Hrishi @ Evatype.com

You know things are getting serious when you start using an actual CRM

Just got access to @typesafeai. Testing it for citation verification, search keyword classification, content validation and market inference in @EvatypeAI. Early results are promising. I may be able to replace Sonnet/Luna with Jev. Seems much faster, cheaper and more reliable.

Thanks for rubbing it in, @stripe 🥲

This was a lot of fun!

I shipped the first version of an MCP server for @EvatypeAI a couple of days ago. Being able to talk to a piece of software I made is absolutely magical. If you haven’t added MCP to your product I recommend giving it a go.

I've seen many sites doing all sorts of complex SEO AEO gigabrain content strategies while Cloudflare is blocking their entire site from LLM bots. Make sure to allow bots if you want to show up in AI search.

Notes from the @TallyForms event at @RamenClubHQ last night. @charlierward did a fireside chat with @MarieMartens. After one year, Tally only got to ~$5K ARR. Marie personally DM'd thousands of people over several months asking them to try out her product. With such low numbers in the beginning, what kept them going? They had very modest goals to begin with. They only wanted to make about $20,000 a month (what a really well-paid consultant would make in Belgium). When they hit that milestone, they threw a big party to celebrate because they felt they'd made it. Tally never stopped growing. So that gave them confidence that they had product market fit. Marie said that they filter all product feature requests through the lens of "does this make the core product better?" And if an idea doesn't fit within that then they don't do it. They may integrate with another product that has that feature but they try not to clutter the main product. Having an inbuilt viral mechanic in your product is extremely powerful if the core demand for the product is clear. Freemium can work really well if the free plan helps the product grow. In Tally's case, that's the inbuilt viral mechanic. One striking observation Charlie made was that unlike a lot of other bootstrapped SaaS companies, Tally didn't really do much SEO and grew mostly through their community efforts and product sharing. Once they had a critical mass of users who were also recommending the product to other users, the product started growing on its own and it outpaced their own outreach efforts. SEO has turned out to still benefit them because they are named in many listicles (like on Zapier) and rank for lots of valuable terms. Because Tally did such a good job of building a trusted and loved brand, it's also paying dividends now in AI citations. AI and MCP have unlocked a new lever of growth for them because it gives non-technical people the power to design amazing forms just by talking to their computer. ---- It was a well-organised event and very inspiring for me as a founder! Also, great to meet Marie and Ward from the Tally team.

🎷🥁 🪘🪈 🔥 open.spotify.com/track/4UkTS…

I just want to spend all my time watching all of @JohnMayer’s livestreams and try to play what he plays.

How to stop sounding like ChatGPT's poor cousin on Twitter - Some of you are copy pasta-ing your posts straight from AI chatbots and it shows. Come on people, you’re not even trying. Here are some obvious LLM slop tells to avoid: ------ “Curious how you’re doing X..” NO. Just ask the damn question like a human - “How are you doing X?” ------ “Wild how [huge thing] quietly happened” NO. It didn’t happen quietly. And it wasn’t wild. If it's an actually important thing, state it plainly and it will be obvious to the reader. If it's huge, mention a number that shows its hugeness. ------ “That’s worth sitting with.” No, it’s not. The English language has better phrases than “sitting with” that you can use here. thinking about pondering about mulling over sleeping on You get the idea. ------ “Most people do X. I do Y [I won’t say it but look at me i’m so clever!]” Get outta here. Would you talk to a friend like that? (If you would, this is why we’re not friends) It's ok to point out mistakes people make, but don't try to sound holier-than-thou. Share how you made that mistake too. And how you realised it and fixed it. Get off your high horse. ------ “X was never the hard part.” Yes, it was. You don't need to rewrite history just because LLMs made some things much easier. One example of this that really annoys me (I'm a developer) is “coding was never the hard part”. Oh for sure, it was. If you have never spent hours banging your head against the keyboard trying to solve a bug, then shut up. “X has become much easier now thanks to AI” doesn’t sound as dramatic. It also doesn’t make you sound like a bot. ------ The one I hate the most is not a specific word or prhase, but this weirdly confident staccato 2-3 sentence structure. “X happened. Y happened. Z happened.” It's so robotic! No human talks like this (at least not yet). You’re allowed to use conjunctions like “and”, “or”, “but”, “because“. You can even start a sentence with one (*gasp*)! It’s ok to use filler words and phrases. Write like you talk. ------ Dis-honorable mentions: “framing” “load-bearing” “delve” "landed" ------ What did I miss?

Londonmaxxing at King’s cross station’s 0-indexed platforms

I'm a member of a few different online communities. The ones in Slack/Discord are so much more alive than forum-style ones in Circle. Chat apps can feel overwhelming at times but forums just feel dead.

I wanted a quick way to burn captions into demo videos on my mac, so I built a script that transcribes (using whisper.cpp) and burns the captions in. Also created a Claude skill for it. Just open sourced it - github.com/hrishimittal/vide…

Does product matter any more or is it all about marketing and distribution?

Fast and Slow - A marketing tip I picked up from @robwalling (sorry about the 500 ummms.. first time posting a selfie video!)