AI’s Impact on Product Development & Angel Investing

As AI proliferates our products and society, I think product management, curation, and editing will become the most critical skills for builders and creators over the next few years. Seed investing will evolve substantially as small teams can iterate to product-market fit faster and get more leverage through technology.

I’m often over-optimistic about how quickly users will adopt new technology (e.g., VR and Crypto), but I already may have been pessimistic about AI, given the traction in the market. Progress on iteration speed (look at this LLaMA timeline) and consumer adoption (Chat GPT has over 100m users) has been staggering. For Chat-GPT, in particular, three factors unlocked something special:

  1. The public internet as training data.
  2. The massive increase in computational power (CPU and GPU) to process and store data.
  3. The large language model and “chat” interface, which many users find familiar.

AI will be incorporated into many parts of our life, whether it’s something we actively interact with (e.g., an AI Assistant) or passively integrated into products we use daily. I’ve been dabbling with the tools (e.g., Chat GPT-4, BardMidjourneyPrismaWindow) and wanted to share some thoughts that draw from my own experience and have yet to see others cover as widely.

Product management is essential

Product management will become an essential skill at all stages of product development because of AI. Prototyping will be easier for new ideas even if creators have weak design and engineering skills. We can go from “what’s in our head” to a prototype much faster if the human-computer interaction and iteration model is simplified to speech or text over code. Product management will also matter more at later stages of product development, where the essential skill is getting the humans in the loop all aligned and rowing in the same direction. There has never been a better time to be a product person.

Remixing with precision

AI supercharges “Remixing” but allows us to add “Precision.” Remixing is the process of taking text, designs, code, or any other body of knowledge and repurposing it for a new use case (I wrote about this for gaming a few years ago). Precision allows us to take remixed content and apply it very specifically to a new use case using AI without having to customize manually. Think about every sign-up flow or password reset process you’ve ever experienced — these could easily be precisely remixed for your use case (without the need for a designer or engineer).

Open base models with personalization

The base LLM models, built on mostly publicly available data, will converge to “good enough” as Bard, ChatGPT, and LLaMA (and its derivatives) all continue to iterate and asymptote towards a commodity. The real magic happens when you layer in personalization, personalization to you, your organization, your area of scientific research, etc. If I used all the information on my laptop, my personal Notion, Apple Health, and my communication channels, this would be incredible training data. I’d get insights about how my intentions and actual behaviors stack up and also could see my productivity and workflow improve an order of magnitude. This could easily extend to teams or entire organizations.

Curating, editing, and validating

AIs are great at confidently giving us inaccurate information and usually require tone and brevity to be tweaked so that the content will land with the intended audiences. As users of tools like ChatGPT, we will all need to become competent at curating the most impactful content, editing down or fleshing out areas to the right level of detail, and validating that the information we read is true. These are all skills we have today, but they will need to be more sophisticated as deep fakes become harder to differentiate from reality at first glance.

Blockchains for authenticity

One of the most challenging areas of AI from a consumer perspective is “validating” if the response from tools like ChatGPT is accurate. Part of the issue is that data processed through machine learning becomes “garbled,” and we’re unable to quickly drill down to a specific set of sources and their relative impact. An area of interest for me is how we can use blockchains and wallets as a way to “sign” digital content with the original creator’s wallet so we know that a piece of content was actually posted by Aadil at a specific time because the image/blog post was minted on-chain.

Seed investing evolves

Seed investing is going to evolve substantially. Seed investors can back teams with live prototypes (or products) because the cost and time to iterate on an idea will considerably reduce. Teams will stay small for longer and require less capital to 1) reach product-market fit and 2) scale their products and services so they will not be able to absorb large amounts of VC money. The era of megafunds may be over, and the generation of high-quality angels who invest small amounts per company (and probably make more investments than before) may be a winning strategy over the next 5-10 years.

I love new technology, and these platform shifts are always inspiring. I can’t wait to see what gets built next and follow the developments closely.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

Create a website or blog at

%d bloggers like this: