electric_muse

It’s a mixture. But yeah, today I pushed the boundary further.

A few days ago I flagged a piece someone else had written with ai. It has a specific cadence and some typical patterns. But many people seemed to buy it before I commented.

So I’ve been running a bit of an experiment lately where I write posts and LLM-ify them in varying amounts to see if an LLM actually produce something more upvotable and when people start to notice.

I started out just saying “rephrase this so it sounds tighter” and moved recently towards just jotting rough notes and saying “make an HN comment out of this” and then editing.

Today was the most AI-influenced set of posts, and people clearly noticed, so that’s definitely the threshold. Fascinating.

Check my post history to see the evolution. Using gpt-5.

[flagged]

[flagged]

The piece was deeply thought-provoking, but I struggled to get through it sensing how much AI was used to write it.

I’ve been drafting a manuscript for a novel lately, trying to see how well llms can help.

I recognize this prose immediately as OpenAI gpt 5.

It loves to describe things “hum” that don’t usually hum, like the author wrote in the beginning. Plenty more descriptions match the cadence and rhythm and word choices I’ve seen writing my manuscript.

I feel like there’s a meta discussion the author was prompting here about consciousness.

Reading the writing of a real human feels more intimate. Reading the auto-tune version of writing makes me feel noticeably less connected to the reader. I know the author still input something to get this output. But there’s something blocking a deeper connection when I just “know” I’m not reading the author’s words.

Edit: llamas > llms

Hard-line answers like that are why we're all building our own cages from a privacy perspective.

Thanks! I’ll look into these. Thought it was simpler lookup of metadata, but these seem like robust offerings that make sense for even more protection.

I have a side project that uses Twilio. Probably going to have to face this dragon at some point and go down the same risk assessment.

What values for Twilio line type intel do you block? (or equivalent values if you don't use Twilio)

For example, is it just `nonFixedVoip` that should get the boot? Or are `fixedVoip` also spammy?

of course - but aren't there valid voip users out there? or should we just consider all voip users as not worth serving?

The visual system is so subjective[1]. Even cultural differences, prior expectations, and our environment can impact “what we see.”

1: https://journals.plos.org/plosone/article?id=10.1371/journal...

What are the latest tips and best practices for acquiring a phone and service without having to deanonymize?

For example, can you just walk into Best Buy with cash?

I just submitted my first big open source contribution to the OpenAI agents SDK for JS. Every word except the issue I opened was done by AI.

On the flip side, I’m preparing to open source a project I made for a serializable state machine with runtime hooks. But that’s blood sweat and tears labor. AI is writing a lot of the unit tests and the code, but it’s entirely by my architectural design.

There’s a continuum here. It’s not binary. How can we communicate what role AI played?

And does it really matter anymore?

(Disclaimer: autocorrect corrected my spelling mistakes. Sent from iPhone.)

They also discontinued the screenshot / recording tool I used. Buggy and slow. But it integrated well.

Are they just thrashing around at this point without vision?

Most of what I’ve seen launch from them has been… undercooked.

Wait, is this real?

That's ironic. I think local MCPs are an auth nightmare.

Just think of all those plaintext auth tokens sitting in well-known locations on your machine.

It's a black hat dream.

We'll see, but I think commercial use of local MCPs is going to be constrained to use cases that only make sense if the MCP is local (e.g. it requires local file access).

For everything else, the only commercially reasonable way to use them is going to be remote streamable HTTP MCPs running in isolated containers

And even then, you need some management and identity plane. So they're going to likely be accessed via an enterprise gateway/proxy to handle things like: - composition -- bundling multiple MCPs into one for easier connection - identities per-user / per-agent - generation of rotatable tokens for headless agents - filtering what features (tools, prompts, resources) flow through into LLM context - basic security features, like tool description whitelisting to prevent rug pulls

MCP is only a protocol, after all. It's not meant to be a batteries-included product.

