frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Open in hackernews

Show HN: Understudy – Teach a desktop agent by demonstrating a task once

https://github.com/understudy-ai/understudy
36•bayes-song•1h ago
I built Understudy because a lot of real work still spans native desktop apps, browser tabs, terminals, and chat tools. Most current agents live in only one of those surfaces.

Understudy is a local-first desktop agent runtime that can operate GUI apps, browsers, shell tools, files, and messaging in one session. The part I'm most interested in feedback on is teach-by-demonstration: you do a task once, the agent records screen video + semantic events, extracts the intent rather than coordinates, and turns it into a reusable skill.

Demo video: https://www.youtube.com/watch?v=3d5cRGnlb_0

In the demo I teach it: Google Image search -> download a photo -> remove background in Pixelmator Pro -> export -> send via Telegram. Then I ask it to do the same for Elon Musk. The replay isn't a brittle macro: the published skill stores intent steps, route options, and GUI hints only as a fallback. In this example it can also prefer faster routes when they are available instead of repeating every GUI step.

Current state: macOS only. Layers 1-2 are working today; Layers 3-4 are partial and still early.

    npm install -g @understudy-ai/understudy
    understudy wizard
GitHub: https://github.com/understudy-ai/understudy

Happy to answer questions about the architecture, teach-by-demonstration, or the limits of the current implementation.

Comments

sukhdeepprashut•1h ago
2026 and we still pretend to not understand how llms work huh
wuweiaxin•55m ago
The demonstration-based approach is interesting for the handoff problem. The hardest part of agentic automation isnt the first run -- its making the agent robust to the cases the demonstrator never showed it. How do you handle edge cases or failures mid-task? Does it fall back to asking the user, or does it have some recovery heuristic? Asking because we found that the failure mode surface matters more than happy-path coverage when you actually deploy these in production.
bayes-song•35m ago
That’s exactly the hard part, and I agree it matters more than the happy path.

A few concrete things we do today:

1. It’s fully agentic rather than a fixed replay script. The model is prompted to treat GUI as one route among several, to prefer simpler / more reliable routes when available, and to switch routes or replan after repeated failures instead of brute-forcing the same path. In practice, we’ve also seen cases where, after GUI interaction becomes unreliable, the agent pivots to macOS-native scripting / AppleScript-style operations. I wouldn’t overclaim that path though: it works much better on native macOS surfaces than on arbitrary third-party apps.

2. GUI grounding has an explicit validation-and-retry path. Each action is grounded from a fresh screenshot, not stored coordinates. In the higher-risk path, the runtime does prediction, optional refinement, a simulated action overlay, and then validation; if validation rejects the candidate, that rejection feeds the next retry round. And if the target still can’t be grounded confidently, the runtime returns a structured `not_found` rather than pretending success.

3. The taught artifact has some built-in generalization. What gets published is not a coordinate recording but a three-layer abstraction: intent-level procedure, route options, and GUI replay hints as a last resort. The execution policy is adaptive by default, so the demonstration is evidence for the task, not the only valid tool sequence.

In practice, when things go wrong today, the system often gets much slower: it re-grounds, retries, and sometimes replans quite aggressively, and we definitely can’t guarantee that it will always recover to the correct end state. That’s also exactly the motivation for Layer 3 in the design: when the system does find a route / grounding pattern / recovery path that works, we want to remember that and reuse it later instead of rediscovering it from scratch every time.

dec0dedab0de•15m ago
What if you had it ask for another demonstration when things are different? or if it's different and taking more than X amount of time to figure out. Like an actual understudy would.
bayes-song•7m ago
That sounds like a good idea. During the use of a skill, if the agent finds something unclear, it could proactively ask the user for clarification and update the skill accordingly. This seems like a very worthwhile direction to explore.

In the current system, I have implemented a periodic sweep over all sessions to identify completed tasks, cluster those tasks, and summarize the different solution paths within each cluster to extract a common path and proactively add it as a new skill. However, so far this process only adds new skills and does not update existing ones. Updating skills based on this feedback loop seems like something worth pursuing.

abraxas•37m ago
One more tool targeting OSX only. That platform is overserved with desktop agents already while others are underserved, especially Linux.
bayes-song•32m ago
Fair point that Linux is underserved.

My own view is that the bigger long-term opportunity is actually Windows, simply because more desktop software and more professional workflows still live there. macOS-first here is mostly an implementation / iteration choice, not the thesis.

renewiltord•31m ago
That's mostly because Mac OS users make tools that solve their problems and Linux users go online to complain that no one has solved their problem but that if they did they'd want it to be free.
jedreckoning•19m ago
cool idea. good idea doing a demo as well.

