Applied AI: Using AI to Go From Insights to Execution

Shh, do you hear that?

Listen closely.

No, seriously, bring your ear really close to your phone or your computer, wherever you’re at.

That found you here.

It’s your data trying to tell you something.

And if you get good enough at listening to it or training AI to listen to it, it will tell you exactly what you should do to take your professional service company to the next level.

The trick?

Figuring out how to go from those actionable insights we talked about a few episodes ago to streamlined execution.

That’s what we’re talking about today.

Hi, I’m Courtney Baker and this is AI Knowhow from Knownwell, helping you reimagine your business in the AI era.

As always, I’m joined by Knownwell CEO, David DeWolf, Chief Product and Technology Officer, Mohan Rao and NordLite CEO, Pete Buer.

We also have a discussion with Daniel Marcous of April about the AI powered app his company has created to make filing your taxes a breeze.

But first, break out those carrots and start tapping one foot because Pete Buer is joining us for a new segment we’re calling, What’s Up, Grok?

Pete, how are you?

I’m good, Courtney, how are you doing?

I’m doing great.

First up, Elon Musk announced this week that a new version of XAI’s Grok 3 is now available to select audiences.

Pete, what do you take away here?

It’s kind of exhausting, isn’t it, all this talk about Elon Musk all the time?

But if we did an analysis of search and media mention terms, Musk is the one word that would appear more frequently than AI.

Wow.

On this Grok 3 and from the announcement, typical hyperbole, as you’d expect in the introduction, an order of magnitude greater capability than Grok 2, which order of magnitude is 10 times better.

I guess we’ll see.

Supposedly outperforms both OpenAI and DeepSeek, of course.

The only real evidence in the article of its advanced capability, though, is reference in demos to a game that Grok 3 created, merging Tetris and Bejeweled together as one thing, to which Musk assigns an assessment of, it’s pretty good.

Worth mentioning and points for transparency, Musk reminds us that the reasoning model is still in beta, so expect some imperfections.

Hopefully, they don’t include bursting into flames on impact.

And ranting aside, what’s the news for business?

So say whatever you will about him.

Musk is a driven innovator and a colorful salesperson, and the storytelling around Grok 3 certainly puts pressure on the competition, OpenAI in particular.

And of course, we love that because inevitably, competition puts greater capability and lower prices in the hands of our business buyers.

So get out there, roll up your sleeves, experiment with all of these applications in the context of your own use cases, and start forming opinions about the one that serves you the best.

And Pete, will you just remind us when it comes to these new models being released?

Because honestly, for the lay person, it feels like every day there’s a new upgraded version.

What should we be looking for in the news that really tells us, oh, this one might be better versus, like you said, a new game development that the…

You know, and I bet you the game development is actually more important than I’m giving credit, you know, that Grok 3 can independently do the work of pulling together two different programs, turning them into one, making it an effective game.

Like, I don’t know how finished it was in the end, but it actually might be a significant accomplishment.

But I’m more interested in specific details than the storytelling around it, you know?

And so what are the new search capabilities?

What are the agentic capabilities, you know, rendered on behalf of a user?

What’s the nature of the data that’s being used, synthetic or real, you know?

Those kinds of hard measures strike me as the ones that matter the most.

Thank you.

And cost, don’t forget cost.

Oh, yes, cost.

There is that.

That is always important as I see my $29.99 coming out of my account each month.

Well, Pete, thank you as always.

Thank you, Courtney.

A few episodes back, I talked with David and Mohan about how to go from raw data to actionable insights.

Today, we’re gonna keep pulling on that thread.

How do you go from actionable insights to what really matters most, execution?

David, Mohan, in our last episode, y’all remember the last episode?

Of course we do, Courtney.

Of course you do.

Okay, so just in case you do need a refresher, we talked about how companies can tap into AI to go from raw data and information to actionable insights.

Today, I want to kind of carry that idea and pull it all the way through by talking about how we can go from actionable insights to strategic execution, like the level up from that.

So to just kind of open this up, what role do you think AI can play in helping leaders fundamentally reimagine how their businesses execute?

Courtney, I was with a customer yesterday, and they were talking about their business and went through how disciplined they were in putting together playbooks for how they engage with their clients and drive execution of the work itself.

But where they were struggling was, how do we make the right decisions when it comes to resource allocation?

How do we know that we’re taking account plans that we put together every quarter and actually drive execution against it so we make proactive strategic progress?

I think what they were alluding to is what you’re talking about here.

It’s one thing to execute really well, to know that we’re going to will it, we’re going to go be responsive, we’re going to be reactive, we’re going to go get the job done.

