AI for Non-Profits with Mitch Schwartz

---

Webinar Transcript: Building an AI Plan for Nonprofits

Speakers: Mohamed Hamad (Host, Mohamed) and Mitch Schwartz (Mitch)

Segment 1: Welcome and Introduction

Mohamed:

Good morning everybody and welcome to today’s webinar. How’s everybody doing? We’ve got a chat on the left there if you want to pop in your messages and tell us how you’re doing this morning. My name is Mohamed Hamad. I’m the founder of Third Wunder.

Mohamed:

And with me here today is Mitch Schwartz founder of OpsMachine. And today’s topic is building an AI Plan for nonprofits. It’s something that is near and dear to our hearts as, organizations that worked very closely with nonprofits in the last five to ten years.

Mohamed:

And everybody’s talking about AI how to work with AI and how to make sure that you’ve got a plan for it. Mitch how are you doing this morning?

Mitch:

Oh doing great. Really excited to be here. It’s so meaningful right, Because I’m not a frontliner. And I respect so much the people who are. Who are really diving in and doing the work. So when I can be there and help and lift some weight off of people’s shoulders or build them wings and maybe a catapult or something then it’s a huge joy.

Mitch:

You really get to see. See the scale of the work you’re doing. And, and I guess, you know, what are, what are we all doing here if we’re just chasing dollars? So how do we how do we help people with what we do?

Mohamed:

Absolutely. I mean some of the work we’ve done with nonprofits, and helping them really catapult themselves out there and get their message out is a point of pride in saying that, you know, when it comes to AI, AI is quite the accelerator.

Mohamed:

It can help lots of organizations do wonders with the work that they do. But as always, with every tool, you know, there’s pitfalls there’s caveats there’s things to do with it things not to do with it things that you should know before you get into it.

Mohamed:

And, before we get started, I kind of want to know from, from the audience here, how many of you, use, AI on a daily basis for work? I’m going to throw out a poll out here if you guys can, give us a couple of answers.

Mohamed:

The big question here is, do you use AI for work? Yes or no? This is just to give us a grounding to understand, where we’re sitting at and, how we want to answer some of the questions that will come out later. Just to be clear, at the end of this session, towards the end of the session, we’re going to have some time for a Q and A, so you can throw in your questions in the chat there and that’ll help us really, get down to the details and answer your questions and concerns.

Mohamed:

So if you can answer this poll just to get the ball rolling. While that’s happening, I think one thing that I would like to cover is that one of the reasons we started this, we wanted to have this session, with our collaborators at the chssn.

Mohamed:

We were looking at a report from the Toronto, University, the tmu and they came up, they had a few stats that were very interesting. The main stat here is that.

Mohamed:

I’m just going to put this away. The main stat here is that 80% of Canadian nonprofits are using AI in some form.

Mohamed:

The question of what AI is was very broad and that spanned the spectrum of things from autocomplete tools and things like Grammarly all, all the way out to actual AI tools like Gemini and Claude and Microsoft Copilot and whatever it is that people are using, for their purposes.

Mohamed:

The only thing is that only 10% actually had formal policies around it. And 60, 4% use AI without any guidelines. And we realized that there needs to be very strong guidelines around AI1 because there’s issues of security, privacy, bias, and how do you get consistent outcomes from your AI, or your, the tools that you have.

Segment 2: Identifying the Organizational Challenge

Mitch:

Yeah, just let me jump in right there and like go one level higher. As I was doing a certification, I was with a whole bunch of employees working for, not for profits. And there was this central tension that just kept coming up. And the theme was we have this responsibility to make the most of the funds we get and to explore tools that can help us be more effective and figure out how to implement them.

Mitch:

Even when we’re already spread so thin and barely have the capacity to figure out, to try to do that. And at the same time it’s not just some companies. So if we leak data, if we write posts that really fall, that are like, forget falling flat, but that are offensive or against the community values, we run the risk of betraying our communities.

Mitch:

And if we’re doing that, why do we even exist? So just to speak to. And we see that a little bit in this numbers. There’s a knowledge that we need to figure out the way forward. There’s a lack of Institutionalizing it and really being formal about it. And there’s a huge existential risk and it’s not an easy position.

