July 23, 2023 • 18 mins
This is an essay readout using an AI voice clone of NFX general partner, James Currier. Generative Tech is the next step in software. It’s a new level of human-machine partnership. It turns deep learning engines into collaborators to generate new content and ideas nearly like a human would. This area is so new, and moving so fast, you can have an advantage in your chosen area in a few months… for now.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:04):
This is James Currier, General Partner at NFAXas investors and builders were optimistic about
the future of AI.
One experiment among many we've done is we'vetrained AI avatars of our own voices to read
out the essays we write.
So this is my real voice you're hearing, butThis is now my AI voice.
I've been created by NFX to provide essayreadouts moving forward as you'll hear.
(00:27):
I am still a pale imitation of the real JamesCurrier, but I'll get better over time.
And if you have suggestions for me, please letus know.
This week's article is called generative techbegins and was first released in October of
2023 as our first essay focused on the subjectof generative AI.
Let's get started.
Generative tech is the next step in software.
It's a new level of human machine partnershipPete turns deep learning engines into
(00:51):
collaborators to generate new content and ideasnearly like a human would.
We now have high quality, cheap, fast AI modelsfor generating text images, videos, software
code, music, voice, 3 d models, and more, noneof which is copyrighted and is not plagiarized.
Some have called it generative AI.
But AI is only half of the equation.
AI models for the enabling base layers of thestack.
(01:11):
The top layers will be thousands ofapplications generative tech is about what will
actually touch us.
What you can do with AI as a partner, NFXstarted investing in the space 2 years ago and
has invested in 18 generative tech companies sofar with more to come founders today.
We are encouraging you to create companies inthis area now to catch the best part of the
tech adoption cycle.
This essay is for you.
(01:32):
If you are an investor or an angel who isfocused on this new area, had to signal and
change your profile so you show up on the listof generative tech investors.
So what's new about generative tech?
Embry Wants, a new internet topology untiltoday.
The internet has been characterized by makingdatabase queries to get stored piece of old
content from the center out to you on the edgeof the network.
Generative tech changes the topology of theinternet because now unique pieces of content
(01:55):
are generated edge of the network in real timeby your action.
That's a major shift, which typically opens upfounder opportunities if web 1 was read only
and web 2 is read right.
And then generative tech is read write generatethen that makes web 3 read write generate on.
Generative tech is now happening in parallel toweb 3 and moving faster.
As crypto hadn't happened, we'd probably becalling this web 3, but we do have crypto.
(02:17):
So we call this generative tech, but othernames could be web 3a generative Beller even
generative internet.
Number 2, human activities will now changequickly.
1 to 2,000,000,000 knowledge workers willbecome faster and better at their jobs.
Some will be able to do jobs they couldn't dobefore.
New types of jobs will be created, and whilesome jobs will be downgraded, threatened,
eliminated.
And that will cause fear and self doubt in tensof millions of workers in the next 36 months,
(02:40):
the expansion of people's abilities,productivity, and efficiency will vastly
outstrip the losses overall generatingtrillions of dollars of value for knowledge
workers and creatives.
Going from 0 to 1 in their minds will never bethe same.
For instance, writers, students, marketers,coders, architects, graphic designers,
musicians, videographers, sales developmentreps, customer service reps, and screenplay
(03:02):
writers, who are paid go from 0 to somethinguseful will now be using these tools to
generate their first ideas until now.
Software has been used to refine our initialideas into something useful.
It was responsible for the second half of theprocess, if you will, of going from 0 to
something useful, but these new generativetools help you with the first half of the
process taking you from nearly 0 to a lot ofinitial ideas, and then the old software tools
(03:24):
pick up from there and take you the rest of theway until now.
Software couldn't solve the 0 to one problembecause it worked for us.
Generative tech will work with us from thebeginning of any projects today.
And for the next few years, this will feelsurprising and in many ways, scary because
those creative moments where you go from 0 toinitial ideas has always felt so uniquely human
because it has been so mysterious.
The idea is once thought to come unpredictablyonly through people's minds and souls emerging
(03:48):
from talent or training Morgan associated withspecial people will now be generated by
something which is not a human not a coworkeror collaborator dot dot dot and something that
is not you.
This will be disturbing the many people.
However, as with most new human machineinterfaces, We'll get through the discomfort
and get used to it.
In the next 10 years, we will expect softwareto collaborate with us.
It will be the new normal Steve Jobs said in1980 that the Apple personal computer was a
(04:12):
bicycle for the human mind.
You might say that generative tech is a rocketship for the human mind.
The makers of these AI models might say theyare actual minds.
No doubt they will get this.
We've been talking about the inevitability ofsoftware based mind since the 50 seconds.
A first example of the dawn of this era we feltas a culture was in 1997 when IBM's deep blue
beat Casperavances.
(04:32):
The neck big leap was when alphago definitivelybeat Lee settle on the game of Go and Treda 16,
starting in 2022.