It’s another to have comprehensive operations which run your business and help you do not the right things, the smart things to advance your business.

And so often we have strategic plans, we have these imperatives, we have account plans, but we don’t find ways to operationalize that.

And that is to orchestrate the execution to make sure that you’re driving crisp execution, not just against whatever is top of mind, but against what needs to be done to execute against the strategy, making those smart decisions, those trade-off decisions that consistently move the ball forward and help the firm to advance.

And I think ultimately that’s the essence of your question of how do we stop just doing a great job getting things done and make sure that we get the right things done that advance the business.

Yeah, I think there are a couple of ways to answer this question, right?

So one is around if you’re getting better insights, how else could you use your time that goes along the way that David was mentioning?

If you can stay ahead of the curve because you got these much better insights using AI, you can potentially plan better and use the time.

We all know execution is the hardest part of any of the business items you can come with great strategy, but it’s all in the execution.

So you have more time for execution.

You can think of it that way.

The other frame I’d like to introduce is using AI for execution itself.

There are two ways to answer your question.

I’ll dwell more on the second and come back to the first as we go along here.

Traditionally, we’ve had automation.

You go from step A to step B to step C.

What we’re seeing today is more of from step A to step B.

In the middle, there’s an LLM or some ML model that is allowing you to do much better predictions that allow you to make step 2 much better.

Your execution is better.

But what is coming with the agentic flow is around asking AI to do broad questions such as, I want to maximize my engagement with this customer.

Here is some history of this customer.

And you go and execute within these parameters.

So you can use the AI for much better execution using these agent workflows as well.

Happy to expand on that.

But Courtney, back to you in terms of which frame was this question asked with?

Yeah.

Oh, which frame?

Do I know the answer to that?

I don’t know.

I thought David would be impressed if I said frame.

You know, Mohan, as you were saying that, obviously execution, it is really hard.

But, and I don’t know if it’s, you know, where the last 10 years of my career have been spent.

But what I find so hard is usually at the individual level, like getting people to execute on what they need to do, barely easy.

The problems for me seem to fall apart when it’s getting all of those people executing, but keeping them aligned and moving towards the same direction.

So many times you find that one person is executing or one department is executing on one thing, pulling this direction, the other team is pulling the other direction, and it is that continual realignment and getting pointed in the right direction that feels like, the thing that makes me want to lose my mind sometimes.

It’s why companies have so many meetings because you just keep having to keep that balance.

That’s why I love the word orchestration.

I think it’s overplayed sometimes.

But really what you’re talking about is how do you take all of the individual workstreams, which in and of themselves, I think you’re right, are probably optimized, right?

If you have a halfway decent manager, they can manage a process, right?

Halfway decent practitioners can manage themselves and get work done and know what they’re doing.

But it is tying it all together, right?

It is getting the trumpet section to come in at the right time compared to the trombone section.

It’s keeping the percussion in line, right?

That image of the orchestra leader, right?

Really queuing in each of the different parts and getting everybody to play together to just make beautiful music, right?

That is really what we’re talking about and how you weave it together.

And I think so often we get lost in the efficiency part.

We forget the effectiveness part, right?

A lot of the reason why orchestration is so important is, to your point, it’s about strategic execution, is about execution towards an objective, and making sure that you’re not only getting things done, but you’re doing the right things to attain the objective.

And along the way, making the trade-off decisions and orchestrating new pieces that need to change and evolve, etc., etc.

And artificial intelligence can help us, because no longer, to Mohan’s point, is it A to B, B to C, it’s not deterministic.

Now we can actually insert an agent to help us make those prudent decisions and manage and orchestrate in real time, where that can be proactive and come back to us and say, hey, alert, proactive, I’m predicting, I’m noticing, I’m seeing X, Y, or Z, you may need to shift execution to hit the objective versus just let it continue the way it is.

I think that’s really interesting.

And Mohan, I would love to get your take on sitting in your seat.

How do you see these kind of tools evolving?

David just gave an example of signaling a flag when maybe one department is way out in front of another.

Is there some other ways that you see AI starting to change what’s possible at this level?

Yeah, I think what’s coming, Courtney, is that you can provide commander’s intent.

We say that in agile methodology a lot.

So essentially, you can tell the AI tools in the coming future to maximize for this customer objective.

Do whatever it is that you need to do.

To maximize for this objective, here is some history and here is just execute on the next best action, right?

So and that can be either autonomous, which could be a little while away, or at least kind of come back to you and say, this is what you need to do next, right?

So as opposed to, if you think about the old mail merge, the automation was about, hey, dear, insert the name, here is the body, insert whatever, sincerely Courtney Baker, right?

