I have been living in villages for about 5 years. I started a pig farm a month back. I have 16 piglets now. I still write software on a daily basis, a mix of client projects and own products. The pig farm needs about 2 hours of cleaning each day. I take care of cleaning. My business partner takes care of feeding.

I plan to grow the pig farm to a capacity of 100 pigs. It is a profitable business with roughly 30% return every 6-7 months. We give the pigs a lot more space and care than I have ever seen in any of those factory-style livestock business videos. With a 100 pigs, I will perhaps spend 5 hours a day in cleaning work - with more tools and employing a couple local folks.

Feel free to check out (links in my bio) or reach out if anyone wants to come and try this out in our little village in north eastern India. The village has large farms, growing all sorts of things.

Comments URL: https://news.ycombinator.com/item?id=47414032

Points: 388

# Comments: 82

Works really well. Extracts companies you have dealt with, people, topics, events, locations, financial transactions, bills, etc.

Comments URL: https://news.ycombinator.com/item?id=47362682

Points: 3

# Comments: 0

I would be interested if MetalRT can be used by other products, if you have some plans for open source products?

I initially started with Google Gemini 3 Flash but I switched to Ollama + Ministral 3:3b. The extraction is not exhaustive and there is much to improve but this is working.

dwata runs locally, runs a web backend and the gui runs in browser. Connects to emails, downloads them. Then we can run the financial template detection. It checks for similar looking emails, grouped by sender. Then sends a sample from each cluster to LLM agent. The LLM is asked to find out the parts of text that look like the data we are looking for. dwata then searches for the variables/values that LLM gave in the email, creates a template by replacing the data with template tags. Saves template to DB. dwata parse the data from each email when extracting data.

Roadmap: There is a long way to go, the extractor needs to work much, much, better. dwata will also work on files soon (bank/CC statements).

I want to extract vendors, businesses, contacts, events, places, etc. Connect to different APIs and process everything locally.

dwata will be able to download and process data from Hacker News API too (or other similar sources) - extract entities you care about.

Eventually, only use Ollama/Llama.cpp with models that fit 6-8GB graphics cards or 16GB unified memory only!!

Comments URL: https://news.ycombinator.com/item?id=47324692

Points: 1

# Comments: 0

I built https://github.com/brainless/dwata to submit for Google Gemini Hackathon, and focused on an agent that would replace email content with regex to extract financial data. I used Gemini 3 Flash.

After submitting to the contest, I kept working on branch: reverse-template-based-financial-data-extraction to use Ministral 3:3b. I moved away from regex detection to a reverse template generation. Like Jinja2 syntax but in reverse, from the source email.

Financial data extraction now works OK ish and I am constantly improving this to aim for a launch soon. I will try with Qwen 3.5 Small, maybe 4b model. Both Ministral 3:3b and Qwen 3.5 Small:4b will fit on the smallest Mac Mini M4 or a RTX 3060 6GB (I have these devices). dwata should be able to process all sorts of financial data, transaction and meta-data (vendor, reference #), at a pretty nice speed. Keep it running a couple hours and you can go through 20K or 30K emails. All local!

dwata: Entirely Local Financial Data Extraction from Emails Using Ministral 3 3B with Ollama: https://youtu.be/LVT-jYlvM18

https://github.com/brainless/dwata

Current open weight models < 20B are already capable of being useful. With even 1K tokens/second, they would change what it means to interact with them or for models to interact with the computer.

Models would be available as USB plug-in devices. A dense < 20B model may be the best assistant we need for personal use. It is like graphic cards again.

I hope lots of vendors will take note. Open weight models are abundant now. Even at a few thousand tokens/second, low buying cost and low operating cost, this is massive.

I feel printing ASIC is the main block here.

LinkedIn is a social network and I wish there was an alternative.

Comments URL: https://news.ycombinator.com/item?id=47097498

Points: 2

# Comments: 0

I am building a data extraction software on top of emails, attachments, cloud/local files. I use a reverse template generation with only variable translation done by LLMs (3). Small models are awesome for this (4).

I just applied for API access. If privacy policies are a fit, I would love to enable this for MVP launch.

1. https://github.com/brainless/dwata

2. https://youtu.be/Uhs6SK4rocU

3. https://github.com/brainless/dwata/tree/feature/reverse-temp...

4. https://github.com/brainless/dwata/tree/feature/reverse-temp...

Comments URL: https://news.ycombinator.com/item?id=47088267

Points: 2

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=47062056

Points: 1

# Comments: 0

We are in the early days and I believe that things will get better as more people will calm the f down. People who have built things for ages will continue to do so, with or without coding agents.

In the long term, I think Open Source will win. I can imagine content management systems, eCommerce software, CRM, etc. to all become coding agent friendly - customer can customize the core software with agents and the scaffold would provide fantastic guardrails.

Self-hosting is already becoming way more popular than it ever was. People are downloading all sorts of tools to build software. Building is better. A structure needs to emerge.