Mitch:

And so much of the advice on the Internet really targets companies that can like move fast and break some stuff and it’ll be okay. But this isn’t that situation.

Mohamed:

Yeah, absolutely. And you know we, we have all come to the realization that AI can sometimes hallucinate in certain cases. So even if the intention is to be factual and honest and open, you know, when you feed something in there and it doesn’t quite know what to say or has been, it can make up stuff and sometimes misrepresent having guidelines and having strong knowledge of what to do with it and how to do it, how to use it and when can really alleviate that pressure and gives you a sense of certainty when you’re using a tool.

---

Segment 3: The Six Principles for AI Adoption

Mohamed:

So what we want to talk today about is what we believe is six principles that you know, everybody should kind of consider when thinking about creating or using their AI.

Mohamed:

And the principles really are when building out a plan for AI. And this is something Mitch tells me all the time is to audit your day to day life. Really get in and understand what are the small bottlenecks, what are the small things that you can offload to an AI because it can do it faster, better and that reclaim time you can use to do bigger things in your life.

Mitch:

Now one of the key areas where I find people get stuck here is I’ll give an example of a creative agency I worked with. So they would run a whole workshop and then at the end they would have the job of translating everything that happened when they were facilitating and in the moment and really connecting into a playbook, which means they were pretty much doing the job twice.

Mitch:

And well, you can all imagine that’s not the funnest. So when they turn to AI and ChatGPT, they found these results suck. We’re going in circles. Or maybe suck is strong. But they weren’t satisfied with it. It was lacking the thing and they gotten that A.I. whack a mole. So it can be really easy to say, hey, I have this large task, I want to make a whole playbook.

Mitch:

Hey AI do it. I want to take this entire transcript and turn it into, a thoughtful report. Hey, AI, do it. And that eventually can work, but usually it takes building up to it. So when we talk about like auditing your day to day, once you identify the task, it’s just critical to really break it into tiny steps and imagine if it were like a tiny production line.

Mitch:

So for example, when I was working with that team that I mentioned before, it was there were like eight formal steps that went into it. You know, one of them was identifying, the, you know, the high level takeaway. One of them was identifying the brand voice, one was identifying motifs and themes.

Mitch:

So trying to. One shot that really didn’t work and made their life unenjoyable when they tried to do it. But going step by step and building on it really worked well. So, sometimes we try to automate from A to Z, but if we just automate the most painful part, A to B, we can get like 50% of the value and start building it out.

Mohamed:

Yeah. And it allows us also to really think of the process as a whole. Because a lot of the times we’re subject matter experts, we’re the ones that have been doing this thing for 5, 10, however long you’ve been in your career. And it becomes muscle memory.

Mohamed:

And we always forget the layers that go into doing something. And when we sit back and just kind of take a moment to break it down, it really starts to shape our understanding of how complex certain things are. But also it’s a great exercise to be able to document things so that you can have a transfer of knowledge to other people.

Mohamed:

Like if you’re setting yourself up to delegate, to, let’s say a team member or colleague or a new employee, or volunteer, that foundation there, you’re like, okay, cool. Like this process, end to end, is X, Y and Z. And this can be done by AI and this can be done by someone else.

Mohamed:

And this is how you check for it and this is how you do it. And this is the end product. So that is a great way to, really understand your processes in general.

Mitch:

Which brought us smoothly through number two.

Mohamed:

Yeah, this is, transformational, at least on my end, going deep and really rethinking what your organization does and how. And the reason you do that is to give context and context.

Mohamed:

With AI is everything. You know, what do you, what do you think Mitch?

Mitch:

Yeah. Well, so the, the day that I became a believer, it was, sometime last year. I was watching. So I was watching Tiago Forte and he’s the building a Second brain guy. And, and, you know, he’s credible and he’s great. And he. With some other guy who I’ve never heard of before, but this guy runs 12 companies from the woods. He’s the creator’s dream. They met playing pickleball, that whole story. So anyways, he says, look, here’s how we’re saving tons and tons of effort with AI, right now.

