Feed aggregator

Article URL: https://twitter.com/rasmushauschild/status/2077483318937862336

Comments URL: https://news.ycombinator.com/item?id=49898507

Points: 1

# Comments: 0

Article URL: https://fiodar.substack.com/p/the-art-of-agentic-software-development

Comments URL: https://news.ycombinator.com/item?id=49898500

Points: 1

# Comments: 1

Article URL: https://blygger.org/talks/2026-09-24-blygger/

Comments URL: https://news.ycombinator.com/item?id=49898495

Points: 1

# Comments: 0

Does anyone have access or a pointer to the origin for the Fong statistical test used in Kano scoring? Lots of websites reference it as "the thing" to use when evaluating Kano surveys, and I found a bunch of academic papers referencing it second hand, but the apparently original paper "Using the Self-Stated Importance Questionnaire to Interpret Kano Questionnaire Results" that seems to have been published in 1996 evades my online searching skills. I'm not looking for a pirated copy, happy to buy it from the publisher, but it seems to have fallen into a black hole. Searching for the paper just returns a ton of secondary references, that quote it.

The reason why I'm looking for it is that the formula looks like a two-proportion z-test that should work if the two samples are independent, but Kano scores are from the same population, so if I understand this whole thing correctly, the main test that the web and the academia copies and pastes doesn't really apply.

I'd love to peek into the original paper and understand what it was based on and why. If you know where I could find a copy, or maybe have one handy to check what Fong actually wrote about it, I'd appreciate a pointer very much.

Comments URL: https://news.ycombinator.com/item?id=49898492

Points: 2

# Comments: 0

Article URL: https://artificialanalysis.ai/models/releases/gpt-6-1-sol

Comments URL: https://news.ycombinator.com/item?id=49898469

Points: 2

# Comments: 0

OpenSEO got to 22k+ stars in <7 months

We were planning to opensource 4 tools (2 of which we have build but they are not opensource yet)

- Meltwater alternative for social listening (already taken: freesociallistening and getting some users from chatgpt)

- Gong alternative for sales roleplay (already setup dosalesroleplay)

- Podium alternative for Review collection and SMS

- AgencyAnalytics alternative for agency reporting, we already have some sort of tool setup to run our own firm we can build it out and opensource it

our right to win is our engineering team based in India where we have already deployed and run 50M+ voice mins, 1K+ brands use our platforms to run paid ads, and we generate more than 5K+ leads/month

one of the biggest constraint we end up facing is that we don't end up speaking with enough customers or pace of improvement is pretty low post we get something to a nascent build stage, believe bunch of us would have faced similar issues here.

what I wanted to understand was that we opensource solutions for the above would anyone be open to try it out and give feedback, and if yes what would it take for you to try it

Questions from my end: 1. which tool do you think we should opensource first 2. what would make someone to switch to our alternative & what can possibly stop from the same happening 3. what are the sources that matter for social listening? 4. what is the license type which would be widely used? 5. how does these opensource tools gets traction/maintainers/early adopters?

advice/feedback/inputs would be really helpful

thanks HN community!

Comments URL: https://news.ycombinator.com/item?id=49898465

Points: 2

# Comments: 0

Article URL: https://www.nature.com/articles/s41598-026-41168-0

Comments URL: https://news.ycombinator.com/item?id=49898459

Points: 1

# Comments: 1

Article URL: https://www.reuters.com/business/anthropics-518-billion-ai-buildout-hinges-largely-deals-that-cannot-be-canceled-2026-09-29/

Comments URL: https://news.ycombinator.com/item?id=49898457

Points: 2

# Comments: 0

Article URL: https://twitter.com/spencerpratt/status/2104944747806695496

Comments URL: https://news.ycombinator.com/item?id=49898450

Points: 1

# Comments: 0

Article URL: https://punkcake.itch.io/spaghetti-celesti

Comments URL: https://news.ycombinator.com/item?id=49898449

Points: 1

# Comments: 0

Article URL: https://circleid.com/posts/openai-agent-bypasses-internet-restrictions-through-dns

Comments URL: https://news.ycombinator.com/item?id=49898442

Points: 1

# Comments: 0

Article URL: https://github.com/GregDixonMXN/annalist

Comments URL: https://news.ycombinator.com/item?id=49898434

Points: 1

# Comments: 0

Article URL: https://newsletter.pragmaticengineer.com/p/shopify-native-mobile

Comments URL: https://news.ycombinator.com/item?id=49898423

Points: 2

# Comments: 0

Article URL: https://martinalderson.com/posts/ai-margin-collapse-gathering-pace/

Comments URL: https://news.ycombinator.com/item?id=49898418

Points: 3

# Comments: 0

Article URL: https://www.nber.org/papers/w35782

Comments URL: https://news.ycombinator.com/item?id=49898416

Points: 2

# Comments: 1

I'm tired of AI coding agents which are good for 30 seconds and then completely fail on the slightest hiccup: a failing test, a temporarily unavailable dependency, an incorrect assumption. Sometimes I end up having to babysit these agents anyway.

Relay is a harness for coding agents which takes advantage of the fact that agents are often good at doing something slightly wrong, and not so good at doing something correctly and completely. Relay repeatedly tries the task and on each failure, uses the output to find a better way to do it.

It does this by running the task in a loop: attempt the task, run the checks in a fresh sandbox, on failure read the error and retry instead, until the checks pass. What it learned from a run is retained, so that it doesn't repeat the same mistake, and when the checks eventually pass, it will ship the change as a normal git PR - no hidden state, no magic, the diff is human readable.

A few specifics:

- it runs locally, is free, and doesn't require an account to run the local agent - it uses a top-tier model to plan/review, and cheaper models to actually write code (for cost reasons) - happy to discuss why this is a good idea and why it isn't - the checks run in an isolated sandbox for each task, rather than "trusting" the agent - currently supports [list actual models/integrations - Opus 5.5, GPT-6, etc] - there's a paid tier for running this unattended across multiple repos with shared sandbox minutes, the actual agent is free.

It's not magic, it's not AGI, and it will fail badly on some problems.

Would appreciate feedback, particularly from people who've deployed agents unattended and found that they don't actually work, or have had to build substantial tooling around them to get useful results.

Comments URL: https://news.ycombinator.com/item?id=49898411

Points: 2

# Comments: 0

The AIxCC competition winner will analyze messaging app code and compiled binaries for vulnerabilities, with technology that could also help commercial customers secure their software.

The post DARPA Selects Xint to Use AI in Securing Military Messaging Apps appeared first on SecurityWeek.

The latest vulnerabilities to be uncovered in the frequently-targeted Citrix NetScaler product lines were being exploited well-before disclosure, prompting disquiet.