Generative tech is going to have an impact onbillions of workers where they live.
This is a very different level.
This is the skillful creation of new things.
So what changed so that generative tech ishappening now?
The recent availability of open sourcealternatives to proprietary generative AI
(04:53):
models from OpenAI is what caused it to tipwide open in the last six months.
In short, Aluther dot i's Pete Neo X 20B.
Launched in February 2022 is the open sourcealternative to open i's GPT 3 for text
generation.
Stability eye stable diffusion.
Launched August 2022 is the open sourcealternative to open eyes Dolly 2 for images and
videos.
Both have been game changers on price, quality,and ease of access.
(05:16):
The cost to generate images has dropped 100 xin the last two months.
The friction to generate output from thesemodels through web and mobile has become about
10 x easier in the last 6 months.
Quality generated text.
Images, code, speech, etcetera is rapidlyreaching human quality.
Many feel we're passing the Turing test inseveral of these content categories already.
It's hard to measure, but you know the qualitywhen you see it as the 2021 Stanford University
(05:40):
AI index noted.
AI, for some constrained application, has movedto a sufficiently high standard that humans
have a hard time telling the difference betweensynthetic and non synthetic pewdas.
We are headed toward generative everything,because of all these changes.
The amount of experimentation has grown 20times larger in the last 2 months.
This accelerates applications providing valueand introduces even more people to the
(06:01):
community.
It's an old story in technology.
The barriers go down and boom.
Morgan explosion.
We are in the early innings, but generativetech is a thing now.
Here we go, and we believe it's going to happenfaster than most people think, unlike self
driving cars.
Generative tech doesn't face regulation anddoesn't need to be perfect to avoid killing
Pete.
Unlike VR, it's already useful.
Needs no new hardware and is getting betterrapidly as a founder.
(06:24):
You can trust that the cost and quality ofnearly every type of content is good enough
today to get your company going, text, images,code, speech, 3 d video, certainly.
By the time you have your team together andseed money raised, it will be there.
Don't overthink it.
Where we are today is just an on ramp.
It's now possible that most of our software andhuman computer interfaces will be significantly
augmented in the next 5 years after 14 years ofnear stasis.
(06:48):
Opens up seams of opportunity for founders,advice to founders right now.
To catch this wave as a founder, you need tomove this week, this month, not in the next 6
months or the next 3 years.
Unless you're on a rocket ship already.
In the fast moving water, I would pause whatyou're doing and consider focusing on this.
We've already invested in 18 generative AIcompanies over the last 2 years.
And we are aiming to make more investments inthe next 12 months when a new sector opens up
(07:11):
like this.
Founders can typically find low hanging fruitmore easily than in areas that are better under
in this more competitive.
So get in there.
Here's what we suggest to founders in order tostart a new company to do new things and create
new markets.
Start new companies that redo old businesseswith this technology at the core, not just as a
feature, add generative tech features to yourexisting product to differentiate it.
(07:32):
Characteristics of generative tech products.
Generative tech products have 2 layers.
The bottom layer is an AI model that is capablegenerating novel output based on inputs that
are unique to the user, like Openized DALL E orGPT 3, to make generalized versions of these
can take 100 of millions.
To make more narrow versions can be less than10,000,000 and the price is dropping very fast.
(07:52):
Open source versions are already starting to beviable.
The top layer is an application.
This is where you can build network effects andembedding effects to produce durable
businesses.
This stack will lower the technical barriersassociated with certain fields.
You don't have to be an architect to generatedrawings of a house remodel.
You don't have to be an illustrator to tellDolly what to draw.
This is what allows generative tech to unlocknew companies and projects.
(08:15):
These emergent companies have certain corecharacteristics that help place them on the
generative tech continuum.
Tierra 3, memory lines 0 to 1 and 0 to 10.
Generative tech begins with solving a 0 to 1problem.
The most successful companies will eventuallyprovide 0 to 10 solutions or put differently
products that serve the complete needs of theuser.
Are uniquely animated by AI models.
(08:36):
You can imagine a version of Latitude's AIdungeon that combines the images and videos to
match the text for complete gameplayexperiences.
Or allows you to create a persistent onlinepersona using images or sounds generated with
AI.
We already see examples of generative AIprojects that provide near finished products.
Projects like Sol, a choose your own adventure,70 seconds inspired sci fi film, uses a
(08:56):
combination of generative tech tools to rapidlygenerate videos.
Write scripts, and generate character voices.
Number 2, replace curation with creation.
Generative tech is personalization in a way wehave never experienced it before.
For 20 years.
We have been chasing personalization throughcuration, ecommerce providers.
Netflix and Spotify, I'll wanna serve youcurated products you're most likely to like
(09:17):
from their central databases.
Facebook, TikTok, and the New York James haveexperimented with curating your experience of
their content.
