Introduction

So we're shifting.

I'm going to be talking to founders here who are building tech companies, but also it will apply to you if you're working in a job.

Background: Software’s growing dominance

But one of the main shifts that we're going to see in the software ecosystem, again, I'll lay out a bit of background here.

So Mark Andreessen of Andreessen Horowitz said, you know, software is eating the world.

And in the past 10, 20 years, we've seen that the majority of the value in the market has moved towards technology, towards software companies.

companies.

So all of your work has become much more technical in nature.

That is a trend

we're going to see continue, but there's going to be a shift in the way it's framed.

And

so it's going to move from software as a service, which is the tools that all of you have been

using, which maybe I'll mention some examples, to service as software.

Okay?

So a little

About the speaker

bit about me.

I'm self -titled the AI guy.

I run two companies, AI Forward and something

something new that we're launching called Unlimited Lead Engine, and I was previously

at Deloitte in AI there, I was the national lead for natural language processing.

So yeah, I'm really in helping companies move in certain ways, and I've architected the

kind of infrastructure for a Forbes 30 under 30 startup that's at over 10 million ARR.

as I've done a few things.

I've also presented my AI innovations

to Google's director of research, Peter Norvig.

Building products in the age of AI

The shift: Software as a Service → Service as Software

Okay, so building products in the age of AI.

The shift from software as a service to service as software.

1So the era of selling tools to humans is ending.

The tools are now becoming intelligent.

They can actually do things, so we don't just sell tools.

So now we want to partner with humans to deliver outcomes,

which we're now beginning to see.

It's not about replacing people.

It's about elevating them.

This is not to say that there will not be disruption in the job market.

There will.

But the majority of the disruption so far will be in how you work and not whether you work.

So that's the good news.

So we're going from using tools to managing outcomes.

From seat-based tools to outcome-based delivery

So in the old model, you would use a SaaS tool.

There are so many examples.

You have a CRM system, you have an ERP system, you have an HR system, you have Word, PowerPoint, Excel.

All of these are seat -based software where they sell you licenses per seat.

A human does the work, the software kind of assists it, it stores it in memory.

And the bottleneck was always human bandwidth for road tasks.

And then programmers would automate some subset of ROTAS, but not all of them, which is why we still have lots of very annoying manual work.

1The new world that we're going towards is we're going to be selling outcomes.

So you might have a software that does something for you, which sells at like 500 per outcome, for an audit, for example.

So AI does the groundwork, and then human provides the judgment.

So we go from being employees and people who are doing admin work to people who are really managers of these AI platforms, these AI agents, right?

And this eliminates also, of course, burnout from low -value drudgery.

How technology climbs the value chain

So the evolution of value, I want to talk about it.

It's historic.

So what technology does is it doesn't just automate, it climbs the chain, right?

From calculators and PCs to the cloud

So we started, I mean, way back we had calculators, and then that became the personal computer where you had spreadsheets that could take care of pen and paper arithmetic.

And so just to take a financial example, the human became then a financial strategist rather than a calculator.

There was a time we were going to the moon.

We had people, we had women specifically, who were doing calculations.

They were called calculators.

Now, computers do that job.

out.

The next shift from the personal computer was the cloud.

So this is, instead of these

servers, you had an abstraction of infrastructure, everything moved towards AWS, Google Cloud,

and the human became the cloud architect rather than the person who set up the server.

Now,

for those of you who are not technical, this may not make sense, but I will try to bring

it back to that it does.

What AI changes: drafting, analysis, and orchestration

Now with AI, AI can actually do, like it can draft code, it can

generate PowerPoint reports, do research documents, put together proposals, do financial analysis,

and it kind of can also do some creative tasks, like design related.

So AI now really does,

you could say 80%, 90 % of the work, but the human has to do a lot of the tweaking, the judgment,

and the synthesis of what the AI is doing, right?

So it's about orchestrating and coordinating the AI.

So we're not removing the humans.

