Danny Nemer

🤖 ← Cursor for Gmail
🚆 $30K MRR ($360K ARR) +8% MoM
PREV: 🍀 | 🚚 @GreenChainAI | 🟰 | 💮

[Danny Nemer @DannyNemer]

AutoText's typing suggestions are now 2x faster!

We're at 800ms RTT per suggestion, still slower than I'd like. For reference, Cursor's gold-standard suggestions land around 250ms.

Inference dominates the latency budget, eating 93%+ of the round trip. After benchmarking dozens of models across every major serverless provider, I've hit a ceiling.

One lever remains: dedicated deployments running open-source models (Kimi K2, Qwen3, GLM, MiniMax, Gemma 4) on hand-picked hardware. That unlocks the accelerators @FireworksAI_HQ and @togethercompute reserve for dedicated tenants, namely speculative decoding, predicted outputs, and their fastest chips. The catch: always-on capacity (no scale-to-zero) starts at $7k/month.

[Danny Nemer @DannyNemer]

@eglyman @zebulgar Another Conductor customer rising up. We call this the Conductor Bump.

[Danny Nemer @DannyNemer]

@mikejulian Switch off of Claude Teams to separate Claude Max plans to save potentially hundreds of thousands of dollars per month. The token limits are 20x higher. It’s the overage usage-based tokens that are 20x more expensive.

[Danny Nemer @DannyNemer]

@Techmeme @markgurman Google Cast support in iOS might mean Apple is retiring the Apple TV (hardware), last updated 3.5 years ago.

Apple officially renamed the streaming service from “Apple TV+” to simple “Apple TV” last October

[Danny Nemer @DannyNemer]

@quinnslcm How about just not having greasy ass fingers?

[Danny Nemer @DannyNemer]

Your Eng team could save $100k+ per month on your OpenAI (non-API) bills:

If your team in using ChatGPT Business plan ($25/user/mo) and a single engineer consumes the same amount of tokens that you get with the $200 Pro consumer plan, you'd be billed $123,000 that month instead of $200.

You'd forgo the Business/Team features (centralized billing, admin controls, etc.), but you'd save so much.

It blows my mind hearing about companies' OpenAI/Anthropic bills for internal usage when the consumer plans are in some cases 20-30x more generous.

[Danny Nemer @DannyNemer]

@AnthropicAI @StainlessAPI Haha, now OpenAI and everyone have to scramble to completely rebuild their SDKs that their entire dev community uses, or migrate to @speakeasydev -- however, OpenAI's SDKs had so much custom code that is not a simple migration of its OpenAPI schema.

Conductor must do this, too!

[Danny Nemer @DannyNemer]

@namcios Cursor has 350+ employees. Do your research.

[Danny Nemer @DannyNemer]

@MTSlive It has done this multiple times before. This is just the latest update.

[Danny Nemer @DannyNemer]

@amir The graph is already outdated by 2x

[Danny Nemer @DannyNemer]

That’s me! 👋 thank you, Fraser!

Another SEO growth hack the eventually worked: creating a viral Reddit post that featured the URL. Tripled sign-ups overnight.

For years, nothing would improve our SEO. Even after building a fast landing page, a docs site with 200+ pages/articles, and a GitHub repo with 300+ stars, we were stuck around #20-60 for the queries that mattered most.

Then, a couple of Reddit posts about the startup did well, and Conductor immediately moved to around #5 on Google for relevant queries. Thereafter, sign-ups grew 3x in a month.

I can’t prove the exact mechanism, but Reddit seems to have helped in a way that X never did for me, even when tweets mentioning the URL got 100K+ impressions.

Google seems to put real weight on Reddit, and LLMs do too.

[Fraser @iamfra5er]

THIS GUY BUILT A $28,033/MO SAAS WITH $0 STARTUP COST BY SOLVING THE MOST BORING PROBLEM IN SOFTWARE NOBODY WANTED TO TOUCH

he was a founding engineer at an AI startup and noticed something nobody was talking about

the AI features were easy. the data integrations were killing them.

weird CSV files. SOAP XML servers. ancient accounting software. the unglamorous plumbing that every vertical SaaS company desperately needs but nobody wants to build

so he built Conductor(dot)is to solve it

first attempt failed. built a generic data orchestrator with no users and burned out

second attempt: a friend connected him with a startup that needed a QuickBooks Desktop integration. he almost said no because it felt too niche

he said yes anyway because it guaranteed one real paying customer

and building for one real user beat building into the abyss every single time

to get his next 9 customers he asked friends to star his GitHub repo so it would show up when people searched for QuickBooks Desktop integrations

that's the whole growth hack. GitHub stars from friends.

every single customer since has been inbound. $0 CAC. 90% net margins.

the thing that made customers love him early? he watched their API logs and reached out proactively when he saw them struggling with setup

"the best support I've ever had in my life" from a customer who never even asked for help

now Conductor runs mostly on autopilot while he spends 90% of his time on his next project

the playbook: find the problem everyone needs solved but nobody wants to work on. be the only one willing to do it. own the niche forever.

[Danny Nemer @DannyNemer]

You can get AutoText here: https://t.co/tMbMnhz5D9

[Danny Nemer @DannyNemer]

I will keep saying it - using AutoText with Gmail is just amazing.

I genuinely love it. Crazy to me how more people are not trying it.

[Danny Nemer @DannyNemer]

@Techmeme OMG

But how would such a relatively small company have news like this “leak” — my guess, to get a counter offer from OpenAI

[Danny Nemer @DannyNemer]

@vasuman Why are they using API key pricing!? Switch to fixed plans (with overage); it’s literally orders of magnitude cheaper.

[Danny Nemer @DannyNemer]

@claudeai Claude for Outlook still pails in comparison to AutoText! Sure, asking questions about your email is nice, but actually having powerful AI autocomplete while you type — like coding — is just another level.

https://t.co/sv9peZv9k6

[Danny Nemer @DannyNemer]

@verge This is as accurate as the rumor that OpenAI was acquiring Pinterest: not happening.

[Danny Nemer @DannyNemer]

Why the hell does Greg have to inform Musk of his position in Cerebras? Musk has nothing to do with Cerebras. It's completely unrelated.

Separately, it is not self-dealing. Cerebras is phenomenal; one of two viable accelerators for inference leveraging specialized silicon (the other is Groq). Of course it makes sense to invest if given the personal opportunity. And of course it makes sense for every frontier lab to want to acquire them.

So Greg and Sam can never invest in companies in which they have expertise due to their job?

[Danny Nemer @DannyNemer]

API key Codex is usage is AT LEAST 100x more expensive than the fixed rate Pro plan.

I hit $400 in Codex in API key usage in a single one-hour session.

Typically, the $200 Pro plan covers me for the whole month, running multiple sessions in parallel 12+ continuous hours every single day.

I assume the same is true for Anthropic. Is this how Anthropic hit $30B in revenue so quickly?

Are all of the organizations/enterprises paying API key prices instead of the fixed plan rates?

[Danny Nemer @DannyNemer]

There are no legitimate text-to-CAD startups. Claude and Codex can write directly to Step and VHL files or use Python to interface directly. No need for a separate wrapping company.

It’s kind of like saying text-to-SQL startups when all frontier models can write SQL that runs in existing SQL environments.