Alex Duggleby

Alex Duggleby

@alexduggleby.com

Founder and Maker or whatever we're called nowadays 🇦🇹. What am I up to? ✍️ https://alexduggleby.com/now 🤘 https://github.com/aduggleby 🏕️ My tribes: Green Tea Enthusiasts, Bootstrapped Entrepreneurs, .NET/React/Azure/Microsoft 365 developers and Lego

244 followers725 following205 posts

Alex Duggleby reposted

After more than a year of labor, my book 𝘏𝘪𝘥𝘥𝘦𝘯 𝘔𝘶𝘭𝘵𝘪𝘱𝘭𝘪𝘦𝘳𝘴 has launched! It's all my best ideas -- and best writing -- on how to grow your company with your existing budget and team. And yes, I read the audiobook myself. :-) (1/5)

Yay my little travel app got approved - I had issues remembering where I went when I was journaling about the day. Turns out my camera roll knew quite well so built an app around it with my agents. Pirate explorer themed obviously…

Also an interesting experiment in doing everything on device. All the image analysis done by Apple’s AI or built in APIs, nothing sent anywhere. I pushed it quite far to get the features I wanted and was pleasantly surprised what Apple offers nowadays. apps.apple.com/at/app/snapp...

apps.apple.comSnapperMapper: Photo Trip Maps App - App StoreDownload SnapperMapper: Photo Trip Maps by Alex Duggleby on the App Store. See screenshots, ratings and reviews, user tips and more games like SnapperMapper:…

Alex Duggleby reposted

New Podcast! Since I’ll be doing more episodes of Smart Bear Live, like the one just done with the great @dmytrokrasun.com, I’ve created a podcast as well as the YouTube channel. It should be everywhere: Spotify, Apple Podcasts, etc.. Weee!

How can PaymentWorks who handles where a ton of money flows only accept SMS as a MFA? Even banks have realized that’s not secure enough. Clients choose us because we are the secure version of what they want for their customer data so why doesn‘t procurement care about security as much?

Pushed back a bit and got this response: “PaymentWorks is run by and through JP Morgan Bank” (nope, they just have a partnership) “and X is confident in choosing this platform for payment.” 🤦‍♂️ confidence in a bank that had the greatest data breach in 2014 caused by missing 2FA… makes sense.

I‘m on a nice cycle streak everyday through the heatwave so far. The secret is leaving at 6:30 and which saves me from the worst heat although I get in today at 11 where it’s already 36 C/97 F. But today is peak day. Been able to watch these grow amazingly fast - feels like two weeks to this size.

Bought a wireless scanner Brother DS-940DW so I can use it with the phone to quickly scan something. Turns out the scanner can only be charged via computer USB port. Charging brick isn’t enough. Needs that data line to charge. Would love to chat with the engineer that signed off on that decision.

I switch between Linux, Mac and Windows daily multiple times for work and spend way to much time trying to remember to use the right CTRL, SUPER or ALT key for Copy/Paste (among other shortcuts). Any folks doing that with some tips on how to harmonize this or do you just power through?

Biggest annoyances have indeed been on Mac so far - POS1 & END behave weird, and don't get me started on backspace for delete!? (And yes I had to replace ALT + Q - which is @ on QWERTZ but closes the current window on macOS with the invert screen color shortcut 😆)

How a customer reacts to "Your account was locked due to a chargeback" greatly influences what hoops I make them run through. If they apologize/explain, I unlock, ask them to create a new sub. Just costs me the chargeback fee. If they don't, teaching them a lesson is priceless.

I will ask them to call the bank to reverse the chargeback, which often doesn't do anything, but usually gets accounting involved and they make a note not to chargeback my company. End result is I unlock them and usually no more chargebacks. ( 2 exceptions in almost a decade AFAICR.)

A future where your agents are negotiating with a suppliers agents all without human interaction sounds like nightmare until you try getting prices for corporate insurance policies (for a 1 person company mind you) and realize nothing can be done online and you're going to waste hours on the phone.

Personally I don't buy from anywhere that doesn't have prices published unless I'm forced too. So I have clear pricing for all of my products - with an optional call if you fall outside the norm. Seems like most experienced founders say it's a mistake - for me, it's a deliberate choice.

Today morning started with: rise and shine, ready for a full day of demo video recording. Today evening ended with: damn you Safari and your little quirks that I have to make exceptions for. So tomorrow: Groundhog Day again! (Well actually cycling first… but then recording!)

I'm torn on submitting an LLM created GH issue against an O/S project. On the one and I understand they're getting swamped by automated issues, but at the same time the issue is real enough that I got hit 3 times yesterday alone so it deserves whatever info I can provide event if it's from an LLM 🤷‍♂️

Previously I forked, let the LLM patch it and build the binary from scratch. But because of maintainer feedback didn't ever create a PR. So is submitting the issue the better approach? I went over it (and made sure the guidelines are followed as best I can) but in the end it feels not 100% right.

It created a way more detailed PR than I ever would, with the code lines, the core dump info I had etc and if I got one for my O/S projects I would use it to ask my own LLM if the issue is valid, but it's a weird conflict in my head. Maybe it's because in the end it only took 5min of my time ...

Poor Hippos! I mean Puzzles, sure I get it, they attack, but what's a Hippo ever done... In more serious news: buy the book, mine is on it's way. Deciding what to build in the age where features are just a prompt away needs a good framework.