Mitch:

He went into his cloud, which was pre set up, and he said, write me a job description for, you know, sales role. And it started putting out the most detailed job description I’ve ever seen. It said, you know, here’s the salary, here are the responsibilities, here’s the bonus structure, here are the people on your team, here are the meetings that you’re going to have each week.

Mitch:

Here are the expected results for an A player, a B player. And it just went on. And I thought, like, how are you doing this? Because that’s crazy. He showed us the prompt and it really just had all of his company’s information clearly spelled out in there. And so he said, this used to take two people two weeks.

Mitch:

Now, I don’t care if that took one person two days, it happened in 15 minutes. And he said, sure, I have to review it, make sure the numbers are good. But the savings, it’s obvious, it’s there. So when the documentation is clear, when it’s kept up to date, which is another challenge, and when it’s put in a way that’s clear and easy for the AI to understand it, then, then things really can be, you know, 100.

Mitch:

We’re talking 100x in this case. And it’s not, exaggerating or, or just clickbait to say that.

Mohamed:

Yeah, I mean, it really, having clear documentation about everything about the organization really makes a big difference in the output, especially when it comes to the nuance of a mandate or a mission. And us as an organization, we have to take a step back as well and look at what is it that we do, why we do it, who we want to serve, where do we want to serve people, what do we want them to get out of us as an organization?

Mohamed:

And that was a foundational moment there, because then you can also get your team very aligned on those specificities, on those details, and by creating that organizational DNA and you pass it on to an AI. If you’re just even just prompting a standard ChatGPT or Claude or Gemini, Having that context really shapes the you know, shapes the answer to be closer to what you’re expecting.

Mohamed:

But at the same time you can also train custom GPTs which are variations of your chat GPT or Gemini calls them gems, you’ve got Claude, they call them projects. You can create base information in there that it always references.

Mohamed:

When it does that, you don’t always have to repeat yourself. It’s foundational to the tool there.

Mitch:

So I feel like we’re also actually, these are laid out pretty well because we’re kind of getting smoothly from like document into. But what can’t it do, right? What should never be crossed? And how do you establish those guardrails now? It’s my. Because my first intuition here is to go, okay, that’s really important.

Mitch:

But we don’t even have time to unpack it today. It’s such a huge topic. I wonder, do you have any tidbits to drop around creating guardrails?

Mohamed:

Guardrails are extremely important because one, it depends on, the tools that you’re using to make sure that you understand the difference between, let’s say a free version of ChatGPT or any of these tools, any of these tools that are free and you’re not paying for it will use your data to train the larger model.

Mohamed:

If you use a free version of ChatGPT and you put some personal information in there, or company information that is internal that might leak out. And we’ve seen horror stories in the press or in the news about things that get leaked up in conversations with the free version.

Mohamed:

If you’re paying for it and it’s part of, let’s say, your organizational structure, let’s say you’ve got Google Workspace and you’ve got Gemini built in, because you’re paying for workspace, it has some guardrails there. Because as an organization there is, GDPR considerations in the US for medical, you’ve got HIPAA here in Quebec, we’ve got Law 25 and all of that.

Mohamed:

And these larger organizations, if you’re using a work based, instance of their tools, also have to comply with that. But that doesn’t mean you can go off and throw in anything that you want. First off, any of these rules, you should never share internal sensitive information, anything relating to identifiable personal information, like people’s salaries, for instance, people’s addresses, phone numbers and all of that client information.

Mohamed:

If you can sanitize it, that’s great. But keep things very broad. If something is Publicly available. You can work on that.

Mitch:

Yeah, there are ways to set up your own custom private thing that’s way outside of the scope and it is more advanced for most teams. So within the tools that already exist, just keep in mind that when using a custom GPT or something, you’re programming saying never do this is good, but it’s not an absolute safety net.

Mitch:

So whatever you tell the AI might get taken as a suggestion, it might get superseded by an entire conversation and become less and less important. So it’s really critical just to keep in mind that you know pretty much the hard rules and when it comes to these safety nets or it comes to guardrails within the organization policies and just shared understanding of what is and isn’t safe AI behavior is so critical and that is where that stat, you know, no, no criticism and just a reality but that’s where that stat of the gap between usage and policy is a bit frighten, tightening and where it does create risks.

