We at
@FoundationCap
believe there is $4.6T of work to be automated. AI companies are leading a transition from Software-as-a-Service to Service-as-Software, turning the table on the very essence of SaaS.
We look at the areas to be automated in two buckets:
1.) Salaries of
nothing is better than the 3 Indian uncles making their rounds at the Databricks conference booths and saying “bakwaas” in Hindi to each other after the booth person finishes pitching them
Prediction:
30% of OpenAI talent will remain at OpenAI
25% of OpenAI talent will go to Google, Anthropic, Cohere
40% of OpenAI talent will go to Sam Altman NewCo
5% of OpenAI talent will leave and start a new company
AI-companies in 2023:
1.) Make a demo of a small feature / widget
2.) Tweet demo
3.) Pray it goes viral
4.) Get VC inbound
5.) Convince them its defensible from what OpenAI, plug ins, and incumbents
6.) Raise
7.) Get disrupted by incumbent
8.) Pivot
Amazing panel at
@OpenAI
old office with
@sama
Best quote: “very few to no one is Silicon Valley has a moat - not even Facebook” when asked what Jaspers moat is
Foundation models, like GPT-4, will prompt a reimagination of nearly every category of software application, creating follow-on opportunities for infrastructure-minded entrepreneurs. Our market map inventories the rapidly evolving space of middleware, or Foundation Model Ops.
Goodbye AIOps: welcome AgentSREs—the next $100B opportunity
"AIOps" is just another buzzword that fails to capture the true potential of AI in IT ops and observability.
We're on the cusp of a fundamental shift in how organizations monitor, debug, and optimize complex software
A Bay-Area born and based founder recently shared with me a struggle to sell an internal knowledge retrieval Slack bot in the Bay Area. It's a common problem many AI-focused startups face: targeting the right customer base. Who needs these tools?
the fact that u just wake up in India and they come offer u choices like: dosa, poha, or aloo/mooli/mehti paratha is truly spectacular
and it’s not some microwave BS but the real thing
the food here is 100000x US
Gen Z friends: “Why is your location on find my friends in your office in Palo Alto”
Me: I’m working
Them: “Omg r u ok? “You need to set boundaries” “I would never allow that” “my manager wouldn’t dare ask me” “on a long weekend, isn’t that illegal”
Me: bro it’s a recession
Been on a journey interviewing 20+ industry leaders across customer success, product, and sales on embedding foundation models and generative AI in their workflows...... here's a short thread about some high level learnings
Historically, RPA's goal was to automate routine tasks but it frequently encountered limitations... Studies have cited failure rates of 30-50% for RPA projects
At
@FoundationCap
, we believe that LLMs are a game-changer for RPA
From volume perspective it’s probably:
1.) “Glean” / or QandA for XYZ role (sales, CSM, PM, marketing) or XYZ industry (legal)
2.) BI / data analytics
3.) horizontal productivity agent
4.) LLM evals / observability
5.) Devins for QA
6.) foundation models for xyz industry /
I have been pitched 500+ AI-related ideas in the past 18 months.
The most common idea, by far, is some variant of "interfaces to simpllify data analytics and dashboarding".
Crazy how much talent and brainpower is going into reinventing BI.
Exciting time to be building in LLM middleware along the following use cases:
- Data / embeddings management
- Finetuning
- Prompt engineering / versioning
- Deployment / optimization
- Observability and monitoring
- LLM analytics
- App development
Just got off the phone with a ML infra CEO who said:
"If a company is questioning whether finetuning is the answer for their LLM use case, they should not be using finetuning."
This isn’t true. Last week, a seed stage company had 10+ lead term sheets. This week, talking to companies with 5+ lead term sheets each.
For the right deal, they will Zoom in from Italy and have the people on the ground chase you.
Noticing a lot more OOO from the VC industry this summer. Highly advise founders to hold off initiating their fundraise processes until after Burning Man (early Sep), if possible.
Sadly, I’m totally serious.
We are at a unique time in history where every layer in the AI stack is improving exponentially, with no signs of a slowdown in sight.
