HNHacker News
TopNewBestAskShowJobs

rancar2

466 karma · joined January 31, 2011

http://www.rancar2.com/
submissionscomments
rancar2··on Show HN: VaultSandbox – Test your real MailGun/SES/etc. integration
Having sent billions of emails between multiple startups:

RE setup and testing: Trust (as is most devops one-time setups). Once the initial email setup is complete, you typically aren’t paying with it much. The black swan outages aren’t really an active concern.

RE PII: email is non-secure and shouldn’t have sensitive data in production either. Also, dev/test shouldn’t have PII in regulated industries as a good hygiene practice (I’ve worked in healthcare, finance, and national security contexts).

Re licensing: I appreciate your openness and clarity on the licensing of the gateway engine as AGPL vs MIT for the rest. There’s a more modern licensing approach with FSL-1.1-MIT. It may be a better fit for customers (ie clear licensing terms when using a paid license and less concerns if the business goes defunct or pivots) and for your business plans.

rancar2··on Show HN: Vibium – Browser automation for AI and humans, by Selenium's creator
I wasn’t able to gather the future state plans beyond what’s noted in the V2 plans:

https://github.com/VibiumDev/vibium/blob/main/V2-ROADMAP.md

What’s next 5 years look like given that you are very good at building long-term projects that last and evolve through time? And for a very specific example, what’s the plan for incorporating new standards like Agent Skills as they quickly evolve and launch?

rancar2··on Show HN: I designed my own 3D printer motherboard
Nice work; I’d love to see a V2. Quick tip: try Flux AI to help accelerate the V2 work!
rancar2··on Launch HN: Mentat (YC F24) – Controlling LLMs with Runtime Intervention
Thanks for the thoroughness! I look forward to the next steps as you all apply this approach in other unique ways to have even better results.
rancar2··on Launch HN: Mentat (YC F24) – Controlling LLMs with Runtime Intervention
Can you share more about the challenges ran into on the benchmarking? According to the benchmark note, Claude 4.5 Opus and Gemini 3 Pro Preview exhibited elevated rejection and were dropped from TruthfulQA without further discussion. To me this begs the questions, does this indicated that frontier closed SOTA model will likely not allow this approach in the future (ie in the process of screening for potential attack vectors) and/or that this approach will only be limited to a certain LLM architecture? If it’s an architecture limitation, it’s worth discussing chaining for easier policy enforcement.
rancar2··on Stopping bad guys from using my open source project (feedback wanted)
On what license to choose, this talk by Adam Jacob is good:

https://www.youtube.com/watch?v=rmhYHzJpkuo

And if you want to read about open source vs source available, this GitHub with the Red Hat lawyer and co-author of GPLv2 provides a TLDR of the sentiment. The reference from Chad gives a deep dive into the discussion and origin of FSL’s language.

https://github.com/liquibase/liquibase/issues/7374

rancar2··on Effective harnesses for long-running agents
Having done this for Gemini CLI to get it to behave well several months ago to have a non-coding LLM CLI without costs, I can attest that these tips work well across CLIs.
rancar2··on Show HN: SyncKit – Offline-first sync engine (Rust/WASM and TypeScript)
This is a write up that may be helpful for you: https://sqlsync.dev/posts/stop-syncing-everything/
rancar2··on Show HN: SyncKit – Offline-first sync engine (Rust/WASM and TypeScript)
Please do in the repo, and thank you for the wonderful contribution on multiple fronts. This is very well done work and documentation. A few tips to help others discover the value on the Readme: Add the key memory benchmark (currently just above the acknowledgments) to the top of the Readme and link to the rest of the benchmarks on the page that you already have (maybe rename it benchmarks.md in the process): https://github.com/Dancode-188/synckit/blob/main/docs/guides...
rancar2··on Using AI to negotiate a $195k hospital bill down to $33k
I was going to say please use and donate to 'Dollar For' [1] which provides this service, which is likely a better choice for this type of problem than trying DIY.

[1] https://dollarfor.org

EDIT: adding in a link to 'Dollar For'.

rancar2··on Ask HN: How to boost Gemini transcription accuracy for company names?
The business edition of Wispr Flow does this well, and includes sharing among teams so you can make sure that the company wide vocabulary is consistent and well recognized.

https://wisprflow.ai/business

rancar2··on Recursive Language Models (RLMs)
For that issue, try Codex until Claude catches up to your style.
rancar2··on Show HN: Metorial (YC F25) – Vercel for MCP
I like the license (FSL) chosen for the project, but it may need some explaining for others. Can you comment on decision for selecting the Functional Source License (Version 1.1, ALv2 Future License), and the intent from the Metorial team with it including any restrictions on potential commercial use of the platform (i.e. free-to-paid without notice)?

For those who aren't aware of what FSL (https://fsl.software/) is: "The Functional Source License (FSL) is a Fair Source license that converts to Apache 2.0 or MIT after two years. It is designed for SaaS companies that value both user freedom and developer sustainability. FSL provides everything a developer needs to use and learn from your software without harmful free-riding."

rancar2··on GNU Health
That’s from the founder of Clear.dental, Dr. Tej Shah:

https://clear.dental/

https://gitlab.com/cleardental

https://www.linkedin.com/in/tej-shah-17829195

rancar2··on F3: Open-source data file format for the future [pdf]
Andrew, it’s always great to read the background from the author on how (and even why!) this all played out. This comment is incredibly helpful for understanding the context of why all these multiple formats were born.
rancar2··on Ask HN: What are you working on? (September 2025)
I especially liked the browser frame page. It’s so beautifully crafted. I would move the inception example onto the homepage and up on the examples as it is a good example of love put into the whole project and its execution. https://quietui.org/docs/components/browser-frame#embedding-...
rancar2··on Claude Code 2.0
A user can append similar system reminders in their own prompt. It’s one of the things that the Claude Code team discovered worked and now included in other CLIs like Factory, which was talked about today by cofounder of Factory: https://www.youtube.com/live/o4FuKJ_7Ds4?si=py2QC_UWcuDe7vPN
rancar2··on US airlines are pushing to remove protections for passengers and add more fees
“The U.S. Department of Transportation (DOT) encourages all airlines to guarantee that young children are seated adjacent to an accompanying adult without charging any additional fee.” https://www.transportation.gov/airconsumer/airline-family-se...