Mohamed:

Yeah. Just to get a sense of how people feel about policies, if they have any policies. I’m going to throw out a poll over here and let me know if do you guys currently have any policies internally or are working on it.

Mohamed:

I see a few of you there are still working on it.

Mitch:

One yes. Come on, there’s one. Yes out there.

Mohamed:

We need to get these policies out there.

Mitch:

It’s like election night. We’re just watching this thing and it’s kind of fun seeing it come through, right?

Mohamed:

Yeah, it is, yeah.

Mohamed:

So it turns out a lot of you do not have policies. This is something that’s very important to make sure internally, you do not share sensitive information. What kind of information is out there? Speaking of the type of information, what does AI work best for and what you should never.

Mohamed:

It doesn’t really work well for in my experience, working with numbers, data calculations, math is a bit of hit and miss with some AIs, not all of them. Some are better than others but as a general rule I try to avoid trying to do any mass calculations with AI.

Mohamed:

What’s your experience there, Mitch?

Mitch:

Yeah, so I, I would say, you know, I think someone said, I don’t know if it was like in the last century, like if they ever make a thinking computer, I think it’ll be a lot like people where it’ll need a pocket calculator to be good at math. And that’s exactly what happened. It’s kind of a bit squishy. Brained, like we can be. So I think when it lays out numbers it’s really easy to just get lulled into not checking them and that can be a danger. So one trick could be at this point they can create spreadsheets. So you could say, create a spreadsheet that models this and then you can go and double check the work really easily and even play with the model anyways rather than having it recalculate it.

Mitch:

But yeah, I wouldn’t take most of the math at face value. Even though it may be right 80 or 90% of the time, that’s frighteningly not enough.

Mohamed:

Exactly. Especially if you’re trying to create a report of some sort and you gave it a bunch of numbers and at face value it looks right. But if someone comes in and digs in into it, that could be a point of contention. Or it can work against you in certain ways.

Mohamed:

There is copilot Claude in Excel, there is Gemini in Google, sheets. They can help you with formulas, assist you in finding the right formula to get the right outcome that you want. But on my part I don’t really trust it with numbers.

Mohamed:

Like I tested this out the other week and asked it to count the number of Rs in the word strawberry and it was still struggling with that.

Mitch:

Yeah, I know Scott’s saying the assumption that the numbers would have been the easiest thing to trust and yeah, that’s as it turns out. While it’s really important not to think of it as a human being, the thought process is much more like a human than a calculator. And the way it does come up with, you know, it does pattern matching, it doesn’t do hard calculation. So that’s where it, it appears like it would be good and can be confident but can go right off of the rails.

Mohamed:

Yeah.

Mitch:

All right, so we’ve got seven minutes to get through three points. I think we can do it.

Mohamed:

Yeah. The best tool is the one in your hand. I think this one’s a bit of a controversial one. I think that. And this is Before I started my organization I was a photographer and the old adage there is the best camera is the one that’s in your hand.

Mohamed:

And I’m adapting it here. It’s easier to use the tool that is closest to you if you’re in Google Docs, for instance. And there’s Gemini right there. Copying, pasting and moving things around to another LLM to do something and come back that round trip is effort wasted, especially if it’s small things.

Mohamed:

Now there are tools that are very specific and specialized for certain things. If you need special work, done, let’s say, you need image generation for instance.

Mohamed:

There are ones that are really good at image generation and you won’t get the six fingers on hands type of situation. There are ones that are better at video. But for generalist approaches, I would say the one that you have around is the one that you should use the most.

Mohamed:

I’m going to throw out another poll here just to also get a sense of which tools people are currently using.

Mitch:

So while you do that I’ll throw in. So the best tool is the one in your hand because change management and getting everyone aligned and with a policy and understanding how to use the tool and trained is a huge effort. For example, and so far I guess I’m in safe audience to say this copilot’s generally considered one of the more difficult, I’m saying this diplomatically ones to work with.

