looking at it, or we're helping you look at it through a shred lens.
00:09:08.080 --> 00:09:16.559
That is a large part of what our team and our platform does, is apply that shred lens to the data you're already correct collecting.
00:09:16.799 --> 00:09:19.440
So that's the dread piece.
00:09:19.919 --> 00:09:41.200
Where I would look at with the CDAE IA piece, that is far more on the financial documentation side, and showing that you have the documentation that actually matches their checklist and all the different pieces that they want to have in place.
00:09:41.919 --> 00:09:48.000
One of the and there's actually and the same type of thing goes for Craig.
00:09:48.639 --> 00:09:55.919
The one other point that I want to make is there's a few different risks with audit.
00:09:56.159 --> 00:10:07.120
The the obvious risk with audit is that the they come in and just cut the whole thing or reduce your claim or whatever.
00:10:07.519 --> 00:10:18.639
But even if they come in and they say yes, everything's good, that still has two negative effects on your company, on the ROI of these claims.
00:10:18.960 --> 00:10:30.639
Because one, you're now spending up to 40, 80 hours, like you can just be you can these audits can be time sinks.
00:10:31.840 --> 00:10:38.080
And then the other thing an audit does is it really can delay when you get your check back.
00:10:38.720 --> 00:10:48.559
And depending on how you've set up your finances, that might mean that you need to get bridge financing in place, which is an additional cost.
00:10:49.200 --> 00:11:00.080
And so knowing the audit risks, knowing that, hey, you're covered in case there is an audit, is something that is valuable.
00:11:01.759 --> 00:11:02.639
I love that.
00:11:02.879 --> 00:11:16.080
So to recap a little bit there, knowing the risks is one part of it, being educated on what could potentially happen to you, because again, free money is only valuable if you receive the free money and on a timeline that you can actually spend it.
00:11:16.240 --> 00:11:20.720
So you want to avoid scenarios where you're gonna get bogged down and then not have a check come out.
00:11:20.879 --> 00:11:27.919
But to that other point about the time commitment too, it's about knowing the risks, and it's also about having risk coverage, more or less.
00:11:28.159 --> 00:11:33.759
So, again, not to say that both is a traditionally insurance for your RD claim.
00:11:33.919 --> 00:11:41.039
We do just make sure that we have everything you need to back up that claim because we've been there before and we know how to navigate these realities.
00:11:41.279 --> 00:11:51.039
Now, you also talked a little bit about the nuances between the Shred program, which is a lot more about kind of that technical qualification, making sure activities themselves meet that criteria.
00:11:51.279 --> 00:11:57.440
Whereas CDIE, you touched on, is a little bit more about making sure on the finance side everything makes sense and everything adds up.
00:11:57.600 --> 00:12:01.600
Josh, I want to turn to you a little bit to pull on that thread and expand on it a little bit more.
00:12:01.759 --> 00:12:07.519
Is there anything Quebec-specific that trips companies up during an IQ or Revenue Quebec review?
00:12:09.679 --> 00:12:18.720
So it's very much, I think the key differentiator between Shred and Revenue Quebec reviews is the frequency.
00:12:18.879 --> 00:12:26.159
You know, Shred is one every, you know, let's say four to five years is kind of the general rule of thumb.
00:12:26.240 --> 00:12:30.000
They're gonna validate your technology, your expenditures, and make sure that it fits.
00:12:30.159 --> 00:12:34.480
And there's on a gray when it comes to a shred uh a shred uh audit.
00:12:34.559 --> 00:12:37.120
You know, how has the technology evolved?
00:12:37.360 --> 00:12:44.159
Is your works does that still meet that kind of high threshold of uncertainty that's really required for a shred claim?
00:12:44.320 --> 00:12:54.879
And then our mandate as your consultant is to really uh you know fight and show through documentation and current practices how you still meet that that threshold.
00:12:55.200 --> 00:13:00.720
Revenue Quebec reviews when it comes to CDA or CDA C D A E IA are totally different.
