At Momentum91, we've been hard at work understanding AI and their use-cases for mid-market and large firms. Late last week, I had the opportunity to share everything we've learnt so far with eChai Ventures at GVFL Limited.
Here are some notes:
๐ ๐๐ข๐ฌ๐ญ๐ซ๐ข๐๐ฎ๐ญ๐ข๐จ๐ง ๐ข๐ฌ ๐๐ฏ๐๐ง ๐ฆ๐จ๐ซ๐ ๐๐ข๐๐๐ข๐๐ฎ๐ฅ๐ญ
Other people also have AI. A lot of folks are building on top of AI and so there is increased competition in almost every vertical. Every niche software has like 3 bootstrapped competitors who are going to market via SEO. Also, product differentiators are difficult to communicate in the age of low attention spans. If you are starting distribution today, it needs a thorough re-imagination.
๐ ๐๐๐จ๐ฎ๐ญ ๐๐๐ซ๐ซ๐จ๐ฐ/๐๐๐๐ค ๐๐
Instead of building systems that are kind of smarter at doing a lot of things, you might want to consider building multiple and separate systems that are very smart at doing a few things extremely well. Enterprises/companies will separately pay for solving specific problem areas and addressing their use-cases. The game is depth, not breadth.
๐ ๐๐๐ญ๐ข๐ฏ๐ ๐๐ ๐๐ฑ๐ฉ๐๐ซ๐ข๐๐ง๐๐๐ฌ
Companies like ClickUp and Notion have done an amazing job at bringing AI within their products. But they are still not AI native. They were conceptualised when AI was not around. They have fit AI retrospectively. There is a large opportunity for building Native AI experiences. Platforms and products that are conceptualised from ground-up in the age of AI. An example of this is a project management system where I can assign tasks to both humans and agents. It will demand a lot of capital and ingenuity but it can be done. Here's a good question to ask: What other 'systems of records' exist that can be made AI native?
๐ ๐๐จ๐ฅ๐ฏ๐ ๐๐จ๐ซ ๐๐๐๐ฎ๐ซ๐๐ญ๐ ๐๐๐ญ๐
AI will solve for analysis, predictions, pattern recognition etc. Most firms do not have accurate data. If you can figure out a way for a business to collect accurate data about their prospects, customers, employees, products, sales, etc - using AI, you can do anything and everything. Figure out a way as to how you will solve for collection, storage and organisation of accurate data.
There were a few other things that we discussed.
๐ Co-pilot and Autopilots
๐ AI wrappers
๐ AI & Funding
๐ Verticalising LLMs
๐ AI Safety
๐ Real-life implementations of AI by companies we have not heard about
This conversation was meant to be a starting point for folks thinking about building in the AI space. It was by no means exhaustive - but sort of a summary of all the research Harsh, Jay, Koushikram and I have been doing for the past few months.
If you want the deck, let me know in the comments and I'll send it over ๐
Thank you Jatin for the opportunity.
Until next time :)
Here are some notes:
๐ ๐๐ข๐ฌ๐ญ๐ซ๐ข๐๐ฎ๐ญ๐ข๐จ๐ง ๐ข๐ฌ ๐๐ฏ๐๐ง ๐ฆ๐จ๐ซ๐ ๐๐ข๐๐๐ข๐๐ฎ๐ฅ๐ญ
Other people also have AI. A lot of folks are building on top of AI and so there is increased competition in almost every vertical. Every niche software has like 3 bootstrapped competitors who are going to market via SEO. Also, product differentiators are difficult to communicate in the age of low attention spans. If you are starting distribution today, it needs a thorough re-imagination.
๐ ๐๐๐จ๐ฎ๐ญ ๐๐๐ซ๐ซ๐จ๐ฐ/๐๐๐๐ค ๐๐
Instead of building systems that are kind of smarter at doing a lot of things, you might want to consider building multiple and separate systems that are very smart at doing a few things extremely well. Enterprises/companies will separately pay for solving specific problem areas and addressing their use-cases. The game is depth, not breadth.
๐ ๐๐๐ญ๐ข๐ฏ๐ ๐๐ ๐๐ฑ๐ฉ๐๐ซ๐ข๐๐ง๐๐๐ฌ
Companies like ClickUp and Notion have done an amazing job at bringing AI within their products. But they are still not AI native. They were conceptualised when AI was not around. They have fit AI retrospectively. There is a large opportunity for building Native AI experiences. Platforms and products that are conceptualised from ground-up in the age of AI. An example of this is a project management system where I can assign tasks to both humans and agents. It will demand a lot of capital and ingenuity but it can be done. Here's a good question to ask: What other 'systems of records' exist that can be made AI native?
๐ ๐๐จ๐ฅ๐ฏ๐ ๐๐จ๐ซ ๐๐๐๐ฎ๐ซ๐๐ญ๐ ๐๐๐ญ๐
AI will solve for analysis, predictions, pattern recognition etc. Most firms do not have accurate data. If you can figure out a way for a business to collect accurate data about their prospects, customers, employees, products, sales, etc - using AI, you can do anything and everything. Figure out a way as to how you will solve for collection, storage and organisation of accurate data.
There were a few other things that we discussed.
๐ Co-pilot and Autopilots
๐ AI wrappers
๐ AI & Funding
๐ Verticalising LLMs
๐ AI Safety
๐ Real-life implementations of AI by companies we have not heard about
This conversation was meant to be a starting point for folks thinking about building in the AI space. It was by no means exhaustive - but sort of a summary of all the research Harsh, Jay, Koushikram and I have been doing for the past few months.
If you want the deck, let me know in the comments and I'll send it over ๐
Thank you Jatin for the opportunity.
Until next time :)