Malus – Clean Room as a Service

https://malus.sh
639•microflash•5h ago•252 comments

Bubble Sorted Amen Break

https://parametricavocado.itch.io/amen-sorting
100•eieio•1h ago•42 comments

Reversing memory loss via gut-brain communication

https://med.stanford.edu/news/all-news/2026/03/gut-brain-cognitive-decline.html
77•mustaphah•2h ago•13 comments

ATMs didn't kill bank teller jobs, but the iPhone did

https://davidoks.blog/p/why-the-atm-didnt-kill-bank-teller
165•colinprince•4h ago•213 comments

Show HN: OneCLI – Vault for AI Agents in Rust

https://github.com/onecli/onecli
59•guyb3•2h ago•25 comments

The Met Releases High-Def 3D Scans of 140 Famous Art Objects

https://www.openculture.com/2026/03/the-met-releases-high-definition-3d-scans-of-140-famous-art-o...
114•coloneltcb•3h ago•25 comments

Show HN: Aurion OS – A 32-bit GUI operating system written from scratch in C

https://github.com/Luka12-dev/AurionOS
14•Luka12-dev•28m ago•2 comments

Show HN: Understudy – Teach a desktop agent by demonstrating a task once

https://github.com/understudy-ai/understudy
36•bayes-song•1h ago•10 comments

US banks' exposure to private credit hits $300B (2025)

https://alternativecreditinvestor.com/2025/10/22/us-banks-exposure-to-private-credit-hits-300bn/
191•JumpCrisscross•6h ago•121 comments

Converge (YC S23) Is Hiring a Founding Platform Engineer (NYC, Onsite)

https://www.runconverge.com/careers/founding-platform-engineer
1•thomashlvt•2h ago

Kotlin creator's new language: a formal way to talk to LLMs instead of English

https://codespeak.dev/
208•souvlakee•4h ago•169 comments

Asia rolls out 4-day weeks, WFH to solve fuel crisis caused by Iran war

https://fortune.com/2026/03/11/iran-war-fuel-crisis-asia-work-from-home-closed-schools-price-caps/
271•speckx•3h ago•173 comments

The Cost of Indirection in Rust

https://blog.sebastiansastre.co/posts/cost-of-indirection-in-rust/
53•sebastianconcpt•3d ago•13 comments

Dolphin Progress Release 2603

https://dolphin-emu.org/blog/2026/03/12/dolphin-progress-report-release-2603/
239•BitPirate•9h ago•38 comments

Show HN: Axe – A 12MB binary that replaces your AI framework

https://github.com/jrswab/axe
70•jrswab•5h ago•58 comments

Show HN: Rudel – Claude Code Session Analytics

https://github.com/obsessiondb/rudel
107•keks0r•5h ago•64 comments

Full Spectrum and Infrared Photography

https://timstr.website/blog/fullspectrumphotography.html
23•alter_igel•4d ago•3 comments

Italian prosecutors seek trial for Amazon, 4 execs in alleged $1.4B tax evasion

https://www.reuters.com/world/italian-prosecutors-seek-trial-amazon-four-execs-over-alleged-14-bl...
149•amarcheschi•3h ago•29 comments

Show HN: Web-based ANSI art viewer

https://sure.is/ansi/
13•lubujackson•2d ago•3 comments

WolfIP: Lightweight TCP/IP stack with no dynamic memory allocations

https://github.com/wolfssl/wolfip
27•789c789c789c•3h ago•2 comments

Scrt: A CLI secret manager for developers, sysadmins and DevOps

https://github.com/loderunner/scrt
7•Olshansky•1h ago•7 comments

Apple's MacBook Neo makes repairs easier and cheaper than other MacBooks

https://arstechnica.com/gadgets/2026/03/more-modular-design-makes-macbook-neo-easier-to-fix-than-...
73•GeekyBear•1h ago•27 comments

Contextual commits – An open standard for capturing the why in Git history

https://vidimitrov.substack.com/p/contextual-commits-an-open-standard
9•vidimitrov•1h ago•1 comments

Claude now creates interactive charts, diagrams and visualizations

https://claude.com/blog/claude-builds-visuals
108•adocomplete•3h ago•54 comments

Long Overlooked as Crucial to Life, Fungi Start to Get Their Due

https://e360.yale.edu/features/fungi-kingdom
40•speckx•5h ago•2 comments

DDR4 Sdram – Initialization, Training and Calibration

https://www.systemverilog.io/design/ddr4-initialization-and-calibration/
12•todsacerdoti•2d ago•0 comments

Avoiding Trigonometry (2013)

https://iquilezles.org/articles/noacos/
187•WithinReason•9h ago•53 comments

3D-Knitting: The Ultimate Guide

https://www.oliver-charles.com/pages/3d-knitting
199•ChadNauseam•10h ago•73 comments

The Road Not Taken: A World Where IPv4 Evolved

https://owl.billpg.com/ipv4x/
23•billpg•3h ago•25 comments

Emacs internals: Tagged pointers vs. C++ std:variant and LLVM (Part 3)

https://thecloudlet.github.io/blog/project/emacs-03/
51•thecloudlet•6h ago•24 comments