@JackEllis Awesome reads, love getting a glimpse of the invisible engineering behind the scenes
English
Richard Sands
205 posts
I’ve spent nearly seven years documenting how we built Fathom: outages, DDoS attacks, migrations, $100k AWS savings, my cofounder’s retirement, and a divorce. I write transparently because I want other people to learn from my mistakes. These 18 posts tell the story:
English
@AmanVirk1 They are very important for us if new versions of the framework or library are out and agents models aren't trained on it yet.
English
Are developers still looking at the libraries/framework docs, their approach towards testing, the APIs, or is that era over?
English
@dtmcculloch @dennishegstad Not impossible 😉. You using recharge?
English
@dennishegstad Claude, replace Recharge. Make no mistakes.
English
If you run a business on Shopify as a merchant, what is an app you’d pay for that doesn’t exist?
I have a 3 hour flight home tonight from Boston and I’m going to try and ship something from takeoff to landing
Shill me your ideas
English
@blakeawood @linear Extremely! Has made it much faster to review others work, helping drive throughput
English
@linear Can anyone tell me if they are using this capability and how effective it is
English
@T4SydneyTrains It is wild, this T4 trainline is closed on numerous weekends, constantly! And here we are with more broken stuff, what is going on? No accountability
English
Allow extra travel time due to urgent signal repairs between Edgecliff & Kings Cross
Some services may be diverted via the City Circle or end and start at Martin Place or Central
This will result in in reduced frequency of service between Central and Bondi Junction
English
@karrisaarinen @linear Keeping billing central and visibility into spend. It blew me away today, log ticket, it just picked it up and the guided diff!! Makes reviews a breeze
English
@ric_sands @linear Right now it’s for larger enterprise accounts but we might expand it later more broadly
What’s the main reason for you wanting it?
English
The software factory is humming along in @linear.
An issue comes in from Intercom to Triage. Linear routes and categorizes it, investigates, opens a coding session, and has a fix ready for review 10 minutes later.
We’re now merging 50 to 70 Linear agent generated fixes per week just from Triage.
(Usage is priced with clear token-based credits. No monthly AI fees, no expiring credits.)
English
Once Claude gets how you work the bottleneck is me reviewing it.
@linear agent is also picking up tickets it thinks it can do, how to get my local memory in there?
English
AI Gateways. What is the go to? Using Mastra SDK in @adonisframework
English
imagine if they banned all models and we all went back to writing code by hand and the last 6 months were just a fever dream…imagine
English
@thdxr The creativity and abstract thinking of people working together is going to be hard to replace
English
i think the models are great and do amazing things day to day
and then i go to use them as a brainstorm partner for creative work and they are just horrible, no amount of steering gets them to be even slightly better
makes you think about what these things actually are
English
I built a custom email marketing editor. React Email, Liquid templating, drag drop, merge tags, Shopify connectors, the lot.
This would have been months and months of work before, what a crazy time!
Claude@claudeai
Introducing Claude Fable 5: a Mythos-class model that we’ve made safe for general use. Its capabilities exceed those of any model we’ve ever made generally available.
English
@AmanVirk1 It really does make monorepos a breeze to work with.
English
@Timb03 I find as soon as you feel friction extending something, do a small refactor. For me avoiding the small refactors over time = big bang refactor needed.
English
What happens when tech debt from AI written code becomes too much? Or does it just not matter?
Does it refactor entire repos and databases?
With Pallyy, over the years the debt built up so much that even AI can't untangle it. I'm using AI to help with this of course, but it couldn't do it on it's own.
So what happens when the code that AI has written becomes too messy, and you are not a software engineer - how do you get out of that?
English
English
@jackfriks @postbridge_ @supabase at 88% net margin, $1950/mo in infra for $35K MRR is not the problem to solve. R2 for storage might save a few hundred bucks but Supabase at that revenue is not where I'd optimize first.
English
my solo business @postbridge_ is at $35,000 USD monthly recurring revenue (MRR) = ±$48,000 CAD for my canadian friends and me
here is main breakdown of major costs to run atm - i'm wondering if i should focus on decreasing at all... maybe im overthinking it.
@supabase database and storage = $860/month
@vercel web hosting = $130/month
@triggerdotdev backend processes = $400/month
@unkeydev API management + backend host = $250/month
@X API = $250/month
other smaller recurring costs = $60 /month
. . .
TOTAL INFRA COSTS = ±$1,950/month
profit margin on gross revenue = ±94%
net profit margin after stripe fees = ±88%
my AI bot says dont focus on cutting costs btw, but my frugal self still dreams about no monthly big bills for some reason
jack friks@jackfriks
revenue grows but so does infra costs, new highs all around and my frugal mind still worrying about costs going up 30% when revenue is up 300%... so silly. this is my DB + storage bill, may switch to R2 for storage as 80% of this is pure egress cost on image and video files
English
@gustavonicot @jackfriks @levelsio @postbridge_ @supabase I believe AI reducez that cost to near $0 now, a prompt could do the r2 move
English
@jackfriks @levelsio @postbridge_ @supabase Just a thought:
Think how much energy it takes to save $400/month vs generate another $400/month.
Growth compounds. Extreme optimization often doesn’t.
Identify the bottleneck to your goals and focus there. Bottlenecks determine throughput.
English
@GergelyOrosz People need to understand every ms of latency in trading can cost large traders $$$
English
Coinbase confirms they deliberately had no multi-zone support for their global trading service: meaning they knew if the AZ went down it operated in, all trading would stop.
So all trading stopped for 8 hours yesterday.
They will now (finally?) look into multi-AZ
Overdue, IMO
Brian Armstrong@brian_armstrong
We experienced an outage at Coinbase last night, which is never acceptable. The root cause was a room overheating in an AWS datacenter when multiple chillers failed. We design our services to be redundant to downtime in any one AWS Availability Zone (AZ), and most of our systems worked this way last night, but not all. Our centralized exchange did not. Exchanges have unique architectures that optimize for latency and co-location of clients. It is possible to make exchanges resistant to AZ failures, but this can introduce latency delays that are not desirable along with breaking customer co-location. Given this incident, we'll revisit these tradeoffs to ensure we're giving you the best possible venue to trade. At a minimum, the duration of an outage should be able to be reduced considerably when an AZ move is needed. Thank you to the AWS and Coinbase teams for working through the night to mitigate the issue. We’ll share the detailed technical summary once it's ready.
English
