Andrej Karpathy built a simple automation pipeline for AI agents to optimize training in 5-minute increments.

an ARA-native review system that automates objective checks so human reviewers can focus on significance, novelty, and taste.

The agent has gone from literal zero, no Cloudflare account at all, without any preconfigured [Agent Skills](https://github.com/cloudflare/skills) or [MCP server](https://blog.cloudflare.com/code-mode-mcp/), to having: * Provisioned a new Cloudflare account * Obtained an API token * Purchased a domain * Deployed an app to production

Humans can be in the loop to grant permission and must accept Cloudflare's terms of service, but no human steps are otherwise required from start to finish.

Fugu models achieve superior performance by dynamically coordinating and orchestrating a diverse pool of powerful models.

Capture sequences and replay them as stable automations.

Map tokens, cookies, sessions, and multi-step sequences automatically.

Kampala lets you reverse engineer anything including websites, mobile apps, and desktop apps instantly.

We want email agent tooling to be composable and reusable. Rather than every team rebuilding the same inbound-classify-reply pipeline, start with this reference application.

Claude packages everything into a handoff bundle that you can pass to Claude Code with a single instruction.

Workspace agents can gather context from the right systems, follow team processes, ask for approval when needed, and keep work moving across tools.

The benefit of using LLMs is that a lot of the initial context gathering can be done in an automated way. An emphasis of focus should be on high signal context – for example, looking through past query history can be high signal in determining the most referenced tables and most common joins, and data modeling solutions like dbt or LookML can provide clear definitions for business metrics.

The most notable improvement comes from CloningQA, which requires end-to-end design of DNA and enzyme reagents for molecular cloning protocols.

Auto-generate screen reader specs from UI designs

The system is designed to handle multi-step workflows like booking trips, clearing inboxes, or running research without constant input, bringing it closer to emerging agent platforms from OpenAI and Anthropic.

The integration also connects to Upwork's AI agent Uma, which helps automate parts of the hiring and execution process once a project is underway.

Ovren puts AI frontend and backend engineers on it - they work inside your real codebase, execute scoped tasks, and deliver reviewable code updates.

M2.7 shows significant improvement in complex editing capabilities for Office Suite (Excel/PPT/Word), better handling multi-turn modifications and high-fidelity edits.

building and shipping an internal beta product with zero manually written code

The AI toolkit for building and maintaining browser automations

Add cloud browser provider system (Kernel + Browserbase) - Add cloud browser providers spec (Kernel + Browserbase) - phase 1: add provider metadata to session state schema

Add dev-tools package with wt worktree manager CLI - New packages/dev-tools with standalone wt CLI for git worktree management - Commands: wt new, wt scratch, wt prune - Uses Vertex AI (gemini-2.5-flash) for branch name generation via gcloud ADC

It has replaced existing tasks for 27% of employed AI work users and created new ones for 21%.

27% said AI has replaced some of their existing tasks (task automation). An example of task automation could be an employee using AI to summarize a document they would ordinarily read themselves.

A routine is a saved Claude Code configuration: a prompt, one or more repositories, and a set of connectors, packaged once and run automatically.

Routines execute on Anthropic-managed cloud infrastructure, so they keep working when your laptop is closed.

Performance: dev-browser: 3m53s, $0.88, 100% success rate — beats MCP configs, Chrome extensions, 'browser skill' stacks.

One Agent can now: open X (Twitter), scroll the feed, extract tweets, return clean JSON. No plugins. No extensions. No orchestration.

GitHub Actions has grown from 500M minutes/week in 2023 to 1B minutes/week in 2025, and now 2.1B minutes so far this week

Open Loop + Infinite Demand = Creative Amplifiers. Content creation & marketing strategy. AI can generate a thousand ad variations or blog posts.

Closed Loop + Finite Demand = Efficiency Plays. AI bookkeeping categorizes transactions, reconciles accounts, files returns. Deterministic rules applied to numbers.

You get your most tenured engineer on-call all the time

70% of alerts resolved in under 5 minutes

Austin built the whole pipeline from his Claude Code terminal using the Notion API. He brain-dumped the desired outcome using Monologue, let Claude Code create the database and data pipeline, and pasted the generated instructions into the Notion custom agent setup.