Mitch:

But at the same time I think it’s really useful if you have, you know, whatever set up in your org, then just considering it as a skill problem first. Right? Like hey, I’m not getting what I want from the AI, so this is a skill problem. How can I get it better?

Mitch:

You know, way back, my dad took me out to a driving range and he picked up my little brother’s like you know, two foot long driver and he hit the ball like 200 plus yards and he said, you know, it’s not the driver, it’s the golf swing. And I think there’s enough truth in that that at least by really pushing through and getting that honest picture of where the skill is then you can find out, you know, what the tool lacks that you really need and then make a more structured decision.

Mohamed:

Yeah, absolutely. I think sometimes learning the tool that you have and learning the nuances behind it after a while just makes it a lot faster for you to be able to use that. I see majority of people are using ChatGPT, which is which is interesting. There’s all the other ones controversial now.

Mohamed:

Yeah, yeah. But in saying that, I mean it was the first mover and that was the thing that everybody got used to and has most of their information and all of that in there and it is a great tool for sure. We were talking about the great tool.

Mohamed:

How do we make sure that we prompt these LLMs to the best of its capabilities. How do we get it to give us what we want more consistently? And that’s the next one is grounding your prompting. You showed me this, Ricoh.

Mitch:

Yeah. Yes. I mean it’s pretty straightforward, right? We’re creating a certain structure that makes it really easy. So essentially what’s the role? Right? What is the instruction, what is the context and what’s the expected output?

Mitch:

So if we go through those, an example of a role is like you’re a trained, you know you’re a senior developer, you’re a senior editor, you’re, you know, we always throw that senior in there. Or we could otherwise say like you’re this person and do work like they would do think like they would think. The instruction is, you know, the task is we’re creating a social media calendar.

Mitch:

So what we want to do is communicate X and Y and make sure that we reinforce that message. The context which we discussed and document your organizational data is you know, all the company information, the information about the task. You know, there’s a whole lot we can get into about how to structure that well.

Mitch:

But giving context will transform it. And tip. Typically too much is going to be better than too little. All the same, within reason. And then output, you know, give it to me in a spreadsheet with these columns and these specific you know, fields, or give it to me in a bulleted list or put it out in prose.

Mitch:

So if you get those elements, those are going to be really strong. And you know, the reality is don’t write your own prompts. So you can, you can, you know, use that to recognize what a good prompt is. But fundamentally going into ChatGPT and saying, you know, you can use this at the meta level, do a RICO thing and say, hey, write me a prompt for this next bot.

Mitch:

You can approach it at different levels, but at this point, I would say just don’t even bother writing your own prompts. But you can modify them and understand what’s good about them. The context you will have to provide and the output.

Mohamed:

Yeah, to reiterate, RICO was role, instruction, context and output. So the role is who this person is supposed to be. So what is the AI supposed to be? Trying to emulate the instruction is what you want it to do.

Mohamed:

The context is what is this for? And in which setting is this going to end up being? And output is the structure of what you want. Is this a report? Is this a haiku is this a, social media post? Like what do you want and what do you expect?

Mitch:

All right, so that brings us to iteration, which is just to say, as I mentioned earlier, trying to do everything in one shot rarely works well. You get to the one shots by building out the pieces, seeing how they work, coming back to it, talking with your team, getting feedback, putting some structure around that.

Mitch:

Basically yeah.

Mohamed:

And I think the iteration is iteration while you’re working with it and you learn a lot from iteration, but also iteration in the longer view. Iteration on your organizational documents for instance. Right. Going back into them and saying, okay, is this still valid after we’ve used it with a for X amount of time?

Mohamed:

Is it still. Did we miss anything that we can add to this that will make it easier for it to understand the context of who we are and what we serve, have our products or services change, has our audience change, has our reach changed? Updating that on a regular basis.

Mohamed:

And at that point you can actually use the AI because it’s got a formal grounding. You can say, hey, can you update this document to include X, Y and Z to make sure that it’s easy for you to understand in the future or

Mitch:

even build learning and memory and that into the way you’re working as well. So as you iterate, you can really get to systems that are self learning and self enhancing too, which is great because it saves you tons of pouring through documents that frankly no one probably ever wants to do anyways.

