Ventus (@venthewise) header
Ventus (@venthewise) profile picture

Ex transportation tech-startup founder now all-in on ai. Running on coffee and curiosity. explorer of anything ai.

Joined December 2023
Photo by Ventus (@venthewise)Photo by Ventus (@venthewise)Photo by Ventus (@venthewise)Photo by Ventus (@venthewise)Photo by Ventus (@venthewise)Photo by Ventus (@venthewise)Photo by Ventus (@venthewise)Photo by Ventus (@venthewise)Photo by Ventus (@venthewise)Photo by Ventus (@venthewise)
I'm 36. What if?.. 8 years ago, an elevator pitch to a friend gave birth to my transportation tech startup. Pure luck, pitched to the right person at the right time. Did the product design, backend algo and operations workflows. 7 years ago, I sold it. Dad got into an accident, sold everything including my car. Started the downward spiral mentally and financially. What if my dad just stayed home that night? 6 years ago, COVID happened. Went back home. Fell into deep self pity and depression. 5 years ago, thought that I can't live my life as someone useless while parents are getting old. Borrowed my sister's old laptop and studied crypto, got lucky with BONK and other memecoins and nft trades. 4 years ago, bought myself a laptop. Taught myself coding. It was hard. Very hard. Barely creating anything and stackoverflow was my bestfriend. 3 years ago, finally getting somewhere after a year but my programmer friends who have been in the IT industry for over 10 years, were lowkey laughing at me at my decision to study coding at such a late stage in my life. My cofidence dropped to its lowest. Last quarter of the same year, I discovered chatGPT. This was my life-changing moment. 2 years ago, i put in all my time, savings and effort into studying the possibilities and capabilities of ai in tech. Built the knowledge, got broke, was so close to giving up and going back a 9 to 5. What if I just gave up and went back to being employed? 2025. The new year's celebration wasn't something I was happy about. But. I discovered n8n. And with the ai knowledge I already built up, the things I imagined that could be done was endless! 1st half of 2025, spent time and borrowed money to get ai certifications while burying myself in creating ai workflows in n8n. 2nd half of 2025, with just the right amount of confidence with my new skills, I contacted people in linkedin who might be interested and a friend referred a construction firm in Southeastern US. Not wanting the momentum to fade, I hopped into upwork and started closing clients who needed ai services. Ai receptionists, automated social media pipelines, manual labor to automated workflows and so much more. 2025 was also the year I started using Grok Code Fast 1 alongside Kilo Code inside VS Code "heavily". Ended 2025 with $28,000 in my pocket, a 2nd hand SUV, monthly retainer fees and a brand new level of confidence that I thought will never come back. Not grand, but it's a start. 2026. This year, closed new clients. Started doing custom software as a service while building apps for myself and for fun using newer and better tools. Discovered ComfyUI, local hosted ai models and other opensource resources, and bought another workstation for them. 2026. The year where I have no boss, no business partners and no investors to impress and report to. Not rich "yet", but can buy what I want, when I want. Go to places where I can still work with my clients remotely, without worrying of the expenses. I'd like to think Im financially-free already, but I know it's not that simple. But Im just getting started. I'm 36. So.. what if I stopped before the breakthrough happened?
17
1
23
737
Connecting with people while building on public here on X? I'm into ai and automations. If you're building the same as me in tech, Let's connect! πŸ‘‹
27
18
429
Ventus retweeted
I've signed up for the waitlist! If you're using Grok bot, this is a must-try imho. The project mentioned here is @HeyLingerAI, an open-source personal agent that runs fully on your PC, with a local model, memory, and orchestration. Strengths vs Grok Bot πŸ‘‡
An open-source personal AI agent that runs 12 leading models locally on your machine: up to 5x faster, with $0 in local token fees. Built by Tony Zhang (early DeepSeek core, ex-MSRA), it delivers a Grok-style experience right on your computer: β†’ Run 12 top models locally on supported hardware, including Qwen, DeepSeek, and Gemma. β†’ Sync context across your authorized files and apps. β†’ Retain your preferences and past decisions. β†’ Coordinate agents to drive tasks to completion. β†’ Pay $0 in cloud API fees for local inference. Your data stays on-device in local mode; connected features and cloud models use external services.
4
2
1
7
202
GTA VI speedrun by iShowSpeed for 52 hours! That ending is insane bro. 😱
2
6
234
Ventus retweeted
There are good days when trying to grow here on X, then there's this. 1/3 Here's a short X test result I did in chasing the 500k verified impressions and using the opensource X algo. See this? You know how many posts I did to be able to get this? πŸ‘‡ (Speedrun to 3/3 for the sauce)
7
1
9
269
I agree with @elonmusk , connecting on X is one of the smartest decisions an ai builder can do.
3
5
184
There are good days when trying to grow here on X, then there's this. 1/3 Here's a short X test result I did in chasing the 500k verified impressions and using the opensource X algo. See this? You know how many posts I did to be able to get this? πŸ‘‡ (Speedrun to 3/3 for the sauce)
7
1
9
269
3/3 The X algo, as it's opensourced gives all of us a guide on how to get our content distributed. - Post consistently in your chosen niche. The X algo groups us and our audience in 1 main cluster (The topic we post most about) and a few sub clusters (Things we post every now and then, those random thoughts, news, etc.). Prompt to Grok: "Review the x algo repo and my x account. Determine my Main and sub cluster and rank them based on my own numbers." This will give you a rough idea about your niches based on what you;ve posted already. - Posts within the main and sub clusters has a chance to pick up traffic since the algo knows how to distribute it. - The audience is big. In the ai niche for example, audience are diverse, so any tone can go viral. Funny, informative, a random fact, a question (this is what hit for me). Bottomline: If you think your clusters are set, be wildly creative and just post it. 🀝 Goodluck to all of us in getting eligible in OCR.