ChatGPT for Excel can connect to apps. Apps from your ChatGPT account can be used in ChatGPT for Excel, subject to your plan, workspace admin settings, app availability, user permissions, and data-source entitlements.

if AI can do only 50 percent of a human's tasks, the importance of the non-automatable tasks likely goes up since they become the bottlenecks, increasing their relative value.

accounting and auditing showing nearly a 20 percent jump on GDPval and even domains like police / detective work showing a nearly 30 percent improvement.

Support teams are high volume and high turnover, and thus need to train new reps in a fast and standardized way. To do so, they have clearly articulated standard operating procedures (SOPs) that guide the work of each rep. These SOPs create clear rules and guidelines that AI agents can model themselves off of.

The difference between analysis automation—inference—anddecision automation is that in the latter the system must make implicit or explicit assump-tions about costs and values inherent in all decisions.

Examples of decisionautomation include route planning and adaptation, such as to avoid bad weather, and systems pro-viding medical diagnosis support.

Decision automation means deciding and selecting appropriate actions among alternatives.This type of automation corresponds to the third human information processing state, decision-making, which the machine is either augmenting or replacing altogether.

An example of low-level automation is the extrapolation or prediction of data over time,such as a system predicting a trend for the output of an industrial plant based on historical sensordata. An example of moderate- to high-level automation is a system integrating multiple sources orinput variables. This could be a display with emergent perceptual features, such as an optical see-through display with a landing strip intended to assist a pilot in landing an aircraft. An exampleof high-level automation is a context-dependent summary of data.

Analysis automation refers to the automation of information analysis and involves inferentialprocesses. It corresponds to the second human information processing state: perception/workingmemory.

An example of low-level automation is assistance in sensor adjustment, such as a system mechanically moving a radarsensor to lock on a detected target. An example of moderate automation is a system organizinginformation according to criteria such as a priority list or highlighting information based on staticor dynamic criteria. This could be, for example, a display highlighting the rate of change in somevariable of interest. This could be indicated by increasing the intensity of some pixels more rapidlythan others in the display.

Acquisition automation corresponds to the first human information processing stage, sensoryprocessing, and it is realized by the system sensing and registering input data.

Types of automation: The types of automation can be understood by viewing a human operatoras a simple four-stage model of human information processing:1. Sensory processing2. Perception and working memory3. Decision-making4. Response selection.

The framework uses two sets of evaluation criteria to help designers determine the appropri-ate type and level of automation for each application. The primary evaluation criteria concernthe impact of the chosen types and levels of automation on human performance. The secondaryevaluation criteria include automation reliability and the cost of decisions or outcomes.

For thisreason, designers can benefit from frameworks that support system design that involves automa-tion. We now discuss one such framework: the types and levels of automation framework [639].Downloaded from https://academic.oup.com/book/60808 by Helsinki University Library user on 01 December 2025

In all cases, getting the right balance betweenautomation and human control is crucial.

Throughout the history of automation, a central challenge in HCI has been posed by the com-plexity of autonomous systems. Users must be able to understand and control them. They need tofind and integrate information from dynamic and different sources. They need to understand howto delegate tasks, supervise their execution, and intervene if needed.

Automation refers to technology that assists users by performing a task or a subtask on their behalf.

Suchtechniques provide different levels of automation.

Historian Mar Hicks documents this in Programmed Inequality: when work is coded as “women’s work,” it becomes a target for automation.

METR conclude that “the length of tasks AI can do is doubling every 7 months”. I’m not convinced that pattern will continue to hold, but it’s an eye-catching way of illustrating current trends in agent capabilities.

The chart shows tasks that take humans up to 5 hours, and plots the evolution of models that can achieve the same goals working independently. As you can see, 2025 saw some enormous leaps forward here with GPT-5, GPT-5.1 Codex Max and Claude Opus 4.5 able to perform tasks that take humans multiple hours—2024’s best models tapped out at under 30 minutes.

f you define agents as LLM systems that can perform useful work via tool calls over multiple steps then agents are here and they are proving to be extraordinarily useful. The two breakout categories for agents have been for coding and for search.

