Pinned Tweet
Somewhere in your docs there’s an article telling customers to click a button you removed six months ago. You’ll never find it by reading all 200 articles. So you don’t. Starting from today, we'll find and fix it for you. 🛡️ Introducing Verification by Ferndesk
8
5
111
6,262
Not enough people know that Opus 5.5 + Codex Computer Use >>>> GPT 6 Astra + Codex computer use. Roughly as good + faster AND cheaper. Used it + @compai MCP to gather all the evidence I needed for SOC II in under a day.
14
5
61
5,732
Connected via the @executorsh MCP!
1
1
455
Hobbies: guitar, chess, stealing customers from billion-dollar companies 😤 The latest is Duplika. They just moved their help center from Zendesk Guide to Ferndesk. "It feels like upgrading from Internet of the 90s" They'd been paying for add-ons just to get a table of contents. A TABLE OF CONTENTS!!! Now their docs stay up to date automatically, their support volume is going down, and they finally have docs that they're proud to share. Honestly, the giants make this hobby too easy. At this rate I'll have to give up chess...
14
1
49
1,879
Someone paid you this month, and you have no idea why. You saw the notification, celebrated, and got back to work. How they found you, what they used before upgrading and where they nearly gave up are all sitting in your analytics and your database. You haven't looked, because getting it out would take hours you don't have. It doesn't take hours anymore... We recently changed our pricing at Ferndesk. When the first customer paid the new price, I wanted to know exactly who they were, so I could go find more of them. So I asked Claude to go through Mixpanel events and session recordings and tell me their story, from first visit to payment. → What content helped them make a decision? → Where did they get stuck? → What could we have done better? It took less than 30 minutes to get a report that would have cost $100s a few years ago. One customer doesn't prove a price works. But their story tells me a lot more than a payment notification ever could. So pick your most recent paying customer. Give your agent the analytics, and ask for their story.
17
2
62
3,245
IT'S 2X CHEAPPPEEERRRRRR!!!!!
Please welcome GPT-6 Sol and GPT-6 Luna to the GPT-6 universe. GPT-6 Sol and Luna build on the advances behind GPT-6 Astra, bringing much of its strengths into faster and more affordable models to support work at scale. We’ve also made caching and inference more efficient, and we’re passing the savings directly to you: 50% lower API prices for Sol and Luna compared with GPT‑5.6 promotional pricing.
4
38
5,059
Bootstrapped and under $1M ARR? Ferndesk is half price for you now. Pro plan, $75/mo for your first year, if you're bootstrapped or pre-seed, under $1M ARR, and fewer than 10 people. Everything included. Verification, screenshots, the AI assistant, all of it. Same product, half the price. I'm a bootstrapper. I've never raised a dollar for Ferndesk. I know what it's like to pay for tools out of revenue you don't have yet, and I know how many "startup programs" want a pitch deck and an accelerator logo before they'll talk to you. Ours takes a minute to apply and nobody reviews your deck. Built by a bootstrapper, for bootstrappers 🌿
12
3
77
3,462
Wilson Wilson retweeted
Here in spirit @euboid
6
1
35
3,015
MCPs would be 1000x better if they didn't disconnect CONSTANTLY. ATP considering moving back to just APIs for everything.
13
1
40
4,602
Our smallest blog posts bring in 4x the customers of our biggest. I only know because I went back six months at Ferndesk and counted paying customers per post instead of readers. The big posts are the ones I was proud of. They rank well and our traffic charts love them, but almost NOBODY reading them ever buys. The small posts were each written for someone with a real pain who's already looking for a way out. Far fewer people read them, BUT far more of them become customers. We'd been chasing clicks when we should have been chasing buyers 😅 300 of the right visitors a day beats 3,000 of anybody. The traffic graph may not look as good. I'm betting the revenue graph will.