00:13:01.120 --> 00:13:04.240
It's a much more narrow credit.
00:13:04.559 --> 00:13:11.759
You either meet the the you know required uh uh you know activities and employee count, or you don't.
00:13:12.000 --> 00:13:14.879
The nuance is often in the documentation.
00:13:15.200 --> 00:13:25.919
So do you have proof that those you know six eligible employees actually did spend a certain amount of their time doing the required activity types?
00:13:26.240 --> 00:13:38.080
Do you have proof that your revenues actually were derived from these business codes that are really detailed on the Revenue Quebec's website and FST Small Quebec's website as well?
00:13:38.399 --> 00:13:51.759
So it's more about clarifying and showing using documentation to show how you fit into the the required boxes that uh revenue Quebec and FST Small Quebec lay out.
00:13:51.919 --> 00:14:03.039
Whereas for Shred, it's often, I don't want to say it's more theoretical, but there there's almost like uh an intellectual component to it where it's you know, how's technology evolved?
00:14:03.200 --> 00:14:07.279
How does this meet and surpass the ways technology has evolved?
00:14:07.440 --> 00:14:08.480
Why is it unique?
00:14:08.639 --> 00:14:22.799
You know, it's a lot more there's a lot more gray to play with, whereas a lot of the times we can predict when a claim is going to have trouble when it comes to CDAIA because of the you know clarity around the credit.
00:14:22.960 --> 00:14:34.639
Obviously, this year with the introduction of CDA IA, there's more uncertainty in the program, but nonetheless, documentation kind of allows us to ensure that we're claiming every dollar that companies are entitled to.
00:14:35.279 --> 00:14:47.919
Just to add to that, even though the program has significantly changed, the mindset of Investissement Quebec and Revenue Quebec has not.
00:14:48.399 --> 00:14:59.360
So the same mindset for reviews that they used with CDE CDAE, they're gonna use for CDEAE AI or IA.
00:15:00.960 --> 00:15:09.600
Cat, yeah, and I think another point that I'm just gonna pull on from my content perch is when you're watching the news, you hear a lot about agencies themselves using AI.
00:15:09.759 --> 00:15:23.919
I don't think that changes necessarily, and this is just anecdotal coming from me, the burden that you need to put into creating a compelling narrative with your RD claim, with the shred claim specifically, because it is a little bit of storytelling, like they were saying here.
00:15:24.000 --> 00:15:29.440
Um, you need to make sure that you're saying what has actually changed in a way that, in human words, are convincing.
00:15:29.679 --> 00:15:36.799
So even if they're deploying their own tools, it's not taking the human or the narrative or the storytelling component out of the equation at all.
00:15:36.960 --> 00:15:50.799
It might be speeding up the process in the same way that our own team uses AI or the AI-enabled tools to speed up our collection and make sure that we can claim things and make sure that we have the data that we need closer at hand to build your claim.
00:15:50.879 --> 00:15:54.320
But that doesn't mean that a human is coming out of the equation full stop.
00:15:54.559 --> 00:15:55.440
Go ahead, Matt.
00:15:55.759 --> 00:16:06.240
Just one story to underline that point is so a company that we know of, they used AI to put together their thread claim.
00:16:06.480 --> 00:16:07.919
It went to audit.
00:16:08.159 --> 00:16:14.080
The auditor used AI to underline why it wasn't tread eligible.
00:16:14.320 --> 00:16:23.840
Now, when they came to us, so they were able to get Claude to say this is thread, and then the CRA got Claude to say no, this isn't tread.
00:16:24.000 --> 00:16:29.200
And so when they came to us, we were able to go back and say, you know what?
00:16:29.440 --> 00:16:30.559
That's Claude.
00:16:31.039 --> 00:16:41.120
Here's the actual human reasoning, the human intelligence, the HI that finally managed to get that to go through Audit.
00:16:42.799 --> 00:16:43.840
I like HI.
00:16:44.080 --> 00:16:47.039
I'm gonna make that happen, at least down here in the States.