Everyday, there are new models, new frameworks, new architectures, new infrastructure tooling, inference providers, etc. that pop up.
Excited about GPT4 and the Google launches? Well
@FoundationCap
is hosting an exclusive hackathon for 30 builders in B2B Enterprise AI next Wednesday in SF.
Tired of living in SF? Want to join all your friends in NYC but continue at a b2b SaaS AI startup founded by Stanford alums?
Ideally: 2-3+ YOE
Pls DM me!
Do u ever just get home and open ur Instagram app to watch Reels before bed and u see this message in ur DMs from ur 25 year old ex coworker you’ve talked to 4 times in ur life
Like bruh
The most fascinating part the
@FoundationCap
Unconference was the amount of people that chose the LLMops session. 50% of the attendees (~50 people).
As we changed from a smaller room to a bigger room, I remember someone said “I’ve been doing infrastructure for 20 years - since
In May, we
@FoundationCap
hosted an unconference with 80+ builders in the
#GenAI
space. We've summarized our learnings in a quick blog post. Key takeaways 👇
The evolution of observability is a case study in solving problems while inadvertently creating new ones. Early pioneers like Splunk and AppDynamics made opaque systems transparent. Their success was clear: Splunk's 2012 IPO valued it north of 1B+
Remember when ur parents took u to the temple when u were like 7 and gave u a 20 dollar bill to give to one of the gods
U walked and had to choose between:
Should I give to Hanuman for strength
Saraswati for the brain cells
Lakshmi for more money
Executives:
2020: The world is ending. Fire everyone
2021: Wait hire them all back. Pay them 3x and give housing in the Ritz w/ monthly bonus. Hire everyone including the dead
2022: Hm money got expensive. Fire them ALL
2023: ChatGPT has been promoted to analyst, PM, SWE 1.
Accuracy matters. Demos on Twitter show cherry-picked use cases, but in the enterprise world, accuracy is king. Implementation is no walk in the park—data integrations, engineering complexity, and security challenges abound
Ever since ChatGPT has constantly been at capacity this week, I’ve had to actually craft my own emails again
What a literal waste of time and productivity slow down
Friendly reminder that you have 13 days left to apply to
@FoundationCap
generative AI challenge. The best eligible startup(s) will receive a $250K SAFE (uncapped, no discount) investment.
Breaking: LEGO's CEO in DC, insisting all block makers pass an architect's exam, because who else knows blocks better?
Kindle's boss also on the scene, proposing only literature PhDs can distribute e-books.
And Uber's head - demanding all ride-share apps need taxi medallions.
We spent the last 12 weeks building a market map on the space, which is quickly emerging (as shown by all the companies below)
Special thanks to
@jerryjliu0
@aparnadhinak
@shyamalanadkat
and others for reviewing!
Blast hosting 15+ founders for a dinner in NYC. Chatted about evals, demos to production and the excitement around the promise of tools like
@joannezchen
@FoundationCap
This great chart from BCV shows the limited number of companies that can construct their own LLM.
What’s interesting is the number of cos configuring LLMs for use cases specific to financial services, like risk modeling, fraud detection, policy underwriting or purchase order
For the non AI-companies that want to train their own LLM similar to Bloomberg GPT:
the infrastructure and tooling is less of a constraint - it’s really only an option if you can hire sophisticated AI talent which is a subset of some of the roles on this slide
these are also
It's a $100B+ opportunity to create agentSREs. Success requires expertise in ML, LLMs, and distributed systems. The field is wide open for multiple decacorns. Building in this space? Email me at jgupta
@foundationcap
.com
@FoundationCap
@ashugarg
At a birthday dinner in SF with a few VC and consultant friends and everyone is debating if their bosses would kill someone for 1B dollars.
What’s fascinating is how sure everyone is on their answer….
@salesforce
one of the most successful software companies of all time, generates only $35 billion in revenue annually, and will grow incrementally. Global companies spend $1.1 trillion on sales and marketing salaries each year.