I think MCP has legs well beyond just the LLM / agent world. Just like USB went from "how I connect my mouse" to "how I charge my beard trimmer."

In fact, I imagine it's going to go full-duplex with all our systems, becoming a more standard way for systems to communicate with each other.

Under the hood, MCP is just JSON RPC, which is a fine format for communicating between systems.

MCP layers on some useful things like authentication and discovery. Both are critical to any kind of communication between systems built by different authors (e.g. various apps and services). Discovery, especially, is the fascinating part. Rather than hoping an OpenAPI spec exists and hoping it's right, MCP has this exchange of capabilities baked in.

I spent the last 9 years building integration technology, and from that perspective, the discovery-documentation-implementation problem is the core issue.

Right now, LLMs basically "solve" the integration problem because they can do the mapping between external tools/resources/formats and internal ones.

But there's nothing that strictly "requires" an LLM to be involved at all. That's just the primary reason to develop MCP. But you could just as well use this as a way for integrating systems, making some bets on interface stability (and using LLMs for cases only when your prior expectations no longer hold and you need a new mapping).

The comparison is perhaps imperfect and overused, but I feel like we're witnessing the birth of a new USB-like standard. There's something right now that it was designed to do, but it's a decent enough standard that can actually handle many things.

I wouldn't be surprised if in some period of time we see enterprise apps shift from REST to MCP for bi-directional integrations.

For the OP, I'm not sure if you're working on an MCP proxy (A) as a commercial offering, (B) as something for your team to use, closed source, or (C) as something open source for fun. But we just built and started selling an MCP proxy/gateway. It handles identities for humans & bots, tool allowlists, and policy setting for an org.

If you don't want to build something on your own because of option B above, get in touch.

This is 100% a problem with the MCP spec: it does not currently provide a way to narrow what tools, and therefore context, flow into the LLM.

I don't really think there's an easy solution at the protocol level, since you can't just make the LLM say what tools it wants upfront. There's a whole discovery process during the handshake:

LLM(Host): Hi, I'm Claude Desktop, what do you offer?

MCP Server: Hi, I'm Salesforce MCP, I offer all these things: {...tools, prompts, resources, etc.}

Discoverability is one of the reasons MCP has a leg up on traditional APIs. (Sure, OpenAPI helps, but it's not quite the same thing.)

I'd be interested in hearing other recommendations or ideas, but when I saw this, I realized that the spec effectively necessitates a whole new layer exist: the gateway plane.

Basically, you need a place where the MCPs can connect & expose everything they offer. Then, via composability and settings, you can select what you want to pass through to the LLM (host), given the specific job it has.

I basically pivoted my company to start building one of these, and we're getting inundated right now.

This whole thing reminds me of the early web days, where the protocols and standards were super basic and loose, and we all just built systems and tools to fill those gaps. Just because MCP isn't "complete" doesn't mean it's not valuable. In fact, I think leaving some things to the community & commercial offerings is a great way for this tech to keep winning.

Perhaps the difference is that you’ve been fortunate enough to never need Airbnb support to help you on something serious.

When things go well, it’s amazing. When things go poorly, you realize how anti-guest the policies and support team behave.

That’s when it changed for me. I realized how pro-host, anti-guest they are. Hotels generally seem to care when something goes wrong. Airbnb support behaves like you have inconvenienced them by raising the issue.

I was surprised to read that, too. I would love to hear more about it.

Thanks for such a great ELI5

You’re right. It’s all part of the natural business cycle, it seems.

As the power shifts to the employees, it becomes possible for employees to reduce output with impunity. Employers don’t have much choice, since you can barely fill the positions you have. It’s more expensive to replace a low performer than to just keep them.

But if some employers are forced to do layoffs, citing the economy, other employers definitely can seize this opportunity to embrace the power pendulum swinging back.

As some anecdata for seasoning, some of the higher ups at Salesforce told me they were pretty ticked off about how the employees seemed to have declined in engagement and productivity. So I bet this is a joyous occasion to let the bottom rung go and move the power pendulum back to the employer. You drop the bottom and expect more from those left. Double win.