This is a very limited approach topersonalization because it is based on calling
existing data.
We have been trying to retrofit people'spreferences into our existing offerings rather
than generating new things that are best suitedto them.
Generative tech replaces curation withcreation.
(09:38):
Generative tech is not a more sophisticateddatabase call.
It can be trained by that database but its corefunction is to generate something new on the
edge of the network.
Generative tech skillfully creates noveloutput, the content, images, or experiences
served to you will not have existed until youask for it or triggered it through some other
action or simply by your presence.
This is happening in the music space boomy,Amper, sound draw, and others are a company
(10:00):
using AI to generate full length, originalsongs, and seconds.
Boomi also gives creators the tools to shareand monetize those creations.
An example of a generative tech layered withSaaS tools.
Critically, boonies AI generates instant musicthat fits anything from a mood to a genre.
That music has never been heard before youdecided to create it in a pre generative tech
world.
You might select a playlist for a road tripcurated by someone else.
(10:22):
In a post generative tech world.
You will generate entirely new songs that fityour occasion, mood, blood pressure, heart
rate, location, and who you're with, etcetera.
Number 3, low friction interfaces.
Perhaps the biggest breakthrough right now ishow easy the generative tech tools are to use.
So much heavy lifting is done by the AI models.
Friction is removed from the creative process.
(10:42):
DALL E and stable diffusion require only simpletext prompts to generate stunning artwork in 30
seconds.
Often, the generation will be automatic.
It will happen just by you showing up.
You could imagine a version of second lifewhere 2 characters enter a house together.
The house could generate something entirely newthat both users' personalities like art
objects, experiences, music, and characters, orif you sell something, it automatically
(11:04):
generates an NFT, or you could imagine displaysof your life that pull from your photos,
videos, texts, and music.
And they'll be good.
These will be the next level of human machinecollaboration.
It's a partnership where you get to besurprised and inspired.
What will generative tech companies do?
Let's break down where the next greatgenerative tech idea will come from and why it
can come from you.
For your brainstorming.
(11:25):
Think, what if I marry 1 or more base layer AImodels with AI model businesses?
If you are founders who build AI and ML models,Go through the same list about and see if you
can create the AI model for that specific areaif you have access to unique data.
If you're the first to see an application area,You can get a leg up by having the best model
for a time in a specific area.
That advantage may not last and more generalmodels might Pete into your data advantage.
(11:48):
Or competitors selling an inferior, but stilluseful model more cheaply as has happened to
the general models already in the last 2 years.
Remember, data network effects are typicallyasymptotic.
If she can make them real time or hyper local,they are more durable.
Metro Netals, another way to land on a greatgenerative tech IDs to think about how that
business might work operationally.
How can it have network effects where every newuser adds value to every other user?
(12:11):
How can it embed itself in a business orsomeone's life?
So they don't wanna stop using it in the longrun.
This is Jasper's job in 2023 to figure outwhere there are hyper local data sets for your
AI model that you can own and maintain yourdata network effects despite competition coming
in later.
Where can you plug into existing workflows or abrowser or an app?
What function does the application serve 3quick functions that are clearly working today?
(12:33):
There will be more over time.
Number 1's initial ideas collaborator is Bellersolve the 0 to 1 problem.
These companies generate rough drafts orcompleted projects.
And incorporate traditional SaaS tools to helpperfect those drafts over time.
We expect these companies to move towardcreating finished products but moving from 0 to
1 is the first big step.
Beller bird's 4 plan generation engine is agood example.
(12:56):
It creates the first draft of a remodel.
So are companies that generate a copy or firstdrafts of code based on plain language Pete.
Companies like Copy.ai or Copismith.
A company like Jasper dot ai shows us how thispath eventually leads to 0 to 10 solutions
Jasper dot ai provides specialized writing andimage capabilities across disciplines.
It's a one stop shop for your firm's writingneeds in all 4 minutes.
(13:18):
They are now in the process of trying to embedthat in a company so they can't take it out
like an enterprise SaaS company does.
These products get the ball rolling on complextasks and let humans take it from there.
Number 2, coaching and tailored feedback, welearn through a process of trial and error.
But we learn faster with a coach.
We expect generative tech to analyze ourperformance, generate advice, or incorporate
(13:39):
tools that allow us to hone our craft.
Many of these applications might make somepeople feel uncomfortable at first, but they
will challenge us to grow with the help of anAI collaborator.
This is going to be the new normal.
Number 3, uniqueness at scale, uniqueness andscalability have historically been incompatible
concepts.
Truly unique things can exist on mass withoutlosing their bespoke qualities.
(14:00):
Generative tech changes this.
The generative engine is capable of providing anew output for every new user or every problem
at scale.
N fxbackthe.com allows you to generate 100 ofnew websites all within one spreadsheet.
These are cookie cutter copies.
They're beautifully designed and unique to theneeds of each user.
