Customer Support AI Explorer (@CSAIExplorer) header
Customer Support AI Explorer (@CSAIExplorer) profile picture

Customer Support Manager & Python Dev. Exploring AI's role in support services, sharing insights on automation and AI-driven solutions. Building @ephor

Joined January 2024
Photo by Customer Support AI Explorer (@CSAIExplorer)Photo by Customer Support AI Explorer (@CSAIExplorer)Photo by Customer Support AI Explorer (@CSAIExplorer)Photo by Customer Support AI Explorer (@CSAIExplorer)Photo by Customer Support AI Explorer (@CSAIExplorer)Photo by Customer Support AI Explorer (@CSAIExplorer)Photo by Customer Support AI Explorer (@CSAIExplorer)Photo by Customer Support AI Explorer (@CSAIExplorer)Photo by Customer Support AI Explorer (@CSAIExplorer)Photo by Customer Support AI Explorer (@CSAIExplorer)
...not even 10 days ago both companies had been very public about how necessary it was to slow down the pace and lauding each other over the idea, only to announce their next frontier models a week later within 65 minutes from one another 🤔 Gotta be one of those SOC checkboxes ("did you publicly disclose your concerns before shipping anyway?")
1
60
Customer Support AI Explorer retweeted
I put a prompt injection into my LinkedIn bio and recruiters are messaging me in Old English and calling me Lord.
634
7,204
538
90,453
4,390,415
Customer Support AI Explorer retweeted
We built our launch video in Claude Code using HyperFrames. Now it's yours. Open source, agent-native framework. HTML to MP4. $ npx skills add heygen-com/hyperframes RT + Comment "HyperFrames" to get the full source code of this launch video (must follow)
2,835
1,785
569
8,314
3,020,025
Customer Support AI Explorer retweeted
This game was 100% designed, tested, and made by Claude Code with the instructions to "make a complete Sierra-style adventure game with EGA-like graphics and text parser, with 10-15 minutes of gameplay." I then told it to playtest the game & deploy. Play: enchanted-lighthouse-game.ne…
74
93
15
1,241
113,721
Done being overwhelmed by notifications, so I'm trying Spec - you should too 👀 Use my link to get on the waitlist: withspec.com?ref_id=K0JXVU8U…
46
The numbers were surprising: Pure LLM accuracy: 13/18 Best persona: 13/18 Worst persona: 8/18 But accuracy missed the point. The vanilla LLM told me WHAT went wrong. The personas explained WHY through frameworks the base model never touched.
1
36
Example: Donald Miller's "StoryBrand" persona didn't just say "customers were upset." It reframed New Coke's failure as "violating the customer's narrative arc." Same data. Completely different insight. One I can actually use in my next product launch.
1
1
28
The big discovery: Personas unlock reasoning pathways that exist in the model but aren't naturally accessed. It's like having the same brain think through different lenses - suddenly connections appear that were always possible but never surfaced. Full analysis: medium.com/@pablo.ambram/whe…
1
26
Just ran an experiment that changed how I think about LLMs. Same model, same business case (New Coke), three different approaches: Raw LLM ❌ LLM with expert personas ✅ Multi-persona debate 🤔 The results? Personas don't make LLMs smarter. They make them think differently. 🧵👇
1
1
29
Is it just me or was Sonnet 4 lobotomized at some point since yesterday? I had to switch to Opus 4.1 for coding because Sonnet was fumbling so badly, and is being generally unhelpful, breaking good code with, allegedly, pretty dumb mistakes...
