Sometimes when I'm on the go, I don't want to spin up Cursor or Claude on mobile because I don't have the time to think through the prompt or my workspace, and I don't want to waste the tokens on my mobile device. But something I'm constantly doing is telling Gemini (in Google search), or Haiku (in duck.ai) to not: - Lie - Ground its answer 100% on facts - No abbreviations - No Jargon ... and so on. You get the point, and I started thinking it would be really nice if I could use a workspace in the browser, but then I hit two road blocks: 1. The browser only accepts files, not folders. ICM is files and folders, so there's nothing to upload. 2. Even if I get it in there, how do I get the chatbot to actually walk the workspace instead of just reading it? Both are solvable: Roadblock 1: one file, not a folder. I was hoping the browser agents would accept a markdown file, but that didn't work and the ai model would respond with a timeout. Then I tried a text file, same problem. Then I noticed that models would accept PDF as a format. So then, I flatten the whole workspace into a single file, stamping each file's path at the top of its contents so the model can still tell the stages apart. Nothing is lost. The router, the stage contracts, the references are all there, in order, in one upload. Roadblock 2: getting the chatbot to run the workspace. I put one instruction at the very top of the flattened file: when my message starts with >, treat everything after it as the request and walk the workspace on it. Then I upload the file, type > and whatever I need, and the chatbot runs the pipeline instead of describing it. So now, on my phone, from a plain browser tab: > how far is the moon from earth comes back ground-in-facts, no jargon, no abbreviations, every rule I wrote once, applied automatically. No Cursor, no app, no wasted tokens. So, at first I wrote a workspace that prevents drift and I don't have to repeat myself. Then I got to thinking, this would be nice for typing in a city, and getting the latest news, city stats, real estate postings etc by using this, and this idea would really stress test if this concept works on ai search.

So the official launch video released about 2 hours ago, here's the link: https://youtu.be/bn0atstgavo?si=6SIAUKYAQBVIQ4N0 Also, the link to the article: https://www.kimi.com/blog/kimi-k3 First off, this has genuinely surprised me, a 3 Trillion parameter model that isn't open source, but it is open weight (Coming July 27th). The bench marks are insane, beating models like Fable and 5.6 Sol. A quick side note, do not trust bench marks blindly, start using personalized bench marks which relate to your field. If you're a web designer, get the models to go through your process and analyze the output of each model, different models have different advantages. The article has some insane use cases, mainly the 3d Open World, I don't know how many prompts or what framework they ran it with but it still looks quite good. Another interesting point, Kimi K3's article points out it's editing and visual effects capability, including motion graphics. Quite some potential there. The API pricing with a few comparisons of what I feel like are the top models are below: (Name: Input/Output per million tokens) Kimi K3: $3.00/$15.00 Fable: $10/$50 GPT 5.6 Sol: $5.00/$30.00 Caveats to each: Kimi K3: - Relatively more expensive for a Chinese model. - It doesn't always stay on task when there's a lot of context. - If you're switching to Kimi K3 mid session with a lot of context, it may get lost. - Quite slow by the only provider up right now, 28 tokens/second (About half the speed of fable and sol) Fable: - Usage limits finish faster than the tokens it generates. GPT: - Multiple cases of important files or entire code bases being deleted, back to 2025 we go!

I am running an ICM knowledge base for my engineering company using GitHub, VS Code, and Claude Code. The efficiency gains in our workflows have been massive, and I see a clear opportunity to offer this to other regional companies as a new business unit. ​However, this dev-centric stack creates too much friction to be presented directly as a product to non-technical clients. ​The model I envision starts with heavy consulting to align the client's processes and culture. Once maturity is reached, it transitions into a managed service: issue tracking, periodic context pruning, and ad-hoc consulting for new agentic features. ​Has anyone here successfully commercialized this philosophy? I am particularly interested in how you package the delivery and abstract the technical friction (like the IDE and repo management) for the end user. Are there platforms already solving this?

Hey everyone! I’m building my first automation for my mortgage business and could use some advice. 💾 I use LendingPad, but my brokerage doesn’t give me API access. Because of that, I’m looking for another way to automate updates inside the system. 🤖 I was thinking about browser automation, but Codex’s computer control seems pretty token-heavy. I also looked into Hermes Agent, but I’m not sure if that’s the right direction. 🛠️ What tools would you recommend for automating a loan origination system without API access? Browser control, RPA, desktop automation, or something else? 🔐 Security is also a big concern since I handle sensitive borrower information.

I have a few questions, and I would appreciate hearing how others handle this in their own systems. How much freedom do you give the agent to interpret the workflow? For example, do you prescribe the exact files, order of operations, and validation steps, or do you allow the agent to navigate the workspace and decide what applies? Where have you found that flexibility helpful, and where has it caused drift? My own instructions are fairly rigid, and I think they need to be. At the same time, I keep wondering whether one of the benefits of using a folder-based structure is that the system should not require every path and action to be spelled out so precisely each time. How do you decide what must be explicit and what the agent can safely infer from the structure?