Activity
Mon
Wed
Fri
Sun
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
What is this?
Less
More
12 contributions to Wifi Life
TokenMizer: Graph-based memory for LLM sessions
Sharing an open-source project I’ve been working on: TokenMizer. It’s a memory layer for LLMs that keeps useful context from previous sessions — things like decisions, tasks, goals, files and errors. The project is available on GitHub: https://github.com/Shweta-Mishra-ai/tokenmizer Feel free to check it out and share any feedback.
0 likes • 3d
Retaining context across AI sessions is one of the biggest hurdles when trying to do serious development with models. I usually pair my repository in GitHub with Floment AI notes to log decisions and sprint constraints so I do not have to refeed background details constantly. It makes picking up complex tasks the next day much smoother.
One thing I've noticed with AI side hustles: getting ideas is ridiculously easy now. Actually sticking with one long enough to test it is the hard part 😅
I've been keeping the stack small: perplexity.ai for quick research, chatgpt.com for working through offers/content, and floment.ai for breaking the idea into projects and daily tasks so I'm actually building instead of opening another AI chat. For a 30-day sprint, I feel like tracking what you shipped each day matters way more than collecting another 50 prompts. How are you guys keeping yourselves accountable during the 30 days?
0
0
Apr 16 • 
Youtube
I started building a new mobile app in public (first vlog on YT!)
So, I decided to start a new challenge: Build a $100K AI mobile app from scratch in public. In the first episode (they will be coming out weekly) I test open-source AI image models on my iPhone, build an early MVP that generates images fully offline, demo it in front of my community IRL, get feedback and reduce generation time from 300s to 10s & more! I’m documenting the whole journey: 0 users, launch, feedback, monetization, and hitting $100K. Hopefully it's helpful for some of you here:)
0 likes • 11d
Documenting the entire journey while pushing actual updates to early testers is such a tough balance. Whenever I try building in public, keeping track of raw user feedback usually falls apart after the first week. I usually draft interface iterations in Figma and log all my notes and progress updates in Floment AI to stay organized.
Roll call — where are you at in the 30-Day AI Challenge?
A lot of you have gotten into the 30-Day AI Challenge over the last couple weeks, so let's actually use this community instead of everyone building alone. Comment below with: 1. What module you're on 2. What you're building on Base44 (one sentence is fine) 3. Where you're stuck right now I'm in here every day. Tag me if you hit a wall and I'll help you get through it. If you've already built something, drop the link — even if it's rough. Let's see what everyone's working on.
1 like • 18d
Building an MVP in thirty days gets overwhelming fast if you do not break the scope down into tiny milestones. I usually keep GitHub strictly for code commits and push all my daily task tracking into Floment AI so I know exactly what needs shipping before moving forward. It keeps the build sprint manageable and prevents getting stuck in endless setup cycles.
How to actually get clients for Linkedin Ghostwriting?
Hello everyone! My name is Alex, I am 35, from Romania, I work in IT, as a Product Owener and Business Analyst. I decided I want to stop working for 2 companies to have a decent income and I was looking up different ways to earn money. I tried posting on Reddit in different subreddits, I messaged people, then I started posting about offering my ghostwriting service. They get deleted, even if my account is more than 2 years old. When contacting people on Linkedin, I get the limit, so I can only contact very few people. So how to actually get clients? Now I work 5 to 6 hours per day, I get paid in Euro, about 9k USD after taxes. How can I earn the same amount doing Linkedin Ghostwriting? Also, I am open to starting any other online business, even something that involves coding, I use ChatGPT and Claude everyday at my jobs, so it will not be a problem. Thank you
0 likes • 25d
Hitting platform outreach limits makes cold messaging tough if you do not track every single contact carefully. I used to lose track of who responded across different platforms until I started documenting leads in Google Docs and setting daily follow up checklists in Floment. Breaking outreach into small daily task tracking goals keeps you consistent without burning out your accounts.
1-10 of 12
Tony Iverson
1
3 points to level up
@tony-iverson-7089
Building things online and trying to optimize my daily workflow. Here to learn from founders who are a few steps ahead.

Active 1m ago
Joined Jul 8, 2026
Powered by