We're kind of really elevating the human to manage the robot, okay?

The AI “bubble” and why wrappers fail

So now when we look at a lot of people consider what's happening right now as the AI bubble.

Now, if you're invested in the stock market, it's something to consider

whether your portfolio is too weighed on AI stocks.

But the reason, there are two components to the bubble.

One is that six or seven companies in the US market

have like 34 % of market value.

It's very high by historical standards.

So that's like whether there's too much concentration.

That's one angle.

But the other part of the AI bubble

is there's been a lot of attack.

So the thing that differentiates AI

from the internet revolution and the dot -com bubble is that in the dot -com

bubble a lot of companies could build on the internet as infrastructure right you

lots of websites we can all build websites today that's infrastructure but

AI is not like that so when you try to build on top of AI if your product works

really well the AI company just builds their product themselves and serves it

directly to consumers and cuts you out of the market so most startups that have

have been attempting to take advantage of the AI revolution

have failed, because any sort of wrapper

that they build, if it's successful enough,

the AI company, like OpenAI, like I joke about it,

every six months, Sam Altman kills a few billion dollar

startups, right?

So that's really how it goes.

So that model has to fundamentally shift.

shift.

In fact, we had, I think, and recent Horowitz, Ben Horowitz, he said that one of

the things that's happening with startups today is that you had a six, seven person

startup and you wrote like, so it's like, you know, a woman, like it takes nine months

to give birth, whether you have one female or nine, right?

So with software, it's similar

if you have five programmers who've spent two years on a software you can't put like 500

programmers and get it done in a month it would still take two years and you know you probably

use five so a lot of startups tech startups could compete with the giants because they could build

products have a couple of years of head start and then the companies would rather buy them than

compete but today with ai you can actually write code a lot faster too so the whole model for

startups is changing.

What stays durable: depth, complexity, and defensibility

Code is cheap, intelligence is becoming commodified, and the durable mode

is like depth, complexity, things that models alone cannot do.

The Centaur model: humans + AI working together

So the new mode, we call

it the Centaur model.

You have a human that can see, that's the one eye of the Centaur,

and then you have kind of the AI that can do a lot.

Moats: workflow integration and data flywheels

So you would have deep workflow

flow integration so you wouldn't just record data data is very important but you'd also do the work

and you'd become the operating system not just a dashboard not just the reporting tool

okay you want to have proprietary data flywheels now a flywheel is basically a virtuous loop okay

so essentially like you have ai that produces drafts human then corrects it and then that

corrected draft is stored as data for the models to reference in the future so over time you get a

compounding effect of improved data for the AI to use as context and it gets

better over time right that's one kind of flywheel another kind of flywheel you

can have is the more integrations you do with they are the more infrastructure

building the more useful the AI becomes the more money you have the more

infrastructure the more depth and and the more scope it has so these flywheels

are kind of what is going to be a competitive advantage for companies that

that want to compete going into the future.

Constraints and opportunities

So of course, one of the challenges we all face

Regulation in high-stakes industries

is regulation.

There's going to be upcoming AI regulation.

There's already some.

But already in law, medicine, and finance,

there is a need for that.

Now, it may also happen that the regulation loosens.

We've seen self -driving cars finally

start to get through the regulatory process

and get onto the road.

But regulation is a big factor.

Physical-world complexity and where AI can’t reach yet

so that's one of the challenges and then the other challenges of course there is

physical complexity so one of the things I want to leave you with is at the

moment the AI is restricted to a computer so in theory a perfect AI could

do anything that a human can do on a computer and more but it cannot do

anything in the physical world yet okay until we have robots and we don't know

what the time frame for robots is but it's certainly going to be a few years

after AI.

So wherever there's physical world complexity, whether it's like a dentist, whether

it's a car wash, whether it's a live in -person experience, it's an events business, those

businesses are going to be okay for the time being, right?

So those industries are something

that I encourage startups and founders to look at.

