I have built a AI powered home improvement platform with live consultation with pros.

I am not sure if I will go live with it.

It allows those professionals experts across the USA provide help to Do It yourself consumers for a fee. Consumers can be anywhere.

So I married sort of like Uber (rent skills) + upwork (rent + fees) + FaceTime + e-commerce. realtime audio transcription that identifies parts you need and builds a list for pros and you to review which you then go shop.

: Meet Handy — AI + Live Experts for Every Fix.

: Instant, intelligent home-improvement help — see it, solve it, and shop for it, all in one live session.

Live Video Calls with Pros Instantly connect with verified experts via real-time video. No scheduling hassle — just point your camera and get help.

AI-Powered Visual Assistance HandyLens AI analyzes what the camera sees, highlights problem areas, and guides both consumer and pro with contextual prompts.

Domain Expertise Specialized AI Packs (Plumbing, Electrical, HVAC, Painting, etc.) ensure every session applies the right technical and safety knowledge.

Actionable Fix Path Each call ends with a clear, AI-generated “Fix Report”: what to do, parts needed, and next steps.

Commerce & Trust Built-In Integrates with retailer catalogs for instant part links, and captures verified pro ratings and summaries for quality assurance.

Paid Live Video Sessions (Primary Income) • Pros earn per-minute or per-session fees for live video consultations. • Example: • $3–5 per minute • Typical 10–20 minute call → $30–$80 per session • No travel time, no fuel, no tools, no overhead.

Pros log in when available (between jobs, evenings, weekends). • Ideal for: • Independent contractors • Apprentices with experience but limited licensing • Semi-retired pros • Pros during slow seasons

This unlocks underutilized labor.

⸻

4. Lead Conversion (Optional Upside) • If an issue requires in-person work, pros can: • Convert the session into a local job lead • Or refer the job and earn a referral fee

I have never had a luck with using Gemini. I had a pretty good app create with CODEX. Due to the hype I thought let me give Gemini a try. I asked it find all way to improve security and architecture / design. sure enough it gave a me a list of components etc that didn’t match best patterns and practices. So I let it refactor the code.

It Fucked up the entire repot. It hard coded tenant ids and used ids, it completely destroyed my UI. Broke my entire grpahql integration. Set me back 2 weeks of work.

I do admit the browse version of Gemini chat does much better job at providing architecture and design guidance time to time.

Do you use AI agents on repos without version control?

I can't emphasize this enough, it doesn't matter how good a model is or what CLI I'm using, use git and chroot (at the least, container is easier though).

Always make the agent write a plan first and save it to something like plan.md, and tell it to update the list of finished tasks in status.md as it finishes each task from plan.md and to let you review the change before proceeding to next task.

> Set me back 2 weeks of work.

How did this happen?

Did you let the agent loose without first creating its own git worktree?

I'll reply to myself since apparently people downvote and move on:

They're useful for allowing agents to work in parallel. I imagine some people give them access to git and tools and sandbox the agents, then let a bunch of them work in separate git worktrees pointed at the same branch, then they come back and investigate/compare and contrast what the agents have done, to accelerate their work.

I think there is value in that but it also feels like a lot of very draining work and I imagine long term you're no longer in control of the code base. Which, I mean, great if you're working on a huge code base since you already don't control that...

Welcome to engineering management.

The average quality of software engineering is abysmal, and the average quality of software engineering management is even worse. I'm not very enthusiastic about where this is headed.

tfw people are running agents outside containers

Yeah this something I need to get to.

Apologies. I meant branch. I nuked the branch. But set me back a lot of time as I thought it may be few things here and there.

I always set gc.auto=0 for this reason, because it means the reflog will be there if I need it (unless I accidentally delete .git)

I see! That is annoying. A good reminder to push or fork the branch for backup.

Haha. This. I had to swallow that $30 a month cost. Because by going through the family sharing, I am saving so much heaxh and time wastage for my family who use YouTube for learning (school, college…), parents who watch videos, listen to music etc. well worth it for me.

Free publicity for Starfield. :)

As a man I disagree.

as a generative AI model, I disagree with your disagreement

Sharing some experience. When hackers steal credit cards and want to verify if it’s working or blocked they also do something similar. Another pattern I have seen is that hackers order multiple items for a seller (retailer website) and have them delivered to multiple addresses over time where they don’t live and don’t care for the item. They however order an item that they want to a address which they can access. This throws off security and police as there is no pattern for machines to track and police don’t have resources to go to all addresses for checking.

i guess for the next order, it has to be a completely new set of addresses?

Doesn’t matter cause still not a huge pattern to be able to figure out and limited resources.

Sharing some experience.

I am doing STT using whisper on my local laptop and recording all the meetings. I am then asking llama and chstgpt to summarize them. ChatGPT does 10x better job. It’s been good to be able to get the main points out of otherwise fuxking boring a. garbage meetings that are at least 30-45 mins long.

Does anyone have info on how to run 70B on windows ? :) would appreciate it.

Not just right now. I have been / am interviewing for VP / CTO roles and by god the process is so draining. By the time I get to interview with the CEO (3/4 times so far) the process gets dragged for weeks and months. The requirements for the positions changes out of nowhere or some exec is on vacation. For example, I interviewed for $25 billion market cap retailer for a VP position. All was great until an SVP which I had already spoken to wanted to talk to me again because they wanted to some question about some very random thing like what would you do if you didn’t know what to do. This after 8 hours of interview with CIO, 4 VPs, 2 SVPs and several directors. None with engineering degrees. They all had been at the company for 13-14 years.

This lasted for 3 months starting October and final rounds in December. Final rejection in mid January. Another Was over 6 months.