An example from the biology world is insilicone medicine.
In silicone medicine employs 3 ai poweredproducts that work together, one identifies new
(14:24):
targets for drugs, while another generates newcandidate molecules from scratch.
Finally, the last engine predicts the outcomeof clinical trials based on previous work.
This is an elegant example of an analytical AIapproach combined with generative tech it's
similar to how snowflakes are generated innature.
Millions fall during every storm, each totallyunlike the one before it.
But imagine that each one of those snowflakescould generate revenue for a business, cure a
(14:46):
disease, or spark delay.
Founder advice, generative tech will haveunusual market dynamics because it's already
consensus.
Typically, major tech shifts roll out slowly.
Many people were still skeptical of theinternet until 2003.
So those of us who believed in lesscompetition.
SaaS was gaining consensus from 1997 to 2005.
Apple didn't open their iOS platform to outsidedevelopers for 18 months after launch.
(15:10):
Webb 3 has been rolling out for 10 years, buteveryone is on board with generative tech.
The VCs get it.
The founders get it.
The incumbents get it, and it's clear that thegame is now on.
What that means for founders is you have tomove very, very fast.
Believe this we matter.
You have to pick your idea very carefully.
What you decide to build, who your targetcustomer is, and what your distribution
(15:31):
channels will matter a lot.
There are patterns for what ideas work.
You can go after a horizontal or a vertical aparticular data type or geography.
There are many choices in the generative techsector today.
As an example, there were 50 plus socialnetworks with the same five features when
Facebook launched.
Social networking was already consensus, butFacebook started with college students at
(15:51):
Harvard, and that turned out to be the rightplace to focus.
You will have to make a similar focused choicein this consensus market, how to be fast
growing and defensible and generative tech.
If you're building a generative tech businessfor enterprises, to grow fast, be prepared to
be a plug into existing systems.
Don't try to replace workflows or replaceexisting software systems.
To be defensible.
(16:11):
In bedding customers existing workflows andsoftware, you're seeing Jasper announced they
want to be the browser plugin that gives allknowledge workers access to all the underlying
AI models text, images, etcetera, they won'ttake it all, of course, but their plug in
approach is correct on both counts, easy toimplement, and embedding in workflows.
Another good enterprise example is tab 9, whichmimics GitHub Copilot code generator.
(16:32):
It doesn't replace the programmer's codeeditor.
It just sits on top for fast growth.
What Tab 9 did for defensibility is to buildhyper local data network effects for each
company that serves around their exact codebase, which locks those customers into Tab 9.
Combining the embedding of the workflow with aprotectable data moat is a good combination for
durability.
If you're building B2C, it's more open ended.
(16:52):
Consumers love novelty and are willing to adoptnew behaviors faster.
Just make sure to move fast and get a networkeffect if you are building for SMB.
Likely going to be in between something brandnew and something that plugs in, speed bumps,
regenerative tech.
There are still questions of issues ofcopyright and safety, and I know firsthand how
real those Currier.
Founders should be concerned with not makingweapons of mass social destruction Currier
(17:15):
still.
As founders of an important company, you needto be a steward of society and not just your
shareholders.
But great founders step into the risks andsolve the challenges quickly.
Don't let those concerns slow you down, betterto be thoughtful and good and also early on the
field.
To be the one figuring it out, rather than leftbehind, wringing your hands with a furred brow
on the sidelines, founder challenges, thebiggest business risk at this stage of the
(17:37):
market cycle is that founders don't move fastenough into the seams opening in the Morgan,
and these technologies will simply becomefeatures and augmentations of the larger
company's businesses.
If you are Figma or Salesforce, you arescrambling to add such features as a Pete.
If you are Snapchat, James, etcetera as afounder, you need to pick your lane very
carefully.
As we discussed above, you want to find thefast moving water.
(17:58):
What do great generative tech founders looklike?
They look like you.
This area is so new, and moving so fast You canhave an advantage in your chosen area in a few
months.
For now, it's October of 2022.
This window will close in months, not years.
Flint your area, find your scene, eithertechnical, distribution, customer focus,
geography focus, etcetera, and hit it hard,then give NFX a call for your seed capital.
(18:21):
We invest 1 to $4,000,000, and we can help.
Thanks for listening to this week's essayreadout of generative tech begins.
As a reminder, this is still an AI imitation ofJames Currier speaking.
We are having fun experimenting with new toolsbut would love to hear your feedback.
Email us at qed@nfx.com.
(18:42):
Subscribe to the NFX podcast on your favoritelistening platform.
Share with you network and founder friends, andstay tuned for weekly essays and much more.
This is James Currier, General Partner at NFAXas investors and builders were optimistic about
the future of AI.
One experiment among many we've done is we'vetrained AI avatars of our own voices to read
out the essays we write.
So this is my real voice you're hearing, butThis is now my AI voice.