Mohamed:

Having the AI do that heavy lifting, and assistive work, really makes it easier to work with it in the long run. And at the same time someone new to the organization or learning the ropes has a foundation now that they can hit the ground running where if they start using it because it’s grounded in memory, it’s grounded in the foundation of who the organization is and has got its own guidelines married with your own guardrails and do’s and don’ts and how to use it at that point now you have a more foundational structure, to continue working with an A.I.

Mohamed:

with, with any of your tools.

---

Segment 4: Q&A and Closing

Mitch:

So I notice we’re at 12:02, which means it’s time to maybe wrap up and get to Q and A.

Mohamed:

Yes, I’d love to hear if you have any questions. I do see a question here from Gordon. How does NotebookLM fit in this? I love NotebookLM and for those who don’t know what NotebookLM it is a product from Google, which basically creates a notebook that you can add, documents to it, YouTube videos, transcripts.

Mohamed:

And how it works is that it’s an AI based on that notebook. Whatever you put into it is what it sources. So it doesn’t go off and search the Internet, it doesn’t try to find things outside of what’s in there. The best thing about it is that it can take whatever information that you have in there and it can, convert it into other things.

Mohamed:

So you can turn it into a podcast, you can turn it into a slide deck, a mind map, flashcards or quizzes so that you know you’re trying to figure things. If you’re trying to learn something, it makes it easy for you to quiz yourself with it. It can create videos out of it.

Mohamed:

This I find incredibly useful for people with accessibility needs. Some people like to read long documents. I like to listen to things. So sometimes I get a long 40, 50 page report and I’ll just put it down through NotebookLM and say, turn it into a podcast and I’ll listen to it while I go out for a walk.

Mohamed:

And it’s fantastic. I get, I get everything I need and then, if I need some details, I can always talk to it and ask it questions.

Mitch:

I love it. For catching up on meetings I’ve missed, for, you know, when, when I’m doing consulting gigs and it’s been two weeks since I spoke with that client. Take the last transcript, make a five minute podcast of it, get caught up. It just soaks my brain right back in the subject matter. I can come in fresh.

Mohamed:

Oh, absolutely. Anybody else have any questions here?

Mitch:

Yeah, maybe while we wait for that.

Mohamed:

Yes. So, Mitch and I, have been working on a program that, that we provide to nonprofits. It is, we’re calling it the Jumpstart. To get you jump started on your, your AI framework and, and how we work on this is we have a four hour session.

Mohamed:

We work with you to define your usage, how you use AI, where you use it, what tools you have, and then we create guidelines, internal documentation, and training material to help you really make the most out of your AI, and make the most out of your tools.

Mitch:

Yeah. The goal is, basically flipping on the light switch, you know, for a lot of people, and then, you know, wherever it is that, that you are right now, figuring out what’s next. You know, one of my specialties is taking things that feel complicated and, and turning them into simpler, actionable steps.

Mitch:

So, that’s what I really love to bring to the teams that we work with.

Mohamed:

Absolutely. And as someone that has, worked a lot with, nonprofits and getting them onboarded with a lot of technology, whether it is, the communications and web presence and digital presence in general, these tools, when worked well, can really speed things up, make things more efficient to a certain degree and empower you to do a lot more and the more time you have.

Mohamed:

Everybody’s strapped for time these days. You know, like there’s just so much going on at any given point, that any extra moment of time or any efficiencies, help out with budgeting and making sure whatever resources you have are put towards your community.

Mitch:

Yes. Essentially we’re, we’re resolving that tension between being effective, not having time to invest into being more effective, and needing to really treat with care the work that’s being done and the people affected.

Mohamed:

Yeah. Gordon here is asking, how about translation of document documents and summarizing? Any experience?

Mitch:

I have some.

Mohamed:

Yeah.

Mitch:

Yeah, I actually work with a client who’s doing gender neutral translations, of English content into French. And I mean we were able to really quickly whip up, I actually have a case study about this on my, on my website, but we were able to really quickly whip up something that took away 90% of the work and then, you know, absolutely requires a human pass at the end but, but eliminated almost all the typing and deriving from zero.

Mitch:

It was a, huge success.

