We are building NeoSmith and we are looking for 3 teams to work with right now at no cost.
Here is the deal in one line: point your LLM to ours and we automatically create a dedicated Small Language Model for your exact workflow. Nothing from your end. No dataset, no labeling, no fine-tuning work, nothing. You just keep running your agents the way you already do.
What actually happens is this. NeoSmith reads your production traces, figures out what your workflow is doing, and builds a purpose-built model trained specifically on your task. Not a general small model. A model that has only ever seen your workflow, your inputs, your expected outputs.
My game launched here today. It is a small colony builder for iPhone, and the thing people ask about most is that it has an ending.
Most builder games never end. This one does, because it tells a real story with a beginning, a middle and an end. You start on Earth with four people. You dig ore, build power, grow food, and research your way to the Moon and then to Mars. Each world has one resource the others need, so you run freight between them. The story ends when you send a fleet of probes out to the stars and your colonists go into a long sleep. What they wake up to is the ending. After that, the worlds are still yours to keep building.
I ve been working on a programming language called Sere.
The basic idea is to combine Python-like syntax and readability with the performance and lower-level capabilities you d expect from languages like C++ or Rust. Sere is compiled, statically typed, and built around LLVM, with features aimed at both application development and systems programming.
I started it because I wanted something that felt pleasant to write without giving up things like native compilation, generics, memory control, async programming, and C/C++ interoperability.
It s still actively being developed, so there are definitely rough edges, but the compiler, docs, tooling, and language itself have come a long way.
A couple of us kept running into the same problem: cleaning and transforming data either meant fighting ad-hoc Python scripts, or losing track of exactly what happened to the data along the way.
I'm the independent developer behind MonoProxy. I built it for a narrow testing problem: the proxy laptop had become another dependency whenever I needed to check whether a phone, tablet, TV, or test device on the same LAN could actually connect and move traffic. MonoProxy 2.4.0 runs one local endpoint on an iPhone or iPad for HTTP, HTTPS CONNECT, plain WebSocket, SOCKS4/4a, and SOCKS5 clients. The part I found most useful was separating three states that are often collapsed into one green light: 1. Listening: the endpoint is ready. 2. Client Connected: another device reached it. 3. Traffic Verified: bytes actually crossed the proxy. The 2.4.0 update also focuses on the cases that expose weak proxy implementations: large downloads, regular and chunked uploads, Range requests, long-lived sessions, and network recovery. A local setup page gives nearby devices the current host, port, PAC details, and platform-oriented commands. The limits are intentional. This is a local-network tool, not a VPN or a remote proxy service. It does not install a certificate or decrypt HTTPS, and iOS can still suspend networking after the app leaves the foreground. Product details and screenshots: https://proxy.monoware.app/ App Store: https://apps.apple.com/app/apple... For anyone who tests devices on a LAN: which part usually fails first for you reachability, proxy configuration, authentication, or proving that real traffic moved?
Most enterprise software fails quietly. Not at launch. At month 4, when the original champion is pulled into a different project and nobody else fully understands why the system was bought.
Watched this pattern repeat across banking, healthcare, manufacturing deployments. The pattern:
Month 1: champion is energetic, adoption is climbing
