AI creating other AI
Where does your value lie if not in the (AI) coding?
Every GTM operator 'needs to' build AI workflows and agents.
Think building them is hard? Think again.
It's becoming more and more simple.
We are at the beginning of a big change.
AI can now create other AI.
That custom workflow that scores and outreach to ICP?
You can build it in few hours with Clay and Smartlead.
That account research algorithm you paid $15k last year?
I've built it in 15 mins with AirOps co-pilot.
___
So, where does your value lie if not in the coding?
First of all, you'll still need coding for advanced stuff.
But here's what imho AI alone can't help with (yet):
1. ๐๐๐๐ถ๐ป๐ฒ๐๐ ๐๐ป๐ฑ๐ฒ๐ฟ๐๐๐ฎ๐ป๐ฑ๐ถ๐ป๐ด
Which business problems are worth solving based on your unique context and situation
โณ Need to be a hands-on operator, in the trenches
2. ๐ง๐ผ๐ผ๐น๐ถ๐ป๐ด ๐ธ๐ป๐ผ๐๐น๐ฒ๐ฑ๐ด๐ฒ
Knowing which AI and workflows are best to solve your specific challenge
โณ Gotta check LinkedIn, X, Product Hunt and stay current
3. ๐ฆ๐๐๐๐ฒ๐บ ๐๐ต๐ถ๐ป๐ธ๐ถ๐ป๐ด
Architect how multiple specialized AIs can share data inputs and output actions among each other
โณ Gotta know some APIs and data pipelines
4. ๐ฃ๐ฟ๐ผ๐บ๐ฝ๐ ๐ฒ๐ป๐ด๐ถ๐ป๐ฒ๐ฒ๐ฟ๐ถ๐ป๐ด
Understanding how to guide the AI to carry out the job in the exact way you want them to
โณ Need to be an excellent, crystal-clear (technical) writer
___
Riccardo Vandra calls this the "Implementation Paradox."
(creds to him who inspired this post!)
He says:
"As building gets easier, knowing what to build becomes exponentially more valuable.
The businesses winning with AI aren't the ones with the best engineers. They're the ones with the clearest problem definition."
I might be biased, but the answer to the above to me is:
GTM Engineering.
They don't just automate because 'we can'.
But they work back from prioritized business challenges.
In 2025, anyone can build AI agents because 'we can'.
But it's much harder to know what's worth building.