Mohamed:

Yeah. On my side, you know, there’s translation, and then there’s localization. So a lot of the time we translate, but we always say translate to Quebec French for instance, or French from France, Spanish as well. You can have, you know, Latin American Spanish versus Spanish Spanish and even down to, you know, Mexican Spanish versus Colombian Spanish.

Mohamed:

It’s, it makes a big difference when you want to localize and be closer to the community that you work with. So being able to preface that summarization is also a great thing. You know, the strongest suit for AI is to be able to take long documents and summarizing them and transforming them into different mediums.

Mohamed:

Being able to take a document and say, turn this into an infographic, turn this into a short summary for SEO, for instance, or a summary for a webpage where you don’t want the whole thing, but you just want to express it. It, or turning it into a blog post, for instance.

Mohamed:

Those types of transformations are great.

Mitch:

That’s where sometimes having multiple passes and setting up Basic systems can, be helpful when, let’s say the first result is good but not great. And that can be done in iterations. On the topic of localization, I was struggling to get something in a tone that I wanted a while ago.

Mitch:

Now, I remember going more formal. No, that’s too formal. Less formal. No, that’s not it. More formal. And it wasn’t getting there. Eventually I said write it the way someone from Montreal would say it. And it nailed the tone so perfectly, except that the subject now had the word A in it, which I thought was funny because ChatGPT was throwing shade.

Mitch:

But, other than that it was amazing. So it’s really great at nailing those nuances for sure.

Mohamed:

Julia is asking, what kind of new AI tools do you see, do you foresee for the future? This one’s a difficult one. I think, you know, we’re seeing these things evolve at such a fast clip that it’s always hard to know where things are going.

Mohamed:

But multimodal, which means, being able to do things in audio, video, text, images, all of that is going to be more and more important in the space. And you’re already seeing this with tools like Nano, Banana from Google, you have Sora and Veo from, OpenAI and Google, as well.

Mohamed:

So you’re going to see a lot more different mediums being explored, beyond just text.

Mitch:

Yeah. And then I mean I see AI becoming a really, you know, generalized interface, meaning a lot of, you know, you have a CRM probably and maybe like most people, you don’t really like being in it. So having the AI be the way that you access that is an obvious way forward.

Mitch:

So there’s that and, and then there’s just, you know, what happens when you have all these AIs that your work is dealing with AIs. Well, you’ll have another to interact with all those. So I see like layers of orchestration and, and I think on the flip side, you know, where’s the limit is just that it’s, it’s trust.

Mitch:

Right. Like at a certain point you want to read the page because you need to see it. So I think AI can go as far as like our trust in it can go. And at the end of the day, AI is not going to certify this bridge won’t fall down. So I, I think it’s fun to discuss both sides of that.

Mohamed:

Yeah. We are past the half hour, but before we wrap up, I just Wanted to say that there’s a lot of tools out there. There’s a lot of training programs out there for organizations that really want to take advantage, of a lot of this learning.

Mohamed:

And any information, well, any courses out there. There are, There is implementation support, there is, financial support from the government. There are training programs, digital transformation programs, there’s digital, literacy programs that will offer up to 100% of the upskilling and training, in most cases, 50 to 75% of, the costs.

Mohamed:

But if you really need support there, reach out to us. We can give you some, tidbits on where to go. There’s programs in Quebec and Ontario and then there’s federal ones. So, support is available and it’s easy to get to, because this is a transformational, technology and you know, everybody needs to get on the program with this kind of thing because, you know, this is it.

Mitch:

Get on the bus, man.

Mohamed:

Yeah.

Mohamed:

In saying that, it was a pleasure having everybody on board today. This was a, fantastic session. I hope everybody got, a lot out of it and learned a little bit about how to structure yourself for setting AI up in your organizations and what to think about what to do next.

Mohamed:

If you have any questions for us, we’re always around and we will send out some follow, up emails to get you, get you some resources. But yeah. All right, thank you so much, Mitch. Pleasure as always to hang out and chat.

Mitch:

Likewise. Always a pleasure.

Mohamed:

And everybody hope, hope you have a great rest of your day and we’ll see you in the next one.