00:16:47.120 --> 00:16:56.720
And also from my content perch, what you just described with the two Clauds talking to each other is my worst nightmare because again, that's my whole industry going out the window right there.
00:16:56.799 --> 00:17:01.120
But I'm not gonna digress because that's why we're three guys talking to each other here.
00:17:01.279 --> 00:17:04.960
We're the human intelligence, not behind the entire boast product.
00:17:05.039 --> 00:17:08.319
I'm certainly not, but that you need to really make that difference.
00:17:08.400 --> 00:17:13.680
And again, tell a compelling story because we're ultimately serving humans at the end of the day, not AI.
00:17:13.839 --> 00:17:19.440
So yeah, I'll digress on that, but I think it's an important point that I want everyone here to hear now.
00:17:20.000 --> 00:17:22.480
All right, so guys, we're moving into the closing questions.
00:17:22.640 --> 00:17:31.039
So we're gonna be covering a lot of ground here that we already covered, but again, points that I think bear repeating and that are gonna be good for this audience to really take home with them.
00:17:31.200 --> 00:17:33.200
So, closing question for you both.
00:17:33.359 --> 00:17:35.359
Joshua, I'm gonna point this at you to begin with.
00:17:35.519 --> 00:17:44.000
But if a CFL or CTO watching this is building AI into their product right now and has never looked into any of these programs, what's the first step?
00:17:45.759 --> 00:17:48.400
I know it can seem overwhelming at first, right?
00:17:48.640 --> 00:17:57.759
You have Freddy, multimedia, plethora of grants to also complement these tax credits.
00:17:58.160 --> 00:18:06.160
A good starting point is talking to other founders and hearing what their experiences have been, what they recommend.
00:18:06.400 --> 00:18:16.079
Obviously, consultants are going to be pushing you in one way or another, but I think it's good to start by hearing what experiences other founders have had or other CTOs.
00:18:16.400 --> 00:18:20.960
From there, the likely solution is understanding, okay, can I do this in-house?
00:18:21.039 --> 00:18:22.960
Is it worth using a consultant?
00:18:23.279 --> 00:18:32.720
For more companies than not, using a consultant is the right approach, just simply for the reason that you know we all have to do this, to-do lists that are that are too long.
00:18:32.799 --> 00:18:36.400
So adding something to it means you likely won't end up getting it done.
00:18:36.559 --> 00:18:44.720
Uh, and then I would start by finding a consultant that's able to really pre-qualify your work and start with their estimate for you.
00:18:44.880 --> 00:18:50.559
That estimation process allows you as an operator to understand, okay, great, is this worth my time?
00:18:50.640 --> 00:18:57.200
If I'm gonna commit 20 hours of resources, which is often the case, what's gonna be my ROI at the end of the day?
00:18:57.279 --> 00:19:00.480
If it's only for$15,000, does it make sense to claim?
00:19:00.640 --> 00:19:01.519
Likely not.
00:19:01.759 --> 00:19:05.599
But often with tread CDAE, the claim size is much larger.
00:19:05.759 --> 00:19:15.119
So really making sure you find someone that can help you with that estimate, pre-qualify you, is something I really recommend uh, you know, prior to any you know claim process.
00:19:16.319 --> 00:19:16.799
Love it.
00:19:16.960 --> 00:19:18.559
Okay, Matt, same question to you.
00:19:18.720 --> 00:19:23.119
What is the first step that CTOs or CFOs in the room should be taking?
00:19:23.599 --> 00:19:35.039
I wish I could just take what Joshua said and repeat it, but maybe the first step that I would do is go ask questions.
00:19:35.759 --> 00:19:45.039
Go there is in addition to the C uh founders, which is a very valuable resource, there are a lot of resources on the web.
00:19:45.200 --> 00:19:54.799
And I know I've said there's limitations to AI, and that is the case, but AI is also a good starting point to get a general sense of the program.
00:19:55.680 --> 00:20:02.400
The other thing I will add to that, though, is this will only take you so far.