Exactly this.

Communicate:

-validation from VCs who should know whether this company has a viable product

-runway to be able survive and build/improve

To:

- prospective employees

- prospective customers

As someone who recently did this kind of funding announcement, it can really help the business to chest beat. It doesn’t have to mean the priorities of the company are wrong.

Good question. We used the pentatonic scale, so that there wouldn't ever be any dissonance when multiple notes were played simultaneously

https://en.wikipedia.org/wiki/Pentatonic_scale

Not familiar with /dev/snd0 -- what would cause the sound to play?

Wow. This seems particularly helpful since it allows you to process more information when your sense of sight is otherwise busy.

Nice work. Somehow this reminds me of Harvest Moon 64 music. I'm not sure I felt these were necessarily catchy (but I think that comes from repetition), but they were incredibly pleasant and interesting.

Also, obligatory re: earworms: https://www.youtube.com/watch?v=iDjX4-LKqCA

Great points! Thanks for hopping in and adding extra color.

Navigating a whole portfolio through a flat/down environment sounds incredibly stressful.

If the driver of these changing environments is predominantly investor perception, public early warnings seem like they would accelerate or exacerbate them.

For example, if you publicly announce you’re expecting flat or down rounds (and similarly lowering your offers on new deals), it’s almost like applying downwards price fixing pressure. The next investor in your businesses feels they can also offer less without losing the deal. In this way, the warning manifests the crisis.

To protect the portfolio valuations, it would then seem strategic to prepare and react in private. But public warnings seem more strategic towards lowering valuations of new deals.

That’s why I’m naturally skeptical about the motivation for signaling.

You're right -- it's a bit harsh. I've been on both sides of the investment committee meetings. As a result of that, I've seen (and delivered) this messaging before.

Matt's an excellent investor and someone I respect and trust. So are the other investors putting up orange flags for their portfolio. It's smart to be prepared early.

But it always seems like investors get giddy with excitement about saying the "sky is falling" and things are about to become far less founder-friendly. This tends to cause a lot of unnecessary panic within the entrepreneur community, especially new founders. And I've seen people make decisions that contribute to manifesting the very situation they're scared of.

I've checked in with a handful of friends with their own funds recently. Most seem to take a cooler tone: deals are still flowing. Things may be slightly adjusted, they admit. Low-quality businesses aren't getting as much attention. But the world is still spinning. And it doesn't sound as bad as one might assume reading this article.

And some interesting dynamics are playing out that are founder-friendly: cash-rich funds from the later stages are coming down earlier. They're competing for the same deals as smaller funds, but they're much less price sensitive. Matt Turck called this out directly. I've directly seen recent deals where this happened. It's kind of breaking the model for some early-stage funds, who need to get their main ownership chunk early and don't have as much capital to follow-on later. They just can't compete with these check sizes and valuations.

So it's also making it a lot harder for investors right now, too.

That's a great nuance. I was referring to it being in their favor for dealmaking.

Here's another one from 1 month ago: https://www.linkedin.com/posts/darian314_founders-activity-6...

VCs love to tell everyone the market is moving in their favor (lower valuations, fewer deals, etc). It’s like car salespeople telling you this car won’t last long.

Some of it may indeed be true, but it’s in their favor to give founders anxiety. It’s a negotiating tactic, not a public service announcement.

Hey, thanks for the (implied?) question.

We are a lot more concerned with making customers feel like part of the company’s growth than worrying about competitors.

Our customers love seeing the roadmap. And we are lucky to have an engineering team that really does deliver what’s promised. So we like to show that (hence the public release notes that connect to items on the roadmap).

This gives customers confidence that we really will deliver what we say we will, and they can plan their adoption of the product around it.

The community also really loves seeing their own suggestions make it into the roadmap. The way I see it, we’re all building this together.

At the end of the day, it’s not the best product that will always win. It comes down to community.