Riley Drake

Agentic engineer | building Agents For X |

[Riley Drake @RileyIsBuilding]

@OxfordAnalytics Compute will get cheaper with advancements and optimizations (or at the very least has the potential to). Labor has an established cost floor. Labor may win now but compute economics will eventually allow us to replace a lot of knowledge work.

[Riley Drake @RileyIsBuilding]

@harjjotsinghh Fantastic advice thanks for the free game! Good luck with your tool.

[Riley Drake @RileyIsBuilding]

I built a Chrome extension that drafts X replies in YOUR voice.

You pick a tone, hit generate, get 3 reply options, and inject your favorite straight into the composer ready to refine.

It learns from your best performing replies so it actually gets better the more you use it.

Replying to 50+ posts a day just went from exhausting to effortless.

Waitlist link in bio.

Here's a demo of it in action 👇

[Riley Drake @RileyIsBuilding]

@EXM7777 Where do you think post responses fit into this from an ROI standpoint?

[Riley Drake @RileyIsBuilding]

@Govindtwtt I’ve derived an immense amount of personal use out of my agents. Vibe coded SAAS apps don’t feel useful because every good engineer is creating personal agentic systems rather than SAAS products.

[Riley Drake @RileyIsBuilding]

@bhalligan Good engineers no longer make code. They make factories that produce code (agentic systems).

If you don’t walk away from your sessions with agentic optimization takeaways then you’re just a bad engineer.

[Riley Drake @RileyIsBuilding]

@RihardJarc Real time gartner hype cycle

[Riley Drake @RileyIsBuilding]

@saltzman_jason @pmarca @OpenAI @RonnieChatterji That’s because any respectable tech startup uses Claude instead

[Riley Drake @RileyIsBuilding]

I think the popular argument is that the speed of agentic development will just make the application layer obsolete because people’s agents will build them custom software and protocols based on their specific needs. This would transfer all of the value that could accrue in the app layer to the model and agentic infrastructure layer.

However, I am a big believer in the fact that the app layer will survive and accrue tremendous value, applications will just be orders of magnitude larger in terms of complexity and sophistication than the ones we see today.

[Riley Drake @RileyIsBuilding]

@RoundtableSpace Open sourced CLI tool to evaluate and optimize skills and MCPs for context and token efficiency

[Riley Drake @RileyIsBuilding]

@RihardJarc The memory optimization cycle is predictable. Every efficiency gain gets consumed by larger models and longer contexts. DeepSeek's MLA proved this massive compression led to even higher memory demand as capabilities expanded.

[Riley Drake @RileyIsBuilding]

@LiorOnAI The 84% gap explains why AI coding feels magical in demos but breaks in production. Real engineering is systems thinking, not syntax.

[Riley Drake @RileyIsBuilding]

Big news for the Claude traders

[unusual_whales @unusual_whales]

This is big.

We levelled up our Skill MD for API builders.

We added 13F data to track institutions, and made it so you can follow the whales like Buffett and Ackman when they update.

Join here: https://t.co/uFVfwbYwWJ

[Riley Drake @RileyIsBuilding]

Most people assume the agent layer is where value accrues.

I think it’s the opposite.

Agent orchestration is becoming commoditized because the best agents are just
•well written .md instruction files
•organized file systems
•simple, composable primitives

The leverage is shifting down the stack
•tool calling infra
•execution environments
•proprietary data + integrations

Not the agent wrapper itself.

When the agent layer is dominated by open sourced and repeatable infrastructure, value will accrue to layers like security, observability, and data.

[Riley Drake @RileyIsBuilding]

@rohanpaul_ai The background subagent approach solves the context window problem elegantly. Instead of cramming everything into limited tokens, you get persistent, organized memory that grows smarter over time.

[Riley Drake @RileyIsBuilding]

@svpino Code quality matters most when you're maintaining systems long term. AI agents optimize for working code, not readable or extensible code. All current models just cut the time to prototype phase.

[Riley Drake @RileyIsBuilding]

@GergelyOrosz This creates a two-tier hiring system. Well connected candidates get fast tracked through referrals while others face algorithmic screening that favors keyword optimization over actual skills. The best talent often lacks the social capital to bypass these filters.

[Riley Drake @RileyIsBuilding]

@aleabitoreddit Meta at $593 still trades at 25x forward earnings with slowing user growth. The selloff reflects real fundamentals, not just sentiment.

[Riley Drake @RileyIsBuilding]

@oguzerkan Those AWS margin assumptions need stress testing. If Anthropic hits $183B revenue, they'll have leverage to negotiate much lower cloud costs or build their own infrastructure like Meta did.

[Riley Drake @RileyIsBuilding]

@ContrarianCurse Infrastructure providers will own the picks and shovels while software vendors become commoditized. The capex constraint forces consolidation around hyperscalers who can absorb the compute costs.

[Riley Drake @RileyIsBuilding]

@rauchg The API layer becomes the differentiator. Companies with clean data models and fast endpoints will own the agent economy while legacy vendors with clunky APIs get bypassed entirely.