frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Open in hackernews

Ask HN: Are subagents making flagship AI models feel like managers?

2•vgoncharenko•7h ago
Subagents are one of the newest power features in AI coding assistants, and I have mixed feelings about them.

The basic idea is good. If a task is messy, you can split it into smaller threads, run them in parallel, and merge the results. For large refactors, multi-file debugging, or research implementation work, this can save a lot of time.

But I am starting to notice a strange manager effect.

I ask for a top-tier model because I want the best reasoning and engineering judgment. Instead, the main agent often seems to switch into "manager" mode and delegate most of the actual work to smaller models. The smaller model produces an okay solution, and then the expensive model turns it into a polished final answer.

So the smart model becomes a manager, the cheaper model does the heavy lifting, and the whole chain burns tokens. Ironically, the manager is often the most expensive part and the least useful part.

What I want from these tools is mostly transparency and control. I want to see which model handled which step. I want the option to disable delegation, or restrict it to certain subtasks. And if the tool is making a speed-versus-quality tradeoff, I want that tradeoff to be explicit.

Subagents are genuinely useful. But if I pay for a flagship model, I want flagship-level work, not a summary of a weaker model’s draft.

Curious whether others are seeing the same thing.

WikiGacha

https://wikigacha.com/
1•AJRF•1m ago•0 comments

Show HN: NERDs – Entity-centered long-term memory for LLM agents

https://nerdviewer.com/
1•tdaltonc•1m ago•0 comments

Show HN: Traces: A new way to share and discover agent traces

https://www.traces.com
1•nwpwr•2m ago•0 comments

Supertoast Tables

https://hatchet.run/blog/supertoast-tables
1•abelanger•3m ago•0 comments

OpenFelix – Open-source AI assistant for macOS (local MLX and any cloud model)

https://github.com/fspecii/openfelix
1•openfelixai•3m ago•0 comments

Anthropic Open SWE Roles vs. AI Replacement Claims

https://grepjob.com/trends/anthropic-hiring-vs-ai-replacement
1•derektank•3m ago•0 comments

DJI is >96% of RemoteID usage in the US [video]

https://www.youtube.com/watch?v=wK31qL40p6I
1•asparagui•4m ago•0 comments

Sick of Microsoft and Google? The Office EU suite is an open-source alternative

https://www.zdnet.com/article/europe-open-source-office-suite-alternative-to-microsoft-google/
1•CrankyBear•4m ago•0 comments

Forest Aboveground Carbon Storage in the Three Parallel Rivers Region

https://www.mdpi.com/2072-4292/18/5/756
1•PaulHoule•5m ago•0 comments

Clerk Dev Outage

https://status.clerk.com
1•linolevan•6m ago•0 comments

I spent weeks building something I could have paid $30M for and I'd do it again

https://github.com/Lakshay1509/NeatMail
1•mafia15•8m ago•1 comments

Running production AI systems at scale (GoDaddy and AWS case study)

https://aws.amazon.com/partners/success/godaddy-agenticai/
2•tmuhlestein•8m ago•1 comments

Depleted oil reserve leaves US exposed as Iran war pushes up prices

https://www.ft.com/content/d546fd5b-4a0c-4b90-8bee-9e85b75458ed
2•Anon84•10m ago•1 comments

Markdown Browser

https://markdownbrowser.com
2•janwilmake•11m ago•1 comments

A Defence of the Constitutions of Government of the USA (1787)

https://en.wikipedia.org/wiki/A_Defence_of_the_Constitutions_of_Government_of_the_United_States_o...
1•simonebrunozzi•11m ago•0 comments

Spamming with Google Groups

https://cweiske.de/tagebuch/spam-google-groups.htm
2•FinnKuhn•12m ago•0 comments

U.S. Lost 92,000 Jobs in February

https://www.wsj.com/economy/jobs/february-jobs-report-unemployment-1d7d1a9b
1•JumpCrisscross•13m ago•2 comments

Can the Most Abstract Math Make the World a Better Place?

https://www.quantamagazine.org/can-the-most-abstract-math-make-the-world-a-better-place-20260304/
1•Anon84•13m ago•0 comments

A particular kind of dark matter explains mysterious signals from the Milky Way

https://www.kcl.ac.uk/news/dark-matter-explains-mysterious-signals-from-the-milky-ways-centre
2•giuliomagnifico•13m ago•0 comments

Show HN: LoRA gradients on Apple's Neural Engine at 2.8W

https://github.com/jmanhype/ane-lora-training
1•jmanhype•14m ago•1 comments

Modular Diffusers: Composable Building Blocks for Diffusion Pipelines

https://huggingface.co/blog/modular-diffusers
1•gmays•14m ago•0 comments

Show HN: Claude skill to do your taxes

https://github.com/robbalian/claude-tax-filing
4•rob313•17m ago•0 comments

Apple's 512GB Mac Studio vanishes, a quiet acknowledgment of the RAM shortage

https://arstechnica.com/gadgets/2026/03/apples-512gb-mac-studio-vanishes-a-quiet-acknowledgement-...
2•oleszhulyn•18m ago•0 comments

Automating the process of validating a business idea

https://www.founderspace.work
2•benga_radu•19m ago•0 comments

Apple M5 Max appears in Geekbench, tops M3 Ultra in multi-core score

https://videocardz.com/newz/apple-m5-max-appears-in-first-geekbench-result-tops-m3-ultra-in-multi...
2•LorenDB•19m ago•0 comments

Always-on detections: eliminating the WAF "log versus block" trade-off

https://blog.cloudflare.com/attack-signature-detection/
1•unknownhad•19m ago•0 comments

OpenWhispr – Open-source WhisperFlow alternative, runs on your Mac

https://github.com/MrPrinceRawat/OpenWhispr
4•mrprincerawat•21m ago•0 comments

Paperclip: Open-source orchestration for zero-human companies

https://github.com/paperclipai/paperclip
3•eclipxe•21m ago•0 comments

Minimum Viable Slop

https://medium.com/@nftechie/minimum-viable-slop-28c6594d9f91
1•awormuth•21m ago•0 comments

Show HN: Anchor Engine – Deterministic Semantic Memory for LLMs Local (<3GB RAM)

https://github.com/RSBalchII/anchor-engine-node
1•BERTmackl1n•22m ago•0 comments