Three ways to build an AI app layer company

Rebuild and rethink the core system of record

It used to be hard to go after the core system because there was so much engineering effort needed to build all the modules
As you can see in this example here, this is the core system in the home services space
Home services is anything to do with your home - like HVAC or cleaning
This is the back office system that people who do those jobs use
and there are 24 modules you would need to build, which would take years of effort
But now, using AI you can build much faster
This is a screenshot from our Fund 3 portco Solea, who are also playing in this space. They've been able to build the core modules in just 6 months
Beyond that, they've built a bunch of net new modules like AI customer service, AI sales coaching and AI route optimization, which add tons of value for the users
Most importantly, their software allows home services businesses to slash headcount because they've automated away a bunch of roles

Build an invisible layer on top of the system of record

it doesn't always make sense to go after the core system, sometimes it makes sense to build a new layer next to the core system
Typically this makes the most sense when the core system is deeply entrenched
In that case, you should partner with them and build the layer on top that intermediates the end user
So we can see this here with our porto Alex, who are an AI interviewer. Instead of rebuilding the underlying system, they've chosen to deeply integrate with them
So when a candidates applies to a job, Alex gets pinged and can screen their resume, interview them and score them. They can also look inside the core system to see if there are any existing employees who are worth interviewing, and then interview them too
In this way, Alex ends up housing the most valuable and up-to-date candidate data, and gets to co-sell with the underlying system which turbocharges their growth

AI-native services

1) The end product is an outcome
2) Underpinning that outcome are a bunch of manual steps, many of which can be automated by AI

So, many examples come to mind here like legal work, audit, staffing, market research etc
So you can take this approach when you believe that you can automate a large swath of internal process and ultimately generate 5-10x more revenue / employee than a human centric firm
That's the North Star
And if you don't have it, then you have fake PMF, because it means that in order to grow, you need to scale headcount linearly
So a really interesting example is that of building an AI-staffing agency, which is shown here through a company we recently backed called Clera
They use AI to reach out to 10s of thousands of candidates simultaneously, have conversations with them, screen them and then match them to businesses -- without any human in the loop
This business is six months old and last month placed 6 candidates, with a team of just 5. They generated $20K fees on each placement.
Obviously if this works, this is very disruptive to human-first staffing agencies because your cost structure doesn't have any labor anymore, which also means you can charge a much lower placement fee, which can be a very unique value prop in the market

Rebuild and rethink the core system of record

It used to be hard to go after the core system because there was so much engineering effort needed to build all the modules
As you can see in this example here, this is the core system in the home services space
Home services is anything to do with your home - like HVAC or cleaning
This is the back office system that people who do those jobs use
and there are 24 modules you would need to build, which would take years of effort
But now, using AI you can build much faster
This is a screenshot from our Fund 3 portco Solea, who are also playing in this space. They've been able to build the core modules in just 6 months
Beyond that, they've built a bunch of net new modules like AI customer service, AI sales coaching and AI route optimization, which add tons of value for the users
Most importantly, their software allows home services businesses to slash headcount because they've automated away a bunch of roles

Build an invisible layer on top of the system of record

it doesn't always make sense to go after the core system, sometimes it makes sense to build a new layer next to the core system
Typically this makes the most sense when the core system is deeply entrenched
In that case, you should partner with them and build the layer on top that intermediates the end user
So we can see this here with our porto Alex, who are an AI interviewer. Instead of rebuilding the underlying system, they've chosen to deeply integrate with them
So when a candidates applies to a job, Alex gets pinged and can screen their resume, interview them and score them. They can also look inside the core system to see if there are any existing employees who are worth interviewing, and then interview them too
In this way, Alex ends up housing the most valuable and up-to-date candidate data, and gets to co-sell with the underlying system which turbocharges their growth

AI-native services

1) The end product is an outcome
2) Underpinning that outcome are a bunch of manual steps, many of which can be automated by AI

So, many examples come to mind here like legal work, audit, staffing, market research etc
So you can take this approach when you believe that you can automate a large swath of internal process and ultimately generate 5-10x more revenue / employee than a human centric firm
That's the North Star
And if you don't have it, then you have fake PMF, because it means that in order to grow, you need to scale headcount linearly
So a really interesting example is that of building an AI-staffing agency, which is shown here through a company we recently backed called Clera
They use AI to reach out to 10s of thousands of candidates simultaneously, have conversations with them, screen them and then match them to businesses -- without any human in the loop
This business is six months old and last month placed 6 candidates, with a team of just 5. They generated $20K fees on each placement.
Obviously if this works, this is very disruptive to human-first staffing agencies because your cost structure doesn't have any labor anymore, which also means you can charge a much lower placement fee, which can be a very unique value prop in the market