48
Customer Support AI Explorer retweeted
You can now deploy AI voice agents to LiveKit Cloud. We handle: • Stateful load balancing • Capacity management • Draining and instant rollbacks • Operational observability
19
29
16
240
68,139
What I like about GPT5 with my limited exposure so far using it in Windsurf: * it follows instructions very precisely. And I mean it is PRE-CI-SE... (even the second sentence of the tenth bullet on a cursor rule somewhere). The most precise rule follower I've seen so far. Not sure if this is good or bad, but at least it's different. * You don't need to heat up the context with every single little detail for it to not mess up. No hand holding, just tell it your end goal, seems to be able to find the way pretty nicely. It's like this one is truly "agentic" at heart (for whatever that means), at least the model that Windsurf calls the "high-reasoning" model
1
74
What I don't like about GPT5 so far for coding: * Conflated some tooling error with my reported task * Conflated output from the task at hand and placed it on the output required from a template (i.e. the template asked for a #summary of something after analysis, and instead it provided a summary of what my instructions had been. Strange!) * May be too verbose oriented for my taste for the purposes of coding (not the output code itself necessarily, but the whole thinking and reasoning process around it). Perhaps, with reason, it's because I'm testing the high reasoning model...
45
Customer Support AI Explorer retweeted
supervision has gained almost 1k stars since the release on wednesday; so cool link: github.com/roboflow/supervis…
19
361
16
3,505
223,921
Customer Support AI Explorer retweeted
announcing: kisuke, a native ios ide kisuke is your fav pocket engineer. you get: - claude code (use your anthropic acc + all CC features) - multi tab terminal - code editor - browser + built-in devtools - port forwarding (auto-detected and available inapp)
507
506
137
10,001
1,180,000
Customer Support AI Explorer retweeted
Introducing Kiro, an all-new agentic IDE that has a chance to transform how developers build software. Let me highlight three key innovations that make Kiro special: 1 - Kiro introduces spec-driven development, helping developers express their intent clearly through natural language specifications and architecture diagrams for complex features. This comprehensive context helps Kiro’s AI agents deliver better results with fewer iterations. 2 - Kiro features intelligent agent hooks that automatically handle critical but time-consuming tasks like generating documentation, writing tests, and optimizing performance. These hooks work in the background, triggered by events like saving files or making commits. It’s like having an experienced developer constantly reviewing your work and handling the maintenance tasks that often get delayed. 3 - Kiro provides a purpose-built interface that adapts to how developers work. Whether you prefer chat interactions or working with specifications, Kiro supports your workflow while keeping you in control of the development process. Kiro is really good at "vibe coding" but goes well beyond that. While other AI coding assistants might help you prototype quickly, Kiro helps you take those prototypes all the way to production by following a mature, structured development process out of the box. This means developers can spend less time on boilerplate code and more time where it matters most – innovating and building solutions that customers will love. Starting today, Kiro is available for free during preview and supports most popular programming languages. Here’s how to get started with @kirodotdev today: kiro.dev/blog/introducing-ki… Excited to see how developers use Kiro, and to work with the developer community to continue to shape Kiro moving forward.
125
377
215
2,117
670,163
Customer Support AI Explorer retweeted
ChatGPT is rolling out a new mode called “Study Together” 👀 Instead of giving you answers, it acts like a tutor - asking guiding Qs and walking through problems step by step. Feels like a big step towards personalized learning (if it works!)
192
529
140
9,719
834,606
Customer Support AI Explorer retweeted
YC’s Hidden Formula: 100 Users, $100/Month, $10k MRR – The Startup Playbook:
48
119
11
1,495
188,831
🔥 Just turned Cursor into Windsurf using only a rules file No extensions ❌ No libraries ❌ No subscriptions ❌ Just markdown and prompt engineering ✅ Here's how I gave my AI IDE a project manager personality:
1
55
## Task List - [ ] Setup FastAPI structure - [x] Configure environment - [ ] Add authentication ## Current Goal Setting up basic FastAPI structure This isn't just a todo list. It's a state machine that controls your AI's execution flow.
1
21
With a "heartbeat" that forces updates every 5 messages or 10 minutes. No more "coded for an hour, forgot to document anything" Full breakdown: medium.com/@pablo.ambram/bri… Just the Gist: gist.github.com/pambram/e937… Your AI assistant having ADHD is a solvable problem.
27