3 Test for Turning Ideas in Assets
How to Validate Before Scaling
This newsletter began with the topic of AI workflows for wealth. I like the name but fear it may be too heady. I'm currently liking the phrase “Ideas into Assets.”
AI is more than an automation tool. More than just an agent or assistant. If the right strategy is applied, AI becomes an asset…
I may or may not change the title of this newsletter, but I am working on a course called Ideas into Assets.
The Ideas into Assets process is seven steps. But with those steps are three different tests.
If you’re working on AI use cases, workflows or any implementation, here are 3 tests that could be beneficial for you.
LLM Tests
The first test we need to do is testing the LLM. There are multiple companies and each has multiple models. ie, OpenAI has five or six models, and Claude has, I think, three or four. Google Gemini has different models too. Each model has a style, personality and a specialty.
So you should take your idea and run it through at least 2-3 different models.
You could test the following:
Which models will work best with your idea?
Which can follow your instructions consistently?
Which model environment/platform is most conducive to your workflow?
^ that last question really gets to the next test.
Team Tests
At this stage we need to go beyond how well the LLM works individually, but how does it work in a shared environment. Do you need a team account or can you just change sharing permissions? Do you have a preference between GPTs, Gems or Projects/Artifacts?
You may also wonder about integrations for speed and ease of use?
Or what about privacy and security?
Finally, we have test when creating client side or public facing.
Public/Client Tests
The last test we need to consider is when our asset or workflow is used in public situations. Maybe we are creating a tool that works with our clients or even leads on landing pages. What are some considerations that we need to think about here?
What LLM works best for public-facing uses?
Do I need a 3rd party integration?
Do I need or want multiple integrations, such as a CRM, a RAG or payment processor?
Conclusion
These considerations are based on your use case. There's no right or wrong way to go about it, but following best practices ensure safety and efficiency.
✓ It's best to begin with the end in mind.
✓ Be thorough and test out different options.
✓ Always keep your eye on what's new and what's changing, because what works today might not be best in a few months.
If you need help learning, testing and staying connected to what's new, check out my AI community launching in January.
Create a Great Day!
Arvell Craig
arvellcraig.com