I've been created by NFX to provide essayreadouts moving forward as you'll hear.
(00:27):
I am still a pale imitation of the real JamesCurrier, but I'll get better over time.
And if you have suggestions for me, please letus know.
This week's article is called generative techbegins and was first released in October of
2023 as our first essay focused on the subjectof generative AI.
Let's get started.
Generative tech is the next step in software.
It's a new level of human machine partnershipPete turns deep learning engines into
(00:51):
collaborators to generate new content and ideasnearly like a human would.
We now have high quality, cheap, fast AI modelsfor generating text images, videos, software
code, music, voice, 3 d models, and more, noneof which is copyrighted and is not plagiarized.
Some have called it generative AI.
But AI is only half of the equation.
AI models for the enabling base layers of thestack.
(01:11):
The top layers will be thousands ofapplications generative tech is about what will
actually touch us.
What you can do with AI as a partner, NFXstarted investing in the space 2 years ago and
has invested in 18 generative tech companies sofar with more to come founders today.
We are encouraging you to create companies inthis area now to catch the best part of the
tech adoption cycle.
This essay is for you.
(01:32):
If you are an investor or an angel who isfocused on this new area, had to signal and
change your profile so you show up on the listof generative tech investors.
So what's new about generative tech?
Embry Wants, a new internet topology untiltoday.
The internet has been characterized by makingdatabase queries to get stored piece of old
content from the center out to you on the edgeof the network.
Generative tech changes the topology of theinternet because now unique pieces of content
(01:55):
are generated edge of the network in real timeby your action.
That's a major shift, which typically opens upfounder opportunities if web 1 was read only
and web 2 is read right.
And then generative tech is read write generatethen that makes web 3 read write generate on.
Generative tech is now happening in parallel toweb 3 and moving faster.
As crypto hadn't happened, we'd probably becalling this web 3, but we do have crypto.
(02:17):
So we call this generative tech, but othernames could be web 3a generative Beller even
generative internet.
Number 2, human activities will now changequickly.
1 to 2,000,000,000 knowledge workers willbecome faster and better at their jobs.
Some will be able to do jobs they couldn't dobefore.
New types of jobs will be created, and whilesome jobs will be downgraded, threatened,
eliminated.
And that will cause fear and self doubt in tensof millions of workers in the next 36 months,
(02:40):
the expansion of people's abilities,productivity, and efficiency will vastly
outstrip the losses overall generatingtrillions of dollars of value for knowledge
workers and creatives.
Going from 0 to 1 in their minds will never bethe same.
For instance, writers, students, marketers,coders, architects, graphic designers,
musicians, videographers, sales developmentreps, customer service reps, and screenplay
(03:02):
writers, who are paid go from 0 to somethinguseful will now be using these tools to
generate their first ideas until now.
Software has been used to refine our initialideas into something useful.
It was responsible for the second half of theprocess, if you will, of going from 0 to
something useful, but these new generativetools help you with the first half of the
process taking you from nearly 0 to a lot ofinitial ideas, and then the old software tools
(03:24):
pick up from there and take you the rest of theway until now.
Software couldn't solve the 0 to one problembecause it worked for us.
Generative tech will work with us from thebeginning of any projects today.
And for the next few years, this will feelsurprising and in many ways, scary because
those creative moments where you go from 0 toinitial ideas has always felt so uniquely human
because it has been so mysterious.
The idea is once thought to come unpredictablyonly through people's minds and souls emerging
(03:48):
from talent or training Morgan associated withspecial people will now be generated by
something which is not a human not a coworkeror collaborator dot dot dot and something that
is not you.
This will be disturbing the many people.
However, as with most new human machineinterfaces, We'll get through the discomfort
and get used to it.
In the next 10 years, we will expect softwareto collaborate with us.
It will be the new normal Steve Jobs said in1980 that the Apple personal computer was a
(04:12):
bicycle for the human mind.
You might say that generative tech is a rocketship for the human mind.
The makers of these AI models might say theyare actual minds.
No doubt they will get this.
We've been talking about the inevitability ofsoftware based mind since the 50 seconds.
A first example of the dawn of this era we feltas a culture was in 1997 when IBM's deep blue
beat Casperavances.
(04:32):
The neck big leap was when alphago definitivelybeat Lee settle on the game of Go and Treda 16,
starting in 2022.
Generative tech is going to have an impact onbillions of workers where they live.
This is a very different level.
This is the skillful creation of new things.
So what changed so that generative tech ishappening now?
The recent availability of open sourcealternatives to proprietary generative AI
(04:53):
models from OpenAI is what caused it to tipwide open in the last six months.
In short, Aluther dot i's Pete Neo X 20B.
Launched in February 2022 is the open sourcealternative to open i's GPT 3 for text
generation.
Stability eye stable diffusion.
Launched August 2022 is the open sourcealternative to open eyes Dolly 2 for images and
videos.