The size of the opportunity for AI disruption is
Recently had the chance to engage with several Chief Data Officers of old legacy, incumbent F500 cos. A common thread?
They all navigated a career landscape where technology and business existed in silos, rarely intersecting.
Aspiring to be the next Langchain?
In the era of LLMops tooling, we observe a pivotal shift towards open-source components, wherein communities form the bedrock of innovation.
Here's some early thoughts on some of the best strategies I've seen working so far:
Shark Tank but for developer tooling companies:
+A $250K investment from Foundation Capital
+ A POC from Turing, which has a network of 3 million developers worldwide
+3 finalists will get to pitch Quora CEO
@adamdangelo
, Turing CEO
@jonsidd
, and Foundation Capital
Excited to be partnering with
@rarjunpillai
and
@atmb4u
on DocketAI and huge congrats to them on their 15M series A.
I had long admired Arjun’s work, and when I reached out to him about attending a Foundation Capital event last July, I learned he left ZoomInfo two days prior
Ran a fun analysis to see what the top 20 B2B Software VC's were investing in from mid September to last week. There were a total of 71 investments.... Let's dive into the insights! ⬇️
MLOps: Demand for solutions < Supply for solutions
FMOps: Supply of solutions < Demand for solutions
In the past, MLOps companies had to convince leaders with extensive handholding to "show" the potential of ML applications. Now, ChatGPT has every board room interested.
As someone who has spent 8+ hours staring at VC data this week, it’s quite astonishing how VC may be one of the very few careers where failure is the overbearing norm
I did some quick analysis today on YC's new founder directory. The conclusion was that YC has a type called ex-Stanford and ex-Google.
For this analysis, I only looked at the industries that were tied closely to B2B software.
Heard from a exec that instead of thinking in where to apply generative AI across different functions in companies (sales, marketing, customer support, etc.), they are doing an inventory of what data / knowledge is queried upon / or asked questions about to prioritize early uses
Tried to make an intro today for a company to a friend and received this back. The bar is very high for a new SaaS application, especially where there are lower budgets
India is quite literally predicted to be the only country in 2030 to not suffer from a labor shortage and will have a surplus of 1.3M technical workers vs other nations.
Excited about the next generation of AI applications, where the most value will accrue! Picks and shovels are still exciting but challenging as the space is quite crowded and as builders collectively learn about the best supporting tools (is it vector databases / Postgres/
Beyond wanting to integrate an LLM into all enterprise operations and being super excited about the idea of doing so", it seems like enterprises are unclear on the "for what". Starting w/ the why first is key at its 1.) reducing friction and 2.) accelerating outcomes
South Asian holidays are either a massive family reunion, that round trip to Delhi or sitting peacefully on your laptop for 4 weeks in your childhood bedroom
I was standing outside a restaurant in Palo Alto on my phone and the 3 owners were all staring at me and made a 100 dollar bet w/ each other on what ethnicity I was
Their guesses were Turkish, Persian and Lebanese
I said no and asked for 100 bucks if they’re wrong again
@RyanFloyd
Miami : crypto bros during a ZIRF + COVID phenomenon
NYC: Bay Area born Berkeley EECS majors that are unsatisfied with an aspect(s) of their life and think NYC will resolve the problem
Super exciting time to be building in AI in SF. 60+ submissions and 200+ people for an “emergency SF” hackathon.
Shoutout to
@nonmayorpete
@rachel_l_woods
for organizing and letting me test my prelim judging skills 😅
1/9 🧵 why are proprietary API-based models outpacing open source adoption in the tech world? I think it could be less about the tech and more about the persona behind the professional. Let's explore….
Last: I’m excited to announce an open call for the “Foundation Capital Generative AI 25”: our ranking of the 25 most influential, VC-backed companies that are defining the generative AI field. Think you should be on the list? Apply:
Does everyone work with a Glean or Vectera or do you work with a bunch of different mini app vendors & libraries to build your search applications? Will enterprises reach for developer tooling or do founders need to offer the end user experience to get a foot in the door?