Building Globally Relevant AI & Deeptech Companies

Event Summary

Proposed Parslee Meetings for Follow Up:

Companies that could use Parslee

1. Blinka, a Commercial Real Estate deal and loan automation platform
2. Fernstone, an Insurance policy automation platform for Businesses.
3. Michael Block formerly at Agent Smyth

VCs willing to have an open discussion:

1. Clint Korver from Ulu Ventures
2. Sherman Williams from AIN Ventures
3. Mehreen Malik from MeiLin Capital
4. Ben Parr from TheoryForge
5. Itxaso del Palacio from Notion VC

{
  "name": "tool-showcase",
  "toolPaths": [],
  "nodeChildren": []
}

• Ulu Ventures
• AIN Ventures
• TheoryForge
• Notion VC :::

Success Looks Different

Panel Discussion

The Defensibility will be Understanding the Customer

• Our experience with Collibra suggests that even in a competitive, noisy market, companies can unlock Go-to-Market breakthrough simply by understanding their customers better, with more complexity, and productizing those insights quickly and reliably.

Getting it Right often requires Taking it Slow

• Our experience is that certain companies benefit from spending a longer time than could be reasonable iterating on their business with a small, lean team. For instance, SaaS products typically take longer to incubate.

Be Cautious with Early Revenue

• The art of navigating pilots is worth it to get out of the sandbox and echochamber. All about the knowledge you capture.
• Equity pool for lighthouse customers, co-development narrative.

Don't Underestimate Brand

• Each company will need to control their own narrative. Ex: "Blow through the benchmarks." This has required a lot more attention to Public Relations, Influencer Marketing, and Digital Storytelling. In addition, thinks like Developer Experience and Developer Community. ElevenLabs focused on influencer marketing, and it was simple to sign up.

• We tend to over-estimate the need for differentiation in product and growth engines. But, because markets can feel so crowded, sometimes just getting the brand right can make an exponential difference. Don't need white space, need great brand. For instance, Lovable pursued a brand-first strategy, and arguably Devin had a head start and a better product and even a better technical team. Yet, Lovable clearly nailed it.

Is there enough Growth Capital?

• DeepMind got acquired because they couldn't unlock growth capital. At the time, there wasn't enough. And now we see something similar, not because there aren't more growth capital firms or more growth capital, but because there are so many startups in a mad race to grow as quickly as possible. And they are consuming a ton of cash.

• Google could provide the patience, and the credibility to tackle some of DeepMind's biggest proof-of-concepts, such protein folding with AlphaFold. So, if a company is onto something but is going to need an ambiguously long runway to achieve it, getting acquired in the right circumstances might be the wise move.

The Traps of Distractions, The Power of Focus

• The combination of using AI for productivity and trying to compete with the accelerated speed of the AI market can lead to distractions and rabbit holes.
• So, as we get to know a company, at Notion VC we are trying to evaluate their ability to prioritize over time. We ask for case studies on "What you said no to, what opportunities you walked away from and why." If they don't have good answers, that's a yellow flag.

The Punch Bowl is going to Disappear Soon

1. a lot of down rounds, growth investors will be hammered on average.
2. a lot of acquisition activity, with the dire need for a consolidation in the market. However, traditionally VCs have been shit at funding rollups, both identifying teams that could pull it off as well as managing governance and a team that can bring exponential expertise.

What market segments are promising?

Don't Dismiss Defense

• There are a lot of opportunities in Defense that are unexpected, Defense budgets are eager for all kinds of solutions that are not explicitly defense. This is in addition to Dual-Use Technologies.

Watch the Research

• The strategics are actually paying very close attention to research publications, academic research, corporate research labs, and platforms like Hugging Face.

Incumbents AI-Washing can make Eager Customers

• A lot of large technology companies and services companies are in the process of rapidly AI Washing, and they are hugely insecure in their lack of true AI capabilities. They are willing to urgently spend money.

Ask about the Success Criteria of Early Adopters

• A challenge we're seeing a lot of is there's a race to be an early adopter of some platform. However, the platform doesn't deliver on either the brand promises or the hidden expectations of those customers. Any company can unlock a new market opportunity if they can actually deliver, and it helps if they have specific success criteria with early partners.

Round Table Discussion

The Frontier of "Knowing" the User and Customer

• There's a huge gap for both consumers and SMB, and Product Led Growth. This is the ability for the AI Systems to capture expressed preferences, and to intelligently managing preferences.... For instance, an AI Travel Agents
• The needs for AI to provide and thus have context for Data-Driven Decision Making needs better ways of memory, context management, etc. Giving better and more custom "advice" or providing more context-aware information could unlock a lot of market potential for "Advisory Services."
• Some have come to call this need Situational Awareness. ^[6va2gm]
• There should be people working on established techniques from Decision Science such as Markov Decision Process, or gated decision making like Decision Quality Framework by Ron A. Howard.

Can you be dedicated to "One Platform" or "One Distribution Mechanism"

• Should we be suspicious of AI companies that are completely designed and built around one platform or distribution mechanism? Like VS Code?
• Celonis got very big focused on building around SAP and has achieved a multi billion.
• Klaviyo built a rapid growth rate by focusing on their partnership with Shopify...

To What extent will our B2B AI Native companies be competing with Consulting Firms and System Integrators

• Systems Integrators are now building their own tools.
• This seems to be depending on workflow expertise -- Beau Seil.
• Accenture is the only big consulting firm that seems to be doing well at this, that they would be bullish about.

To What Extent will our Vertical Wrappers be competing with core Model APIs

• barraged with pitches from fledgling AI/fintech founders and very few companies of the hundreds that I see do anything more than serve as a wrapper on existing LLMs.  99% of what they do can be done by Perplexity or Claude.  

Beware of Customer Paralysis

• There seems to be a new paralysis at the customer level. Customers are experimenting and its creating false positives for portfolio companies. AI Pilots are questionable with unclear success criteria and an inability to understand traction levels.

Should we be worried about the Insane Valuations for the Core Model providers?

• Open AI is $10B run rate, Anthropic is at a $5B run rate. How to fulfill the revenue needs to justify the backfill valuation? Won't get backfilled.
• IT Services market globally is $1.5 trillion. In order for there to be any good return on new model generation it needs to be $2 trillion.
• Need to be a B Corp.

What's going on with emerging "Models" and AI Paradigms?

• We shouldn't think that LLMs are the end all be all. There's lots of progress in Multimodal Models, Chain of Thought. Relationship Systems Models like Graphon , State Space Models. Better than LLMs.
• LLMs seems to only work well right now for use cases where errors are easy to see and there are processes to fix them. That's why Code Generators are so popular and work so well.
• AGI seems like a distant fantasy right now, despite the hype.

Is the AI Hype hiding a broader challenge in Early Stage VC?

• Seed to Series A -- 22% graduation rate before, now it's 9%. Now Series A multiple is now 1.4x multiple. 24 months to 30 months. Expected revenue has been doubled.