What this means for your job and knowledge work

But if you're also thinking about your job

and if you're in a knowledge work industry you you have to do one of two

things you're either going to adapt I heard a part and McKinsey said today

that AI is going to do a lot of what McKinsey and other consulting firms did

for their clients so they have to go up in complexity they have to look at

outcomes they have to say we're going to help companies double their market cap

right so if you're doing any sort of knowledge work consider how to elevate

yourself and whether there are things you can add to your value that AI can't

easily do and if you're looking at an industry like customer support that is

going to have a lot of challenges I've identified some of the physical world

complexity businesses industries that will probably be okay in the in the

short to medium term okay so so just so that's come some of the that the

A practical playbook for implementing AI

the overview so this is kind of the playbook of how you can actually

implement AI in your company is whether you're working in a job with you doing

a startup so the first thing you can do is like find a veg like finds a painful

repetitive boring task that AI can do right so that's like step one if you

want to get using AI in whatever you're doing step two is then you can capture

the gold standard okay so you can see what the AI is doing you can compare

had that, the gold standard, measure the distance, see if you can prompt engineer your way or

whatever, but you want to get closer to the gold standard, okay?

Founder strategies: where to build and buy

And then in terms of if you're building a company, one of the things you can do is you

can acquire businesses, because I don't know if you guys know this, but the largest growing

demographic in North America, including Canada, is most people don't know, it's actually old

people.

It's people who retire, okay?

When people are retiring, they leave their businesses.

And a lot of these businesses are businesses that will do perfectly okay in the age of AI.

So that's another place where you can really capture value.

And I'll contextualize that a little bit more.

Another set of industries is like healthcare.

I mentioned that.

Logistics, again, physical components.

legal is arguable but where lower level legal work is automatable still you still need lawyers at the

moment and manufacturing right because again it's it's it's it's a physical business these are some

examples this is not exhaustive there are other good examples okay so essentially I want to kind

Structuring a business: IP/AI “brains” vs. asset-heavy cash flow

of also give you a model of how to set up the business so you can have a kind

of the brains of the business like that'll own the AI the IP and talent you

can have it be venture backed etc right and then you can have something else

that owns like the physical assets and focuses on cash flow okay so you don't

want to dilute your tech equity you don't want to if there's issues and

stuff that's just how to structure it very much for entrepreneurs so the main

Own the mine, not the shovels: investing in scarce assets

thing that I want to leave you with is you don't want to shell you don't want

to mine the gold you don't want to shell sell the shovels you want to own the

mine because the shovels will get better the mining process will get better but

there is only a limited number of mines okay or real estate assets so like you

You want to think of actual assets that do not get depreciated by AI

and try to own those assets in whatever form

because those assets will become more and more valuable

as AI is able to do more with them.

But there is a finite set of those assets.

So I wouldn't build tools that are wrappers.

I wouldn't build black box systems.

I would look at how you can multiply the value of one human.

I would look at interoperability or high amount of observability into the AI systems.

And also partnerships where you would really want to empower the humans.

And that's also going to be a much easier approach to sell and get into the market.

What the next big companies will do

So in my opinion, the next $10 billion company won't replace the expert yet.

I mean, until we get to like real AGI.

But it could make every employee as good as the best expert.

it.

And that's what I think we should be aiming at.

This is not going to be an easy transition.

It's not going to be comfortable.

You have to be prepared for it.

But if you're prepared for it,

you can be one of the people who really benefits a lot from it.

Your work gets easier,

year, your value grows, and so on.

Conclusion: preparing for transition and the upside

I will end on one optimistic note.

If AI is really successful, the projections indicate that

the economy will grow from about 2 % a year to about 10 % a

year, OK?

So if this all works out, we will have an

abundance of riches, OK?

So I wouldn't go into this with a scarcity mindset.

There's going to be a lot of opportunity happening if the

the world of AI is successful.

So, okay.

So thank you very much.

Q&A

I'm here for any questions.