Future Proof Your Products

One question that keeps coming to mind is how do we future-proof the things that we’re currently building?

Based on my experience with engineering and working with highly capable models like Claude’s latest Opus 4.7 or even their Sonnet models, I’ve found that AI is 1,000% the future. There is no denying that it can extremely increase people’s productivity. Prototypes and MVPs that used to take weeks, possibly even months, now take days, sometimes even hours.

Now, what does this mean for SaaS?

Expectations will go up. It is inevitable. There are several things that need to happen in order for you to future-proof your products, and they all happen to be making sure that AI can use them to their advantage.

Here are the 3 core layers:

1) MCPs

MCPs are, simply put, an agent API, a way for AIs like ChatGPT and Claude, allowing them to access your product without even touching any sort of interface. This is also the most secure way currently for these agents to access data through an authentication system.

2) Skills

Skills can be driven to give users immediate access to some sort of value. The point is to create a top-level layer that allows agents to think in a way that leverages the MCPs of your SaaS.

3) Open Your Information Moat (AEO)

Make information around the way that your product works open and accessible. Without this, AI isn’t going to have any way to understand what your tool does. The more in-depth the information is, the better.

These are the three core areas that I’ve been working on for the past year, and starting to see the value of everything coming together in an agentic world.

1 Like

Love these ideas, especially around AEO.

Agents will need structure and a lot more depth . AI makes building fast, but not clear without guardrails. AEO makes understanding necessary.

From what I can tell, AEO isn’t just opening information. It’s exposing how your product thinks:

  • what problems it solves
  • how it makes decisions
  • what inputs matter
  • what outputs mean
  • where it’s strong or weak

I liked this idea, so I made a Rubric to evaluate the Glare documentation pages:

Fully agree here! AI likes being able to grep content. Also, when things are organized, it makes it 10 times easier to parse.

Super cool for the rubric @Bryan!