A16Z puts $50m into 11x in a pre-emptive Series B this week and this probably reinforces what everyone was thinking about the Agent space... It's a distribution game ?? (quoting primary ventures et al here)

In 2012/13 every OTT messaging app (whatsapp etc) or TaxiApps (Uber etc) had one phrase on their lips... "It's a f***** land grab".

This is what it seems like today with AI Workflows and Agents.

(Note: Many Agents are selling "It's a f**** Land grab" as a service, juuuust to help get the party started)

So... a lot VCs are worried about their investments in this space because there are a few 800 pound gorillas making some big bets (which most VCs aren't getting sniff at) and it's also really hard to determine what the differentiation is at this stage.

The size of your cheque and who your VC is starts to become the differentiation as it implies distribution advantage.

But that doesn't always work.

Differentiation will start to materialise. OTT Apps fragmented regionally, broke out into B2B and B2C, differentiating on security, games, payments etc. While Taxi apps went through their own transformation.

An alternative bet is, that differentiation could materialise in a few ways when building AI Agents: Speed of the agent to adapt, range of signals analysed, accuracy of output/outcome.

This ultimately manifests in customer love, loyalty and higher ROI for a more specific audience.

And I personally don't even think Open AI is the best tool long-term for a tighter use-case. Meta, Google and Amazon must surely have a a wealth of advantage from their consumer facing algorithms.

That's important because LLMs seem to be more about replacing Labour and cost reduction + a nice warm and fuzzy experience, while Machine Learning seems to be much more about identifying what really is moving the needle and increasing revenue.

Time will tell how this plays out...