Both have been game changers on price, quality,and ease of access.
(05:16):
The cost to generate images has dropped 100 xin the last two months.
The friction to generate output from thesemodels through web and mobile has become about
10 x easier in the last 6 months.
Quality generated text.
Images, code, speech, etcetera is rapidlyreaching human quality.
Many feel we're passing the Turing test inseveral of these content categories already.
It's hard to measure, but you know the qualitywhen you see it as the 2021 Stanford University
(05:40):
AI index noted.
AI, for some constrained application, has movedto a sufficiently high standard that humans
have a hard time telling the difference betweensynthetic and non synthetic pewdas.
We are headed toward generative everything,because of all these changes.
The amount of experimentation has grown 20times larger in the last 2 months.
This accelerates applications providing valueand introduces even more people to the
(06:01):
community.
It's an old story in technology.
The barriers go down and boom.
Morgan explosion.
We are in the early innings, but generativetech is a thing now.
Here we go, and we believe it's going to happenfaster than most people think, unlike self
driving cars.
Generative tech doesn't face regulation anddoesn't need to be perfect to avoid killing
Pete.
Unlike VR, it's already useful.
Needs no new hardware and is getting betterrapidly as a founder.
(06:24):
You can trust that the cost and quality ofnearly every type of content is good enough
today to get your company going, text, images,code, speech, 3 d video, certainly.
By the time you have your team together andseed money raised, it will be there.
Don't overthink it.
Where we are today is just an on ramp.
It's now possible that most of our software andhuman computer interfaces will be significantly
augmented in the next 5 years after 14 years ofnear stasis.
(06:48):
Opens up seams of opportunity for founders,advice to founders right now.
To catch this wave as a founder, you need tomove this week, this month, not in the next 6
months or the next 3 years.
Unless you're on a rocket ship already.
In the fast moving water, I would pause whatyou're doing and consider focusing on this.
We've already invested in 18 generative AIcompanies over the last 2 years.
And we are aiming to make more investments inthe next 12 months when a new sector opens up
(07:11):
like this.
Founders can typically find low hanging fruitmore easily than in areas that are better under
in this more competitive.
So get in there.
Here's what we suggest to founders in order tostart a new company to do new things and create
new markets.
Start new companies that redo old businesseswith this technology at the core, not just as a
feature, add generative tech features to yourexisting product to differentiate it.
(07:32):
Characteristics of generative tech products.
Generative tech products have 2 layers.
The bottom layer is an AI model that is capablegenerating novel output based on inputs that
are unique to the user, like Openized DALL E orGPT 3, to make generalized versions of these
can take 100 of millions.
To make more narrow versions can be less than10,000,000 and the price is dropping very fast.
(07:52):
Open source versions are already starting to beviable.
The top layer is an application.
This is where you can build network effects andembedding effects to produce durable
businesses.
This stack will lower the technical barriersassociated with certain fields.
You don't have to be an architect to generatedrawings of a house remodel.
You don't have to be an illustrator to tellDolly what to draw.
This is what allows generative tech to unlocknew companies and projects.
(08:15):
These emergent companies have certain corecharacteristics that help place them on the
generative tech continuum.
Tierra 3, memory lines 0 to 1 and 0 to 10.
Generative tech begins with solving a 0 to 1problem.
The most successful companies will eventuallyprovide 0 to 10 solutions or put differently
products that serve the complete needs of theuser.
Are uniquely animated by AI models.
(08:36):
You can imagine a version of Latitude's AIdungeon that combines the images and videos to
match the text for complete gameplayexperiences.
Or allows you to create a persistent onlinepersona using images or sounds generated with
AI.
We already see examples of generative AIprojects that provide near finished products.
Projects like Sol, a choose your own adventure,70 seconds inspired sci fi film, uses a
(08:56):
combination of generative tech tools to rapidlygenerate videos.
Write scripts, and generate character voices.
Number 2, replace curation with creation.
Generative tech is personalization in a way wehave never experienced it before.
For 20 years.
We have been chasing personalization throughcuration, ecommerce providers.
Netflix and Spotify, I'll wanna serve youcurated products you're most likely to like
(09:17):
from their central databases.
Facebook, TikTok, and the New York James haveexperimented with curating your experience of
their content.
This is a very limited approach topersonalization because it is based on calling
existing data.
We have been trying to retrofit people'spreferences into our existing offerings rather
than generating new things that are best suitedto them.
Generative tech replaces curation withcreation.
(09:38):
Generative tech is not a more sophisticateddatabase call.
It can be trained by that database but its corefunction is to generate something new on the
edge of the network.
Generative tech skillfully creates noveloutput, the content, images, or experiences
served to you will not have existed until youask for it or triggered it through some other
action or simply by your presence.
This is happening in the music space boomy,Amper, sound draw, and others are a company
(10:00):
using AI to generate full length, originalsongs, and seconds.
