OK, now AI-tools let us build new software products in days rather than months. And one of the main things to think about for tech vendors and outsource companies: the gap between the cost of 'working demo' and 'production-ready' product has become huge.

A note: Ideas are my own, not my current employer.

I've been working in B2B fintech. Here I met many founders of startups. Now many of them at the time of our conversation already have designs or prototypes of apps. Previously the entry price to launch an app was a blocker even to try an idea. Now a working prototype cost is very cheap.

Even before AI-era many founders were shocked seeing the difference in price estimates provided by cheaper outsourcing services and specialized development services. Reason often was in the dev scope: experts often see risks and try to prevent them. For example: reflecting trades happened on a centralized exchange with blockchain is expected to be a simple part — "just send requests to the API". But guaranteeing that both counterparties will get their funds in case of any issues with transactions or order updates is not easy to design and implement.

Now AI-based tools made entry level projects extremely cheap. So I can imagine disappointment of company owners who already saw that their working application created in days with Cursor from design mockups and now they see that to have a fully operational product they need to pay 100x their previous expenses.

My thoughts:

  1. As the result of this shift the outsourcing companies will have even more difficulties justifying the hourly rate of their developers for clients. So the competition in the niche for starting of new products or building MVP will be even more severe. Not a new problem for them but still.
  2. There will be more need for "transition" services that will help to assess the pile of code previously generated by AI, audit it and make it compatible if the growing business needs (scalability, performance, regulation etc). It's where products survived the first rounds of competition will end up. It will be pricey. Many AI tools will go for this niche as well: creating documentation for the code and auditing for example.
  3. Outsourcers and tech vendors will adopt a clear three-tier pricing:
    - cheap rates for initial AI-assisted product launches
    - Premium fees for transforming prototypes into production-ready systems
    - Moderate pricing for ongoing feature development of established products
    The 2nd tier will be even more in demand: a few people like to write documentation, read spaghetti code. But the "boring" things are where the profit is, right?
  4. In fintech, vendor reputation will matter even more. New vendors focusing on cheap development services will face reputation damage when problems arise. Established vendors offering tier 2-3 services will likely distance themselves from these entry-level providers.