Why is on-chain automation still so clunky in 2025?

Been thinking about a nagging problem in DeFi that doesn't get enough attention: the janky state of automation.

It's 2025, yet if you want to do something as basic as a dollar-cost average (DCA) buy or set a real stop-loss on-chain, you're usually forced to use an external service. You're either paying fees to a keeper network like Chainlink/Gelato or trusting some random bot. It feels like a weirdly centralized and fragile solution for a supposedly decentralized world.

This leads to the question: why can't the blockchain just handle this itself? Why can't we just tell the protocol, "execute this for me when X happens"?

I was digging around and saw that Aptos Labs is trying to tackle this with a feature they call "Event-Driven Transactions." The idea is to bake automation directly into the L1. So you could, in theory, schedule transactions, set triggers based on price, or chain actions together without an external keeper.

This sounds great on paper, but it immediately made me wonder: If this is such an obvious solution, why aren't more chains doing it?

There must be reasons why giants like Ethereum and Solana have historically relied on third-party services for this. My guesses are:

→ Bloat + Complexity: Maybe building this into the core protocol is incredibly complex and could slow the chain down or introduce new bugs. State management for millions of pending "if-then" transactions sounds like a nightmare. → Security Risks: Does this open up new attack vectors? What if the price oracle it relies on gets manipulated, triggering a cascade of wrongful liquidations? → Economic Model: Is the keeper-as-a-service model (like Chainlink's) just more sustainable? Maybe the fees generated by keepers are essential for their security and it's a model that works.

So, I'm just throwing this out to the community, especially the devs:

1. What are the real technical trade-offs of building automation directly into an L1 versus using an external network?
2. Am I missing other projects that are already working on native on-chain automation? I'm genuinely curious about other approaches.
3. For those of you who use keeper services, what are the biggest limitations? Would a native solution even solve your main problems?

It feels like we're at a crossroads where we either accept the external bot/keeper model as "good enough" or someone figures out how to make automation a native function of a blockchain. What do you all think is the more likely future?