Boomi also gives creators the tools to shareand monetize those creations.
An example of a generative tech layered withSaaS tools.
Critically, boonies AI generates instant musicthat fits anything from a mood to a genre.
That music has never been heard before youdecided to create it in a pre generative tech
world.
You might select a playlist for a road tripcurated by someone else.
(10:22):
In a post generative tech world.
You will generate entirely new songs that fityour occasion, mood, blood pressure, heart
rate, location, and who you're with, etcetera.
Number 3, low friction interfaces.
Perhaps the biggest breakthrough right now ishow easy the generative tech tools are to use.
So much heavy lifting is done by the AI models.
Friction is removed from the creative process.
(10:42):
DALL E and stable diffusion require only simpletext prompts to generate stunning artwork in 30
seconds.
Often, the generation will be automatic.
It will happen just by you showing up.
You could imagine a version of second lifewhere 2 characters enter a house together.
The house could generate something entirely newthat both users' personalities like art
objects, experiences, music, and characters, orif you sell something, it automatically
(11:04):
generates an NFT, or you could imagine displaysof your life that pull from your photos,
videos, texts, and music.
And they'll be good.
These will be the next level of human machinecollaboration.
It's a partnership where you get to besurprised and inspired.
What will generative tech companies do?
Let's break down where the next greatgenerative tech idea will come from and why it
can come from you.
For your brainstorming.
(11:25):
Think, what if I marry 1 or more base layer AImodels with AI model businesses?
If you are founders who build AI and ML models,Go through the same list about and see if you
can create the AI model for that specific areaif you have access to unique data.
If you're the first to see an application area,You can get a leg up by having the best model
for a time in a specific area.
That advantage may not last and more generalmodels might Pete into your data advantage.
(11:48):
Or competitors selling an inferior, but stilluseful model more cheaply as has happened to
the general models already in the last 2 years.
Remember, data network effects are typicallyasymptotic.
If she can make them real time or hyper local,they are more durable.
Metro Netals, another way to land on a greatgenerative tech IDs to think about how that
business might work operationally.
How can it have network effects where every newuser adds value to every other user?
(12:11):
How can it embed itself in a business orsomeone's life?
So they don't wanna stop using it in the longrun.
This is Jasper's job in 2023 to figure outwhere there are hyper local data sets for your
AI model that you can own and maintain yourdata network effects despite competition coming
in later.
Where can you plug into existing workflows or abrowser or an app?
What function does the application serve 3quick functions that are clearly working today?
(12:33):
There will be more over time.
Number 1's initial ideas collaborator is Bellersolve the 0 to 1 problem.
These companies generate rough drafts orcompleted projects.
And incorporate traditional SaaS tools to helpperfect those drafts over time.
We expect these companies to move towardcreating finished products but moving from 0 to
1 is the first big step.
Beller bird's 4 plan generation engine is agood example.
(12:56):
It creates the first draft of a remodel.
So are companies that generate a copy or firstdrafts of code based on plain language Pete.
Companies like Copy.ai or Copismith.
A company like Jasper dot ai shows us how thispath eventually leads to 0 to 10 solutions
Jasper dot ai provides specialized writing andimage capabilities across disciplines.
It's a one stop shop for your firm's writingneeds in all 4 minutes.
(13:18):
They are now in the process of trying to embedthat in a company so they can't take it out
like an enterprise SaaS company does.
These products get the ball rolling on complextasks and let humans take it from there.
Number 2, coaching and tailored feedback, welearn through a process of trial and error.
But we learn faster with a coach.
We expect generative tech to analyze ourperformance, generate advice, or incorporate
(13:39):
tools that allow us to hone our craft.
Many of these applications might make somepeople feel uncomfortable at first, but they
will challenge us to grow with the help of anAI collaborator.
This is going to be the new normal.
Number 3, uniqueness at scale, uniqueness andscalability have historically been incompatible
concepts.
Truly unique things can exist on mass withoutlosing their bespoke qualities.
(14:00):
Generative tech changes this.
The generative engine is capable of providing anew output for every new user or every problem
at scale.
N fxbackthe.com allows you to generate 100 ofnew websites all within one spreadsheet.
These are cookie cutter copies.
They're beautifully designed and unique to theneeds of each user.
An example from the biology world is insilicone medicine.
In silicone medicine employs 3 ai poweredproducts that work together, one identifies new
(14:24):
targets for drugs, while another generates newcandidate molecules from scratch.
Finally, the last engine predicts the outcomeof clinical trials based on previous work.
This is an elegant example of an analytical AIapproach combined with generative tech it's
similar to how snowflakes are generated innature.
Millions fall during every storm, each totallyunlike the one before it.
But imagine that each one of those snowflakescould generate revenue for a business, cure a
(14:46):
disease, or spark delay.
Founder advice, generative tech will haveunusual market dynamics because it's already
consensus.
