HOW and HOW NOT to approach Moat in AI products

What the heck is Moat?

"Moat" in the context of a startup or business refers to a concept that originated in the field of investing.

The term is derived from the water-filled moats that surrounded medieval castles, providing them with a practical and effective defence.

In business terminology, a moat represents a company's sustainable competitive advantage.

1. Brand Moat: Strong brands, like Apple or Coca-Cola, use their reputation as a competitive edge.

2. Network Effect Moat: The value of businesses like Facebook or Amazon increases as their user base grows.

3. Cost Moat: Companies that offer lower-cost goods or services than competitors have a cost advantage, potentially due to economies of scale.

4. Regulatory Moat: Legal barriers or regulations can deter new market entrants, creating an advantage.

5. Technology/Innovation Moat: Superior technology or faster innovation gives companies a competitive edge.

6. UI/UX as a Moat: Your user experience is something that is just irreplaceable. You have a secret sauce that makes your UX irreplaceable and hard to copy.

7. Domain Knowledge as a Moat: A company’s knowledge of a particular problem that they are solving could be a Moat.

MOAT is elusive. Now, more than ever!

For a technology startup, technology innovation has always been considered the best MOAT. A lot of great companies are built on it. Google is the best example. The algorithmic advantage Google started generating in search is to date not beaten.

But what’s changing? 

With AI and foundational models, technology is becoming more and more commoditised. Building a product with key technology differentiations is becoming easier every day!

Rethinking MOAT for AI Startups

Here are a few paths you CAN bank on and a few ways you CANNOT bank on to create a Moat, if you are building an AI startup.

DO Bank on these »

1. Depth of Domain

Horizontal use cases vs Vertical use cases: The most generic use cases, with a wide market, are the ones the foundational players (read Google, Microsoft, OpeanAI) play in. Writing is one of those horizontal use cases.

Any tool that is solving the most obvious problems with the largest market stands the risk of getting bulldozed by one of the future releases of GPT, Bard etc.

For a startup, the way ahead is to build for vertical, deep problems. One could take the industry vertical approach (solve deep problems in banking, finance, healthcare etc) or the technology vertical (go deep into solving specific technological issues)

2. User Experience [Conditions Apply]*

Yes, UX CAN be a moat for your product. But NOT every problem requires a UX uplift.

> Chat is NOT your friend. You are competing with Chat: With ChatGPT and Bard in the block, every tool that is built on top of LLMs is competing with the Chat UX. What is the delta you can bring into your UX, compared to a chat UX? In most use cases it is super tough to bring in enough delta to lure your users.

> You are competing with intelligence: Intelligence is no longer a differentiation. It is a commodity. You need to go deeper.

Does the current user experience have a core problem that required a specific workflow?

For example,

• Context length limitations: Context length limitations prevent Chat UI from being an effective way of delivering the UX. You need something deeper, to learn about the wants of the user and deliver that.

• Dynamically changing context that needs version tracking: If the context itself changes very often. For example, if you want to read a stream of emails and learn from them to answer queries, it becomes a key advantage for you.

• Security of data: The data security needs of your user prevents them from using a general-purpose system. You need to find out workflows to make it happen without compromising on data security.

All these could be interesting issues that prevent the normal UX from performing well.

3. Network Effect

I cannot think of a single company that has created this yet, but an AI startup CAN have a network effect. Something so strong that users would just want to stay in the product because they have strong networks built into the product.

In the traditional software world, all social networks thrive on this. It feels like science fiction the moment I think about how that effect could come into place in the AI world.

• Will there be a bot who knows so much about you that you do not want to part with and go to another one? For emotional reasons or practical reasons?

• Will there be a new social network emerging where AI agents can connect with us so well that they feel like humans?

I only have questions. No answers on this one.

DO NOT Bank on these »

1. Community CAN NOT be a Moat

Now this could be a hot take that is questionable. Here is why I think the community cannot be a moat in a space that is so dynamic.

Communities are built around one or more of these three things > Inspiration/Admiration, Content, and Learning.

If you think that you cannot keep a group of people motivated with AT LEAST one of these, for a long enough period, you do not stand a chance to create a moat from the community.

2. Cost CAN NOT be a Moat

The big problem with AI applications is the downstream cost.

There is always a hard limit to which you can make cost your advantage without burning yourself too much!

3. Regulatory and Compliance factors CAN NOT be a Moat

This is a tricky one. Being a space that is so new, the regulatory landscape is evolving.

If you see a regulatory requirement as a key Moat in your product, it might work very well in the short term, but it will be highly unpredictable to bank on for the long term.

Take Crypto as a classic example. Hundreds, if not thousands of startups have lost their life in this regulatory mess.

Concluding…

If you are a startup founder breaking your head over finding the Moat, think about these.

AI is going to be a foundational shift in the way software is going to be consumed.

As history proves, it is always new thoughts and ideas that will come and succeed in these foundational shifts!