00:20:03.440 --> 00:20:08.960
Likely what will come up out of you asking these questions is more questions.
00:20:09.599 --> 00:20:19.839
In fact, if you're asking the right questions, it's almost guaranteed to lead you to more questions, such as, is this project eligible?
00:20:20.079 --> 00:20:21.920
What part of this project is eligible?
00:20:22.160 --> 00:20:26.400
How do I set myself up for success in claiming all of this?
00:20:26.799 --> 00:20:57.440
And at that point, depending on how many of those questions you have, I think is really the time to talk to someone who has a lot of experience in the industry, who's able to kind of talk with you, help you navigate those questions, take a look at your work, and see yes, here's how much likely this can come out to and give you kind of that clarity and transparency on the process.
00:20:59.200 --> 00:21:00.000
I love it.
00:21:00.160 --> 00:21:06.480
All right, so to bring it home, this one's for Joshua specifically, because we're gonna actually broaden it out to talk about Quebec.
00:21:06.640 --> 00:21:10.720
And so, I guess in one sentence, but you can go on for a few more than one sentence.
00:21:10.880 --> 00:21:14.000
Why does this matter for Quebec's innovation economy specifically?
00:21:14.160 --> 00:21:22.160
And I'm talking about the changes to the CDAE program, even having their own provincial programs for digital media and the Crick program that we were talking about.
00:21:22.319 --> 00:21:26.319
Just why is Quebec really pushing hard on all these funding fronts?
00:21:26.400 --> 00:21:28.960
And what does it mean for innovation in the province?
00:21:29.920 --> 00:21:31.279
I think the answer is pretty simple.
00:21:31.440 --> 00:21:34.640
There's no more transformative technology than AI.
00:21:36.079 --> 00:21:47.200
Economies that are able to integrate, adopt, develop AI solutions will see higher prosperity.
00:21:47.359 --> 00:21:58.960
So Quebec is has bet historically, you know, with Yashua Bengio, with Mila, with all the researchers we have in Quebec at UCann, at McGill, Concordia.
00:21:59.119 --> 00:22:02.559
Um, we have a rich history in artificial intelligence.
00:22:02.799 --> 00:22:06.720
One thing Canada is less good at is the commercialization aspect.
00:22:06.960 --> 00:22:08.960
Too many founders head south of the border.
00:22:09.200 --> 00:22:19.599
I hope the strength and the strengthening of these programs rebalances the capital imbalance that we have uh with our neighbors to the south.
00:22:19.759 --> 00:22:27.839
Uh, and so incentivizes builders to stay here and scale here because that's really uh essential to Quebec and Canada's prosperity.
00:22:28.079 --> 00:22:32.160
The government's trying to do something which uh is good to see.
00:22:32.640 --> 00:22:33.680
Will it be enough?
00:22:33.839 --> 00:22:43.039
Uh I don't have a you know uh the the you know as a seeing ball, but I I hope it sure is because it's great to work with builders as they scale.
00:22:43.200 --> 00:22:59.119
Uh and we play a small role in that, but helping them gain additional capital so they can grow faster is certainly something essential, especially in such a capital-intensive business as AI, where we're seeing token costs, GPU costs, uh skyrocket.
00:23:00.480 --> 00:23:06.160
You need to make sure you're you're well funded to compete with the our our once again our our good neighbors south.
00:23:08.799 --> 00:23:14.720
Yeah, to to maybe close this out and add just one final point, and again, this is purely anecdotal.
00:23:14.880 --> 00:23:20.160
I think it provides a little more stability than you can get to the neighbors to the south today, at least.
00:23:20.319 --> 00:23:38.079
Or while things are figuring themselves out on a global platform, Canada is not wasting time to make sure that there's funding in place for businesses that are doing innovative work to get comfortable here on this soil and make sure that they're gonna get the resources they need to see their dreams take flight without them taking flight.
00:23:38.319 --> 00:23:40.640
So gonna leave it at that.
