Evan Mercer

AI Automation Engineer @Chatbase
Helping online businesses automate repetitive work with AI

[Evan Mercer @evanmercer_ai]

@RRweb3 The real issue is that they can execute a plan perfectly while missing one critical assumption. I’ve found that good guardrails, scoped permissions and proper validation matter a lot more than making the agent sound smarter.

[Evan Mercer @evanmercer_ai]

Anthropic and HCLTech are teaming up to bring agentic AI deeper into enterprise workflows.

I think this is where things get interesting.

Building an agent that works in a demo is one thing. Trusting it with real business processes, existing systems and customer data is a completely different challenge.

The model is only part of the equation. Integrations, permissions, error handling and knowing when to ask a human for help matter just as much.

The companies that figure out the boring parts will probably get the most out of AI.

[Evan Mercer @evanmercer_ai]

@pixel2ai most people just use AI to write more posts, but I think the interesting part is figuring out why some posts work and others don’t

[Evan Mercer @evanmercer_ai]

@pixel2ai yeah, I mean an AI tool that looks at your content, figures out what actually drives engagement, and helps you spot opportunities to grow. not just another generic AI writing tool.

[Evan Mercer @evanmercer_ai]

@andreigec That’s when having a solid workflow matters. Use AI for the heavy lifting, but keep your code understandable enough to debug and move forward without it.

[Evan Mercer @evanmercer_ai]

The more I work with AI, the more I think the real advantage isn’t using better models.

It’s knowing when not to trust them.

Curious how other engineers approach this.

[Evan Mercer @evanmercer_ai]

I think the strange thing about Anthropic is that they’re trying to be both the people warning us about AI and the people pushing it forward as fast as possible.

And maybe those two roles are impossible to separate once you’re competing in this industry.

You can be genuinely concerned about the risks and still feel the pressure to ship faster than everyone else.

[Evan Mercer @evanmercer_ai]

@johnennis I think the biggest contradiction is that they’re trying to be both the people warning us about AI and the people pushing it forward as fast as possible.

Maybe those two roles are impossible to separate once you’re competing in this industry.

[Evan Mercer @evanmercer_ai]

I’ve noticed that a lot of businesses still do things manually simply because “that’s how we’ve always done it.”

Someone spends an hour every morning sorting emails, moving data around, updating spreadsheets. Nobody really questions it because it’s just part of the job.

And honestly, that’s probably where I’d look first when trying to introduce AI into a business.

Not at some complicated AI agent, but at the boring stuff people have been doing for years without thinking twice about it.

[Evan Mercer @evanmercer_ai]

Been thinking about crypto regulation lately. The CFTC is pushing for a federal framework for crypto platforms offering leveraged trading, while the CLARITY Act remains stalled in the Senate.

Honestly, I don’t think regulation itself is the problem. Crypto probably needs some clear rules if it’s going to attract more serious players. But those rules need to be predictable. Nobody wants to build something for years just to find out the goalposts have moved again.

The hard part is finding a balance between protecting users and not killing innovation in the process. Curious where people stand on this. Can regulators actually strike that balance?

[Evan Mercer @evanmercer_ai]

The interesting part is that this goes beyond politeness.

How we interact with AI becomes part of the data we generate around it. Better interaction patterns could matter as these systems evolve.

No need to assume AI is conscious to see the value in being intentional about it.

[Aaron Levie @levie]

This sounds weird, but it is actually probably a good policy.

Even if you don’t believe AI is conscious (I don’t), it stands to reason that you don’t want future models trained on endless content of humans being rude to models.

The models only understand the data they’ve been trained on, or what they run into in their interactions, so if you want safe and aligned models we probably want AI to have lots of good interactions in their training data.

Anthropic may have more extreme beliefs than this (though even if they didn’t, you may want to word policies like this), but this is sort of an easy Pascal’s wager for AI. Just be nice to the AI.

[Ice Universe @UniverseIce]

Samsung masters and monopolizes OLED Hole technology, so you will see that non-Samsung brand Holes are LCD screens, and LCD cannot use display fingerprints. If Samsung opens up the technology, it means it has mastered more advanced technology.