Typically, major tech shifts roll out slowly.
Many people were still skeptical of theinternet until 2003.
So those of us who believed in lesscompetition.
SaaS was gaining consensus from 1997 to 2005.
Apple didn't open their iOS platform to outsidedevelopers for 18 months after launch.
(15:10):
Webb 3 has been rolling out for 10 years, buteveryone is on board with generative tech.
The VCs get it.
The founders get it.
The incumbents get it, and it's clear that thegame is now on.
What that means for founders is you have tomove very, very fast.
Believe this we matter.
You have to pick your idea very carefully.
What you decide to build, who your targetcustomer is, and what your distribution
(15:31):
channels will matter a lot.
There are patterns for what ideas work.
You can go after a horizontal or a vertical aparticular data type or geography.
There are many choices in the generative techsector today.
As an example, there were 50 plus socialnetworks with the same five features when
Facebook launched.
Social networking was already consensus, butFacebook started with college students at
(15:51):
Harvard, and that turned out to be the rightplace to focus.
You will have to make a similar focused choicein this consensus market, how to be fast
growing and defensible and generative tech.
If you're building a generative tech businessfor enterprises, to grow fast, be prepared to
be a plug into existing systems.
Don't try to replace workflows or replaceexisting software systems.
To be defensible.
(16:11):
In bedding customers existing workflows andsoftware, you're seeing Jasper announced they
want to be the browser plugin that gives allknowledge workers access to all the underlying
AI models text, images, etcetera, they won'ttake it all, of course, but their plug in
approach is correct on both counts, easy toimplement, and embedding in workflows.
Another good enterprise example is tab 9, whichmimics GitHub Copilot code generator.
(16:32):
It doesn't replace the programmer's codeeditor.
It just sits on top for fast growth.
What Tab 9 did for defensibility is to buildhyper local data network effects for each
company that serves around their exact codebase, which locks those customers into Tab 9.
Combining the embedding of the workflow with aprotectable data moat is a good combination for
durability.
If you're building B2C, it's more open ended.
(16:52):
Consumers love novelty and are willing to adoptnew behaviors faster.
Just make sure to move fast and get a networkeffect if you are building for SMB.
Likely going to be in between something brandnew and something that plugs in, speed bumps,
regenerative tech.
There are still questions of issues ofcopyright and safety, and I know firsthand how
real those Currier.
Founders should be concerned with not makingweapons of mass social destruction Currier
(17:15):
still.
As founders of an important company, you needto be a steward of society and not just your
shareholders.
But great founders step into the risks andsolve the challenges quickly.
Don't let those concerns slow you down, betterto be thoughtful and good and also early on the
field.
To be the one figuring it out, rather than leftbehind, wringing your hands with a furred brow
on the sidelines, founder challenges, thebiggest business risk at this stage of the
(17:37):
market cycle is that founders don't move fastenough into the seams opening in the Morgan,
and these technologies will simply becomefeatures and augmentations of the larger
company's businesses.
If you are Figma or Salesforce, you arescrambling to add such features as a Pete.
If you are Snapchat, James, etcetera as afounder, you need to pick your lane very
carefully.
As we discussed above, you want to find thefast moving water.
(17:58):
What do great generative tech founders looklike?
They look like you.
This area is so new, and moving so fast You canhave an advantage in your chosen area in a few
months.
For now, it's October of 2022.
This window will close in months, not years.
Flint your area, find your scene, eithertechnical, distribution, customer focus,
geography focus, etcetera, and hit it hard,then give NFX a call for your seed capital.
(18:21):
We invest 1 to $4,000,000, and we can help.
Thanks for listening to this week's essayreadout of generative tech begins.
As a reminder, this is still an AI imitation ofJames Currier speaking.
We are having fun experimenting with new toolsbut would love to hear your feedback.
Email us at qed@nfx.com.
(18:42):
Subscribe to the NFX podcast on your favoritelistening platform.
Share with you network and founder friends, andstay tuned for weekly essays and much more.
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
The Breakfast Club
The World's Most Dangerous Morning Show, The Breakfast Club, With DJ Envy, Jess Hilarious, And Charlamagne Tha God!
Crime Junkie
Does hearing about a true crime case always leave you scouring the internet for the truth behind the story? Dive into your next mystery with Crime Junkie. Every Monday, join your host Ashley Flowers as she unravels all the details of infamous and underreported true crime cases with her best friend Brit Prawat. From cold cases to missing persons and heroes in our community who seek justice, Crime Junkie is your destination for theories and stories you won’t hear anywhere else. Whether you're a seasoned true crime enthusiast or new to the genre, you'll find yourself on the edge of your seat awaiting a new episode every Monday. If you can never get enough true crime... Congratulations, you’ve found your people. Follow to join a community of Crime Junkies! Crime Junkie is presented by audiochuck Media Company.