There is not coverage beyond one adult already in the US. With an additional adult and one child, the airlines already adds in fees. It’s also non-transparent when booking that they have made sure the easy path is the charged path, especially now that airlines make you pay to guarantee being seated together prior to flight checkin 24 hours in advance of takeoff.

rancar2··on Ankit Gupta Joins YC as General Partner
The thesis was dated prior to the hard tech wave, and it is a perfect time to update it as Boston has a lot of the talent. Plus, Link Ventures is based in Boston with $1B AUM focused on AI. We have the talent here in Boston. That said I’m biased though as a cofounder one of our lovely local AI groups, and I used to have board meetings every two weeks with the Link Ventures executive team at my last startup.
rancar2··on Show HN: Greppers – fast CLI cheat sheet with instant copy and shareable search
You may be looking for do thing like this tool: https://github.com/closedloop-technologies/autocomplete-sh
rancar2··on AI not affecting job market much so far, New York Fed says
I’m in Portugal at the moment and went to their version of Target using Sensei [1] technology. I was with my little one who was grabbing things off the shelves while I put them back, and I was using a stroller not a store cart. The system worked flawlessly including consumed cafe purchases (coffee, pastry, and bread) along with per kg fruits and vegetables. One of the odd benefits is that I felt a bit safer with my little one roaming freely as I knew he was being carefully watched.

[1] https://www.sensei.tech/

rancar2··on Some thoughts on LLMs and software development
My favorite quote to borrow: “Furthermore I think anyone who says they know what this future will be is talking from an inappropriate orifice.”
rancar2··on Launch HN: Embedder (YC S25) – Claude code for embedded software
Check out https://xnilio.com/
rancar2··on Claude Code is all you need
You can make Gemini CLI much better by making it behave more like Claude Code. Claude Code has some lovely prompt engineering at the system and subsystem level that can be replicated with Gemini CLI. I’m having great results already. I am still perfecting process and prompts to be a fully agentic system that can do well on benchmarks but more importantly do the right work with steerability, which was an absolute pain with Gemini CLI out-of-the-box. If you are interested, I can publish some of the basics now and then I can keep you posted as I develop it into a more robust system. Just email me at randycarlton@gmail.com with the subject: SaaS.bot (where this work will likely reside).
rancar2··on Show HN: I've been building an ERP for manufacturing for the last 3 years
3D print manufactures that print custom parts need to use their manufacturing and billing operation outside the mold. Unixhero provided another example, but there are a lot of examples where big ERPs aren’t the right solution and/or require a lot of money/customization/ongoing costs to make it work well beyond its total ROI.
rancar2··on Show HN: I've been building an ERP for manufacturing for the last 3 years
Operations and planning are performed when one’s company fits the mold of a classic business. The system you are building suits the companies that break the mold, which is a growing number as personalization/hyper customization grows.
rancar2··on Show HN: I've been building an ERP for manufacturing for the last 3 years
TLDR Brad: I’d suggest you look at any customers with potential scale, who should have a custom SKU/BOM/order mix. 3D printing and full custom made to order products will have this mix, and it gets increasing painful when scaling past 100s of orders per month. One will find there will be in-house systems and workaround to deal with this complication, which should be fixed decently once a company is in the 1000s of orders per month by necessity of survival.

A bit more background as there is various bits of advice in these threads, and I will provide my take with scaling such a startup. Third-Party ERPs from the big vendors are purchased by Finance and are needed for validation pre-IPO and into the IPO (no one is going to trust something else without proof of success in publicly traded companies and it will be a red flag if there is no use cases in reputable publicly traded companies). ERPs are financial focused (like EHRs in healthcare), and their vendors will happily upsell the other addons like MES/BOMs, which are fine for generic manufacturing with limited SKUs. However in a world of customized/personalized SKUs, traditional ERP/finances solutions cannot be easily used to run manufacturing operations. I’d recommend focusing on integrating into ERPs (tack on custom IDs to the related objects) and automating them rather than building the full financial accounting/taxes into the platform. For example, your platform will still track the BOM details, but the totals will get synced for overall financial reporting for the various ledgers and not all the sub-assemblies which the ledgers don’t care about. This keeps the MES purpose built (and the big vendor ERP keeping simple books) and the ultimate source of truth what’s happening on the floor without getting into the accounting details that matter for tax optimization and not manufacturing operations.

rancar2··on Ask HN: Who is hiring? (August 2025)
Why is Ryan, the current staff product designer, leaving? Oh wait, it’s right there in the job post with a video from him. Good luck to Ryan with his courage to start his own startup!
rancar2··on Anthropic tightens usage limits for Claude Code without telling users
There was a batchmode pulled from the documentation after the first few days of the Claude Code release. Many of have been trying to be respectful with a stable 5 agent call but some people have pushed those limits much higher as it wasn’t being technically throttle until last week.
rancar2··on MemOS, the first 'memory operating system' that gives AI human-like recall
"MemOS demonstrates significant improvements over baseline memory solutions in multiple reasoning tasks over OpenAI baseline with a +38.98% average improvement."

https://github.com/MemTensor/MemOS?tab=readme-ov-file#-perfo...

← PreviousPage 3 of 6Next →