Steve McLeod@steveofmcleod.com

It's newsletter time! This week's newsletter is the big announcement: after 1 year and 9 months, Kill the HiPPO is now published! Kill the HiPPO is my book on how bootstrapped software companies decide what feature to build next. killthehippo.com/posts/kill-t...

I've been challenged by the SO to provide a local LLM (ultimately with a Cowork like harness) for use on our private docs etc and was wondering what hardware/model combos are actually using at the moment? Will a top Mac Studio do or no way around 4090 for smth slower but close to frontier perf?

Kimi K2.6 looks promising but unless I'm mistaken quick research says ~20T/s needs either 4 top of the line Mac Studios with Exo or even PC with 4x 4090. If that's what it takes, ok but looking for some ppl's experience. (It's off the books until mem prices drop again, but at least it's a target.)

The „agentic engineering“ founders finding out they were hacked don’t realize they are the lucky ones. It’s the ones being exploited in secret and continuing to do so for long time that are the ones I’d worry about. In an age where data is worth so much, finding a leak = gold.

Today in Hamburg: Kebap shop wanted 5% extra charge because I was using a business credit card (N26) to pay. Different card was fine. I mean I don’t like the surcharges for international/business cards either but it’s the first time I’ve seen it passed on to the customer.

Also I’m fascinated how they explain this to a normal consumer - I mean I happen to know there’s a difference but the average person off the street? Also 5% is steep. I don’t think it’s that high anywhere - must be compounding through the POS provider.

Zero Human Businesses… isn’t that just arbitrage while the masses catch up to using AI? Maybe you can make some money in the meantime but won’t every AI builder start trying to build one of those and quickly saturate the market? Won’t personally bet on that one but it’s wild what’s being created!

Almost a decade into a SaaS 99% of the "bugs" reported are documentation issues, user errors etc... But it's the 1% actual bugs that always leave me asking myself: how on earth has no one encountered that case in all this time!? Especially when it doesn't feel like an edge case.

My new end of day routine is to ask Codex to suggest improvements to the test suite, refactoring for maintainability/reusability and reviewing/simplifying code comments. With the test suite I haven't had a refactoring go wrong (so far) and the code comments are also vital (past decisions/fixes).

But there are still things I do out of habit from 6 months ago that thankfully aren't as necessary any more - but habits die hard, even if they were learnt so quickly in relative terms.

I've added a special "Copy LLM prompt" to my error messages (when I run in dev mode) which not only has a basic "replicate this in e2e then fix it" prompt but also appends a bunch of technical data, breadcrumbs for my last 10 steps etc. Quite a productivity boost...

Stripe's Managed Payments just went public 🥳, but am I doing the calculation right that as a EU company (with EUR account) charging a customer in USD is going to cost me 8.75% + 25c? 3.25% for the international card + 2% currency conv. to EUR + 3.5% for managed payments Where's my mistake?

I guess I can settle in USD using Wise/Revolut (does Stripe support two settlement currencies though?) and bring it down to 6.75% but I remember everyone going bonkers about Paddle's 5%, so I am wondering if I'm just plain wrong here somewhere. (and isn't subscription billing on top?)

Yay for any European alternative, but I don't understand is why we don't build on what's already there. SEPA Instant is free AFAIK and supported across Europe (and chargebacks are left to the courts not a card processor). But let's see: europeanbusinessmagazine.com/business/eur...

europeanbusinessmagazine.comEurope's $24 Trillion Breakup With Visa and MastercardA 130-million-user payment system backed by 16 major banks just launched to challenge Visa/mastercard

I’m spending about as much time getting in Claude Code fixing Openclaw as I am using it. Running it on a VPS is definitely less out-of-the-box awesome than a Mac mini. Using oxylabs proxy definitely unblocked a lot of things. See my fork if you’re interested: github.com/aduggleby/op...

github.comGitHub - aduggleby/openclaw: Your own personal AI assistant. Any OS. Any Platform. The lobster way. 🦞Your own personal AI assistant. Any OS. Any Platform. The lobster way. 🦞 - GitHub - aduggleby/openclaw: Your own personal AI assistant. Any OS. Any Platform. The lobster way. 🦞

Codex 5.3 is "feeling" and definitely more Zen than me 😆 "There's no rush, so maybe waiting a moment isn't a big deal. I’ll proceed to pull the image on the server now using Docker Compose. It feels good to have a clear plan to follow"

Will people in the future look back at the introduction of LLMs and think "probably alien tech" just like we do when we look back at the pyramids because we don't understand how it was possible?

I am so grateful my biz lets me do crazy things like go from a completed foggy Vienna four hours by train to the west of the country to ride my bike for a few hours and then head back for another four. Totally worth it - especially with SAD. And even got support tickets done on the train!

Recently got back into stamp collecting and I’ve realized it’s one of the few areas where it’s quite easy to get your hand on items from a long gone era. This one’s about 150y old - a very different world for the original owner. But I’ve still yet to cross the 2 century mark though…

Good old fashioned in-flight shenanigans: www.saxrag.com/tech/reversi... Often overlooked: "SNI leaks all the domains they visit". HTTPS often doesn't stop a MITM/state from knowing you visited beardgroomers.com + Hyperscalers mostly use SNI for SSL. But improvement to SNI are being worked on.

saxrag.comUnlocking free WiFi on British AirwaysI was recently flying between HKG & LHR via British Airways. I’d done the same flight back in 2023, and remember relying on the in-flight entertainment for the 14 hour journey. However, this time on m...