17
7
85
3,838
Most founders get to a conference by buying a ticket. I had to take a drug test. I also needed dozens of documents, guarantor paperwork from multiple people, and hours sitting in random offices. It was a full week of work, for a visa that can still get refused. So why bother? Because building as a solo founder is SO lonely. When something big happens at Ferndesk, the first one to hear about it is usually an AI. Almost everyone I know building a business like mine lives inside my computer. I’d like to spend a few days in a room with people who chose this life too. Hopefully the visa comes through. If it doesn’t, at least I’m officially drug-free 😆
31
2
107
6,925
When Youform launched, I thought: do we really need another form builder? Then they hit $18K MRR. I’ve known the team since they started. There were already hundreds of alternatives, and I genuinely couldn’t see how they were going to grow. But they kept going. They listened to their customers, kept improving the product, and grew it to that milestone by March. As someone building in another extremely crowded market (Documentation software), this is SUPER inspiring. It’s our bootstrapped company against huge VC-backed teams. Of course, when they said they wanted to use Ferndesk, I jumped at the opportunity. I love the company, and I think they’re going to win. Now Ferndesk saves them around five hours a week on docs. Super proud to be part of their journey 🫡
24
4
127
6,813
Wilson Wilson retweeted
We’ve raised $34M, led by Roo Capital and Grand Ventures. 18 months ago we started Comp AI with one idea: agents will be responsible for compliance busy-work. Today, 1000+ companies like @UseCorgi, @opencode and @dubdotco trust us to get them audit-ready.
218
48
637
574,473
The era of the budget SaaS tool is over. Ferndesk launched at $39/mo. I thought a low price would get me customers fast, and customers would get me feedback. Instead it got me a product I had to keep watering down. The best thing we could build, an AI that verifies every article against your codebase, doesn't work mathematically for a $39 customer. So I couldn't ship it. I added limits everywhere and nickel-and-dimed on features just to keep the margins alive. And the $39 customers churned anyway. The pain wasn't strong enough. Half of them vibe coded their own version and left. We start at $149 now. The customers who pay more stick around longer, are easier to serve, and are actually serious about their implementation. They’d much rather have someone manage their implementation VS building it themselves, even with AI. So it honestly looks like the era of budget SaaS is dying. Be the serious option or don't exist.
57
14
295
15,661
I had both Fable 5.1 and GPT 6 astra create me an animated, pixel-art character + animation for the @ferndeskhq mascot, Fern. Fable 5.1 Left / GPT 6 Astra Right details below ⬇️
5
4
33
4,497
This isn't exactly an apples-to-apples comparison: - Harness: Claude Design VS Chatgpt work - Both ran in the cloud on high effort. - Output format for both was HTML - Both with zero references other than our mascot img The Astra version was so bad that I gave up on it completely, even after 5 iterations. Even more annoying - Astra used gpt imagegen for style references but still created a malformed design that didn't match the requirements. Fable went in completely blind and created something usable on the first go. I iterated at least 3x more with Claude, not to fix motion/composition - but to improve on the character's design. Genuinely not sure if I'm doing something wrong, but after all the Astra hype this was extremely disappointing lol. At design tasks, Anthropic models have always cleared.
4
1
7
788
I've been bootstrapping for 5+ years and I've never been good at planning. I've run (and barely stuck with) sprints in the past. At Ferndesk I mostly ran on vibes and checked the data whenever I felt like it, which was usually after something was already going wrong. That changed a few weeks ago. Now an AI agent does my weekly planning with me every Sunday evening. → It pulls Stripe, Mixpanel, my inbox, support tickets and our core metrics. It even pulls my whoop sleep + fitness data to gauge my energy levels. → Fills in a scorecard for the week → Shows up with ideas on what I could do next → Then it interviews me. What went well, what drained me, what I avoided → We pick ONE big bet for the week and focus solely on that. Three Sundays ago, it told me our trial-to-paid rate wasn't that great and that nothing else I was excited about mattered until that started trending upward. So I dropped everything and spent three weeks on it. Now it's trending upward. On vibes, I'd have kept shipping features and wondered why the numbers weren't adding up. The data has always been there, but it was tedious synthesizing and analyzing it every week. No longer! This is why you need to be contextmaxxing.
16
2
94
4,744
I sell self-updating documentation. Docs that never go stale. So imagine my face when I ran Ferndesk's own help center through Verification and 63 of 131 articles came back with issues. It wasn't the stuff I'd have guessed. An integration partner changed a URL on their end. I changed a policy, told customers about it in support tickets, and never updated the article. I renamed a field in Settings and the docs still had the old name. Screenshots from a version of the dashboard before our big redesign. I'd have never caught any of that by reading through 131 articles. So I didn't. What Ferndesk does instead is check every article against our codebase, our live app, our tickets, and our integrations. It drafts the fix, retakes the screenshots, and I approve or dismiss. 63 articles, fixed before lunch!!
7
1
59
3,528
Shipped 🚢 @ferndeskhq now has native support for API references! → REST, @GraphQL and Stainless support → Fast, bespoke, beautiful API playground → Fully SSR'd + indexable by AI and search engines Rebuilt from the ground up 🔥
6
3
51
2,870