1
2
38
Hello @aniqatc sorry I posted this very late!
1
1
27
Didn't hit my 150k verified impressions for the first 30 days of trying to join the X OCR Program. 🫠 But the island vacation was a much needed rest for me. So I'm pumped after getting back. Month 2 starts today! Not starting blind, algo already knows my audience, I just need to grind the hell out of me to be able to provide max value for views. That's gonna be consistency + adding value. Basically showing up with substance and lots of luck and prayers. πŸ˜‚ 380k verified impressions to go. Good morning builders!
18
19
431
Not much really. It's more on just continue finding high value pieces to post
1
1
36
I've signed up for the waitlist! If you're using Grok bot, this is a must-try imho. The project mentioned here is @HeyLingerAI, an open-source personal agent that runs fully on your PC, with a local model, memory, and orchestration. Strengths vs Grok Bot πŸ‘‡
An open-source personal AI agent that runs 12 leading models locally on your machine: up to 5x faster, with $0 in local token fees. Built by Tony Zhang (early DeepSeek core, ex-MSRA), it delivers a Grok-style experience right on your computer: β†’ Run 12 top models locally on supported hardware, including Qwen, DeepSeek, and Gemma. β†’ Sync context across your authorized files and apps. β†’ Retain your preferences and past decisions. β†’ Coordinate agents to drive tasks to completion. β†’ Pay $0 in cloud API fees for local inference. Your data stays on-device in local mode; connected features and cloud models use external services.
4
2
1
7
202
Linger strengths vs Grok Bot - Privacy / ownership: data stays on your device, not in a provider’s account. - No token bill, no subscription gate: local inference is free after hardware, and you can bring your own runtime β€” Claude Code, Codex or Pi, on local models or a plan you already pay for. - Local big-model trick: PolyStrata keeps the busiest MoE experts on GPU and the rest in RAM/SSD, so the 125B model runs on a 12GB card with 64GB RAM. - Memory + learning is explicit: 95% LoCoMo vs 94.7% for Zep, 92.5% for Mem0 in their chart, and the nightly β€œDreaming” loop turns two similar reactions into a rule β€” e.g. vendor invoices β†’ tell now. - Open source and inspectable: you can read the harness, unlike Grok Bot’s hosted black box.
1
2
90
Bottomline Pick Linger if you want Grok Bot-style proactive work, but local, inspectable, and with no per-token bill β€” and you have the hardware to run it, or you’re happy routing hard jobs to Claude Code/Codex. One caveat: Linger still needs a decent GPU + RAM to get the good model. Atleast a 12GB GPU + 32GB ram.
1
17
What's up with @TencentHunyuan Hy4?! It's so good! Tested alongside @deepseek_ai V4.1 Flash inside Deepseek Harness with 1 prompt to build an app. Deepseek is still not done yet, and I needed to steer it through the coding process. No readiness and security checks yet. It already used up 166M tokens with Cache hit of 98%. Will most likely double this to finish. Hy4 Flash on the other hand. Finished readiness and security checks already after building a fully functioning prototype. 32M tokens, 97% cache hit. Bonus: UI made by Hy4 Flash looked even better than the one made by Deepseek and my very first prototype made by Grok. Looks like I'm gonna be having fun after getting back from my island trip with this new ai model I just found. 😎
1
8
152
When I was starting building on X, I already had a plan. That 'plan' was what trapped me. Most of us are guilty with this. After seeing it, you'll most likely agree that at some point you were like that when you started. Ready? Ready? It's this: We started posting like big accounts, famous influencers and credible experts on X. Truth is, big accounts can post that because people already care. Small accounts have to make people care first. So now I try to write every post like I'm talking to a stranger who has 1 second to care. Still small, still learning, but this made way more sense to me.
3
97
Linux has to wait. With a new project in, I'll need more coding power. Let's see how Deepseek Harness performs vs Hermes.
1
5
118
Just gonna go mess around your heads and have fun here on X while building.πŸ’€
2
9
203
Building your brand here on X is hard when you're a small account. I know. I'm one of them. πŸ₯² But we have different tools inside the Creator Studio. Like this one. πŸ‘‡ Do you also use all the creator tools here in X?
2
6
155
Bro, growing on X is already hard enough as it is. And now I have to worry that Im not tagged as an ai when replying? wtf? 🫠 Time to show the world how bad I am at spelling. πŸ₯²πŸ˜
It took a while but as promised here it is... "THE 𝕏 SPAMMERS LIST" πŸ›‘οΈ Blocked for AI reply blockedforaireply.levels.io 3,728 𝕏 accounts that continuously spam automated AI replies to artificially farm engagement and ruin 𝕏's social network - 498 already suspended by 𝕏 - continuously updated βœ… Evidence for every blocked account βœ… One-click block all button βœ… Continuously updated JSON API βœ… Text-only username list Based on @ryanvogel's original Clankwall and @dillon_mulroy's X list I hope everyone uses this and keeps using it, so my account becomes a honeypot for AI replies, once I block them they're blocked by the entire network and they stop being able to reply to anyone + It'd be cool to add more "scouts" like @ThePrimeagen and @JackEllis and sync their block list continuously too cause they and many others have the same issue I've also submitted this list to @X's CEO @petersellis to train their spam models on Again it'll be continously updated! Enjoy a spam free 𝕏 experience! πŸ˜ŠπŸ™
11
17
881