So instead of that, here you’re kind of providing sort of more outcomes based objectives to the tools.

And for the tool to figure out the set of things it’s got to do because it’s not a linear process anymore, and come back and say to maximize on this objective, here are the steps, and I’ve thought through this, I’ve paused through this, I’ve sensed it, I know the context, and here are a few options to execute.

Either the execution back can be by the human back or can be autonomous.

That’s where I think things are going.

Yeah, that’s so interesting.

So you see it as, and I really like this, so you see this as, hey, it knows what the company’s objectives are, it knows what’s most important on an individual level in being able to tell you like, hey, here’s what you need to work on today.

I mean, there’s certainly times, I think we’ve all been there when you get, you’re overwhelmed and you have a space, you have an hour between meetings and you just, you kind of have a minute when you’re like, what should I be doing?

There’s so much.

That’s right, that’s right.

You’re providing the outcomes that you wish to see and let the tool figure out the steps.

Well then, that means that’s the end of the never ending to-do list in a way.

You know, it basically is providing, not just defaulting to what’s in your email or what’s in Slack, but really what’s most important to drive the business forward.

Yeah, I mean, I think you’re going to see a total change in enterprise systems to that point, right?

If you think about it, the enterprise systems of today are workflow-based, they’re information-based, it’s data entry, it’s lists, it’s these things.

I think we’re definitely moving more towards a intelligent interface that is telling you in the moment, like prioritization is a word that keeps coming to mind as we talk, right?

The AI can do the prioritization, it can take that objective that Mohan is talking about and saying, oh, right now you need to be focused on this, right?

This is what you should go to, exception handling, noticing things that are going wrong.

Those types of things that aren’t black and white, but take a sense of quote judgment, right?

That the artificial intelligence, in its pattern matching, can take in and apply logic and rationally deduce, here’s what should be important right now.

This is where AI becomes not just the co-pilot that we think of today that’s helping me write an email, like a true leadership management co-pilot that is helping me run the business, right?

I think that’s the power of the future of AI.

Totally.

I mean, the future is in these sort of non-deterministic adaptive flows, right?

So it’s not a sequence of steps, but there could be new variables introduced, new scenarios introduced, and essentially hit the outcome that you’ve set for the tool.

I think it’s so exciting to think of this.

I totally agree.

I am curious for both of you.

I think it’s such an interesting thought process of where we are most likely headed.

What do you think for leaders, executives today, are there things that they should be doing today to prepare for where we’re going?

Hope and pray?

Invest?

So gets there faster?

I mean, that’s always a good idea.

I think leaders that are already think in terms of outcomes they want and not in terms of procedural command and control are already at an advantage here.

Because that’s essentially how they decompose the problem.

So they will be at an advantage because the execution steps are left to somebody who can do it even better than you can do yourself.

In this case, an AI tool.

What happens today is if your command and control, if you’re prescribing how everything has to work, it is not a good recipe to be ready for the new world.

So I would say much more of having an outcomes-based roadmap, understanding your strategic contribution that you’re going to make in terms of how you provide value to your clients, having that kind of vision, seeing how it can be improved with AI or good starting points.

I think another thing that I see is there is a struggle to trust these agents.

There is a struggle to have confidence and to use them for some of the bigger, more important things and I would challenge leaders to begin to experiment and to use different tools to begin to increase your reliance on them and your confidence in them, to be able to get yourself to a point that as these more advanced tools come on board, as they start to operate businesses, not just execute work, that you have the confidence to be able to get on that bandwagon.

Because I think the people that get left behind are the ones that are not going to embrace it at that inflection point when it takes off and you’ve got to be ready.

A big part of that is just the psychological, am I ready to delegate and empower to agentic AI?

It’s a great point.

It’s a really good point and one that I feel like is perfect to leave with everybody to think about today.

David, Mohan, thank you as always.

Super fun.

Thanks, Court.

Awesome.

Who among us wouldn’t like a little AI assist orchestrating and executing on our work?

We all want the time we spend working to be in pursuit of something valuable and important.

That’s one of the benefits of using a platform like Knownwell.

Whether you’re a CEO, a VP of client success, or an account manager, you can be assured that whenever you log into Knownwell, you’ll find the highest priority items that you and your team can be working on.

Go to knownwell.com to find out more, and let us know if you’d like to see your data in the platform.

Daniel Marcous is the co-founder and CTO of April, a product that uses AI to help people and companies file their taxes with ease.

He sat down with Pete Buer recently to talk about how to build trust in AI products and more.

Daniel, so great to have you on the show.

