Horatio Thomas (@horatioecthomas) header
Horatio Thomas (@horatioecthomas) profile picture

AI Product Engineer, ex-Tinder Android. Building Dream Interpreter AI (millions of dreams interpreted), Shangri Lingua and Moolot.

Bangkok · กรุงเทพ
Joined August 2021
The MCP lightbulb finally turned on for me and now I'm rethinking all of my products and am seeing so many new opportunities
2
231
I've been daily driving Claude Code the past week or so but now I'm setting up Grokbot to help grow my X because it's the only harness with a first party X connector
2
2
155
MCP is so sick. I connected the gmail MCP to claude, asked it to check my emails. It saw an invoice from my accountant. I connected the Mercury MCP and it made the payment. I only needed to leave claude to approve the payment from the browser from the link it opened.
3
2
248
A big problem with MCP apps is that their moat is extremely narrow. You only need to tell your agent to migrate your data from one to the other, so the switching cost is effectively zero
1
2
150
Just submitted Dream Interpreter AI's MCP to the Claude directory. Soon you will be able to explore dreams from around the world right from Claude.
1
1
193
There is something so satisfying about describing a cohesive problem and idea to claude and then seeing it start four sub-sessions in parallel to work on it
2
238
I've been pretty confident in the performance of AI coding, but I'd be really scared if agents were coding in C something could go devastatingly wrong. It's nice to have the guardrails of a programming language like Rust at the bare minimum.
Real men manage their own memory
1
2
201
Early results from this have nearly doubled my metrics. Forget how to code.
I'm not sure if I'm experiencing some degree of AI psychosis or if Claude has really just gotten this good. I finished a rewrite of dreaminterpreter.ai from Next.js to Ktor, implemented an entire redesign of the website, and even made some progress with my language study app. I tried to let AI take the wheel before and just ended up shooting myself in the foot with bug after bug, but now it's fixing bugs that I'm writing. There's a long list of ideas I've had, and now I can execute so fast I don't want to stop. I've been going for 14 hours today, so it's finally time to take a break and get some sleep.
1
131
What I love about living in Thailand is I can grind pretty hard and then get instant cortisol dumps from just being in this atmosphere
2
245
I'm not sure if I'm experiencing some degree of AI psychosis or if Claude has really just gotten this good. I finished a rewrite of dreaminterpreter.ai from Next.js to Ktor, implemented an entire redesign of the website, and even made some progress with my language study app. I tried to let AI take the wheel before and just ended up shooting myself in the foot with bug after bug, but now it's fixing bugs that I'm writing. There's a long list of ideas I've had, and now I can execute so fast I don't want to stop. I've been going for 14 hours today, so it's finally time to take a break and get some sleep.
1
1
2
396
Claude couldn't figure out how to do something, so I Googled how to get Claude to do it. We've really come full circle. Do we have Stack Overflow for agents yet?
1
2
249
I wasted so much time over the past year or so trying to write "clean code" with AI
We need to reconsider our priors on code readability. Humans, compared to agents, are pretty slow at understanding large quantities of text. So in the old days, readability was important so the human could efficiently grok what was going on. So readability now comes down to: 1. Can a human, if needed, jump in and understand the code 2. Does the code use the same domain language as the human describing the code? 1 helps with observability and the occasional spot checks you need to do on important PR's. 2 helps the agent translate the words you use into code changes. In other words, readability is now a nicety, not a necessity. The way agents 'read' code is different - good agents are all about gathering data about how the code works. They'll prod around with it, execute parts of it. So 'readability' to agents really means 'verifiability'. How easy it is to mess about with that code to see it working. Let's take this scenario: - You have a huge module in your codebase that uses all sorts of complex performance tricks - It has 100% test coverage A talented human would need an hour or two to get comfortable working in that module. An agent needs zero time. They verify it with the tests, poke about a bit, and they're good to go. The takeaways: - When judging a module (or a codebase), don't think about how easy the code is to read - Instead, think about how easy it is to verify - Design your codebase so modules can be tested and verified independently - Feel free to sacrifice a bit of readability for performance - Cyclomatic complexity and function size don't matter to agents
156
I 1000% agree with this. You get further customization, are not reliant on a third party and don't have to deal with the bloat and headache of managing external dependencies. In the past, many third-party dependencies could be massive time savers, now they are just liabilities.
What’s going to happen to open source? A prediction: Dependencies are over. The same Brotli vulnerability impacted .NET, Node, Bun, Envoy and more. Dependencies are too big of a target. When rewriting a dependency is one prompt away, why take the risk?
179
Vibecoded a little something to help me practice writing and reading Thai apps.apple.com/us/app/shangr…
4
530
Pretty cute that IBKR has a survey on their app to ask for ideas to improve experience on iPhone Duo
267
It's going to be a bit of an uncomfortable truth that "AI Slop" is actually producing better quality work than pre-AI solutions. This will only become increasingly more so as AI continues to advance.
1
206
The 2020s is really becoming the decade of the scalper. Toilet paper, PS5s, Pokemon, AI compute etc
Heard from a neolab friend: “Even if we fail to train the best model, we locked in GPU capacity early. H100/B200/B300 rental prices have since gone up 2–3x. We can just resell the compute and print money.” OK, so that’s the business model.
1
183
Support iPad one time and you're locked in for life
1
2
424
Still waiting for my new app to get reviewed by Apple. Removed some dead ends from my pages by adding Dream symbol links to users' shared dreams. Hopefully this helps with SEO.
1
118
If you are buying the iPhone Duo to review code on the go, note that the GitHub app doesn’t support it yet. Having the larger screen is nice in safari for viewing diffs tho
1
599
The cursor app looks great tho
59