Welcome.

Thank you, Pete, nice to meet you.

As you know, AI Knowhow is about business leaders making sense of AI in their worlds.

Can you kind of enter and help us understand in the way of context how you spend your time and where AI fits in?

Yeah, I’d love to.

So I’m Daniel and I’m a co-founder and CTO in a company called April.

What we do in April is that we’ve started the first AI-powered tax solution to help American citizens file their taxes, plan their taxes, and optimize their taxes in a year-round manner.

And we’re doing this in this kind of way that’s called embedded fintech.

So we embed those tax experiences in your bank, your payroll company or your financial platform of choice.

So you can take care of your taxes in a single place, have no surprises of what’s coming for you next year.

Be confident that you’re doing the right thing for your finances.

And this is highly powered by AI behind the scenes to an extreme level of personalization that I’m sure we’ll talk about.

Looking forward to that.

And the coin just dropped for me on the name, April, when you said US taxes.

Maybe April and sometimes October.

Yeah, that’s a future subsidiary.

So taxes are something in life that it’s not a good idea to get wrong.

And you wouldn’t have reason to know this, but several episodes back, we covered a couple new applications, chatbots from H&R Block and Turbo Tax that were given out dicey advice.

And of all the places you could give bad advice, you know, tax is sort of top three in terms of ending a person in jail.

Can you tell us a little bit about how April works to avoid the kinds of issues that we saw with those guys?

Sure, sure.

So indeed, intimately familiar was the topic and this piece that you mentioned that covered those chatbots.

I’d say first and foremost, it’s really important to understand that AI in fintech is sort of a next level of risk for AI and people are visibly afraid of using AI in fintech and companies are also afraid of putting AI in their user facing product.

So when you talk about AI in fintech, you can see two main use cases.

You can see AI in the back office, which we’re heavily using in April to come up with the calculations, the personalizations or so on, but it’s not directly communicating with the user.

And there’s the second part of the user facing AI, which we often see in chatbots of, hey, the April chatbot, here’s my test question, please help me get a sense of what’s going on in here, where those things sometimes happen to hallucinate and just basically make up answers with confidence, which is even more dangerous than doing things, I’m not sure, but actually making up something with confidence.

And that’s where you need to be extremely careful.

So there were some bad reviews for players in the industry trying to do that in a way that’s user-facing and in a way that’s sort of like a one-size-fits-all chatbot.

Let me answer all of your tax questions using the knowledge over the wide web.

And unfortunately, the knowledge in taxes over the wide web is far from accurate and helpful.

So we were actually taking a dramatically different approach, and the key to the April chatbot is actually rooted in the data.

And not the specific model, which is more of a commodity these days, with all of the foundational models out there.

But really investing in curated articles to say only the right things that are written by the main experts, exactly what we want, exactly how to do those things in the April software.

And having our chatbots limited only to the context of these documents and the official IRS documents, so they are far more accurate and less likely to hallucinate.

And we actually got some decent coverage of this chatbot in Time magazine some time after that, comparing ours to those that you just mentioned.

So it’s been a nice ride.

Yeah, congrats.

That’s awesome.

As you have expressed and as I feel, there is an inherent anxiety in when you’re dealing with your life savings and the tax man to rely on AI in your business.

The explanation that you just gave was confidence giving to me, but I’m sure you must have to deal with getting customers over a hurdle to be willing to try.

How do you do that?

Sure, sure.

There’s the answer of me, Daniel, as a person and what I think.

And I’m personally the tech-optimistic kind of person.

And that the value unlock that I see in employing AI, even when there’s some risk is far greater of what I can achieve with it than my fear of it.

And then there’s the April Way, which takes some of Daniel’s perspectives, but also acknowledges the fear in the community and uses two key pillars to try and solve for this stress and uncertainty.

The first one is the familiarity of common brands and common practices that you use in your day-to-day life, in your finances.

So when April is embedded in another financial software, financial solution, like a bank or payroll that you use in your day-to-day, and we also share the data from that other financial software, which is 100% accurate, then we’re far more likely to make you confident that the data we’re using is accurate and we’re just making something up.

And it’s already a brand that you trust, you know how to interact with, and that really removes a lot of the stress.

And then the second factor, which I think is really important, AI that actually makes financial decisions for people or financial optimizations, proactive advice, so you can choose, needs human oversight.

And a lot of the things we do is not using AI as a stand-alone system, but using it in a manner that’s often referred to in literature as collaborative AI.

It’s a way to marry the advantages of the AI’s creativity and vast knowledge sources with the skills of specific domain experts.

And we have a meticulous process with domain experts from TACS continuously testing, refining, and reviewing what our AI is outputting and the way it evolves, so we can direct it in the right place where we’re extremely confident that I would use it for my personal life, and I would put it out there for other people to benefit as well.

So brands you know and experts you can trust.

I think that’s good advice for anyone bringing something AI powered to market in nonsensitive areas as well, like why not follow the formula?

Yeah, yeah, definitely.

So we waved earlier at the notion of extreme personalization.

Can you tell us a little bit more about how that works with April?

Sure, sure.

I know it’s somewhat of a buzzword these days, but I do strongly believe it.

Basically because of, if you think about taxes, taxes are the entire financial picture of someone’s life.

You get all of their income and all of their financial holdings from wherever they manage their finances, and you get a really strong, solid, accurate picture of someone.

That’s extremely personal on its own.

And then actually using that as an input to AI and machine learning models that can analyze this data properly, and more than that, can use it to come up with helping you making or adjusting financial decisions is extremely powerful.

And in order to do that, the right way, which will actually benefit Pete, and not just give you something like a dashboard of your net worth, but something that is actually impactful for you to optimize on, needs to be extremely personal to your financial situation, which encompasses your entire financial picture that’s included in tax.

You need to really look at a lot of different data points.

So what we do in April is that we try to personalize every step of the way, and doing that in real time.

So trying to unpack this vague sentence that I just said here, we’re like in our software, we’re trying to get to know Pete.

We’re starting to collect tax data that we need for filing purposes on Pete’s financials.

And as we do that in real time, every single second that you answer a question that we ask, that you connect one of your financial institutions that we pull data from, that you upload a tax document that we OCR and service into our system, we update this, personalize, something that you can look at as a decision tree to say, okay, this is what we understand about Pete exactly right now after this new data point, and this is what we need to understand next.

And we’re cutting away all of these different branches from the decision tree that are irrelevant for Pete, which is basically how we’re able to tailor something specifically for you that takes minutes to file your taxes with instead of the nine hours quoted as average by the IRS.

Because we’re not doing the traditional one size fits all.

Here’s everything someone needs to know about a human, but every human is different, and we don’t actually need all of this data.

So this personalization also helps, like it more than helps you cutting friction, it also is more secure and private, because we’re just not asking what we don’t need to know.

So fascinating, my wheels are turning, and I find myself like wandering off the back end in aggregated form or cut into segments.

Like, do you see an opportunity for thought leadership?

You know, the 10 things that guys like Pete need to think about when it comes to filing their taxes, or, you know, I can see so many ways that could be useful to individuals, to banks, to whatever, accountants.

Definitely.

There’s so much potential on how we can help Pete using this tax data that we’re collecting in a very accurate manner.

And this is basically what we’re doing.

So we’re like this full cycle of tax services, which basically renders us sort of this virtual CPA or virtual financial advisor that wealthier people do tend to have on their side and are able to afford, but we’re democratizing it and opening up these capabilities in a virtual manner to the average Joe like you and I.

Maybe you are not the average Joe, but I am.

I’m very average.

The question dawns on me.

The fear around AI is that it takes jobs away.

Do we need fewer tax accountants if we have April?

Well, the interesting thing is that we have fewer tax accountants by nature.

It happens to be if you look at the statistics of how many people go to accounting school within time, then you see that the segment is dramatically shrinking.

We’re actually using AI to fill this gap that’s formed.

I won’t say that we won’t need accountants.

I’d say that unfortunately, there are less accountants.

I’d also say that in the AI-powered world, then the role of an accountant will change.

Also, there’s a lot of interesting opportunities for accountants to work in companies like April developing the next AI-powered accounting software.

Because as we said earlier, domain expertise is extremely important here.

Thank you so much for sharing.

It’s been a pleasure to have the conversation and to meet you.

Same here.

Thank you for having me.

Thanks as always for listening and watching.

Don’t forget to give us a review on your podcast player of choice.

And we’d really appreciate it if you would leave a review.

It really helps other people find this show.

At the end of every episode, we’d like to ask one of our AI friends to weigh in on the topic at hand.

So, hey, Gemini, what’s happening?

By the way, you might be our favorite because you’re what we built our Knownwell platform on.

In this episode, we’re talking about how to go from actionable insights to strategic execution.

What do you recommend?

To go from insights to action, start by using AI to help you prioritize the most promising opportunities and break them down into smaller manageable steps.

Then, test and iterate based on what you learn, letting the data guide your AI-powered strategic execution.

Now, you’re in the know.

Thanks as always for listening.

We’ll see you next week with more AI applications, discussions and experts.