Home security company Vivint, which uses Agentforce to handle customer support for 2.5 million customers, experienced these reliability problems firsthand. Despite providing clear instructions to send satisfaction surveys after each customer interaction, The Information reported that Agentforce sometimes failed to send surveys for unexplained reasons. Vivint worked with Salesforce to implement "deterministic triggers" to ensure consistent survey delivery.

All of us were more confident about large language models a year ago," Parulekar stated, revealing the company's strategic shift away from generative AI toward more predictable "deterministic" automation in its flagship product, Agentforce.

human experience makes science possible

During one-third of a century of experimental work, I have been operating on the philosophic premise that all thoughts and all experiences can be translated much farther than just into words and abstract thought patterns.

projectionsby

while true; do waitforsave assignment.ly; lily assignment.ly; say "done"; done

Then it automatically withdraws money you can safely save and deposits it in the Oportun account you specify—either the Oportun savings or retirement account. Funds from your Oportun accounts can be moved back to your linked account at any time.

Another simple way is to set up automatic deposits from your checking account into your savings account. Set the deposits to occur on the same day each month (like the day after your paycheck hits the account). This way, you’ll be saving a fixed amount of money regularly without even giving yourself the chance to use it for something else.

If you’ve got some technical savvy, the open-source tool Privacy Bot can send mass data deletion requests from your email address.

IFTTT (/ɪft/, an acronym of if this, then that)[3][4] is a private commercial company founded in 2011, that runs online digital automation platforms which it offers as a service.[5] Their platforms provide a visual interface for making cross-platform if statements. As of 2020[update], they have 18 million users.[2][6][7] IFTTT has partnerships with different providers of everyday services as well as using public APIs to integrate them with each other through its platform. They supply event notifications to IFTTT and execute commands that implement the responses.

The processing limit for shortcut text, etc. seems to be approximately 2.6 MB.

In those countries where manpower is not available, or some process where human efforts are limited, Automation is justified. Howeverin the countries with huge potential of young manpower automation can be limited.

We will try to add two tests for response code in order to know that our request was successful. Another test we will add for response time <  2 sec in order to understand how fast request was processed by server. If it will be executed slower then for 2 seconds, our test will fail. In this case I use 2 seconds just for example it might be greater or lower number, but 7 seconds is usually a maximum time for request execution. So in order to add tests, go to “Tests” in request section of application and add this few lines : tests["Status code is 200"] = responseCode.code === 200; tests["Response time is less than 200ms"] = responseTime < 2000; When this is done hit on Send button again and execute your first test.

Developers use tools like Postman for API test automation to create WMS system integration technologies that help you increase efficiency by automating manual processes. The WMS connects all of your warehouse locations so that you can easily access information about your inventory, orders, and shipments from any computer.

I believe the final policy shall contain robust rationale and, in the best way possible, avoids the perception of rAIcial discrimination

The industry has changed from being fragmented to consolidated

If there are any date formats you regularly use that aren’t the default for Shortcuts, those are excellent choices for creating utility shortcuts.

C’est la confusion entre mécanisation et automatisation qui nourrit l’inquiétude de vies soumises au calcul.

The beauty of FAST framework Overloaded with information → Filter Drowning in busywork → Automate Descending into chaos → Structure Doing things over and over → Templatize

Since this can involve a lot of repetitive actions on a long file, I use this javascript code along with the QuickAdd plugin, to do it for me.

“You have to assume that things can go wrong,” shared Waymo’s head of cybersecurity, Stacy Janes. “You can’t just design for this success case – you have to design for the worst case.”

“Find Notes” can now retrieve more than 26 notes at a time

Create Tab Group

For example, whereas C programmers have argued for years about where to put their brackets, and whether code should be indented with tabs or spaces, both Rust and Go eliminate such issues completely by using a standard formatting tool (gofmt for Go, rustfmt for Rust) which rewrites your code automatically using the canonical style. It’s not that this particular style is so wonderful in itself: it’s the standardisation which Rust and Go programmers appreciate.

The single-line commit trailers ^Update-Info: *(.*) on version tagged commits are used to assemble a small changelog during update, which is presented to the user. The single line can contain important information/links to relevant fixes and changes.

Danger runs after your CI, automating your team's conventions surrounding code review.

This is an Appium plugin designed to manage and create driver session on connected android devices and iOS Simulators. Why Appium Device Farm? Automatically detects connected Android, iOS Simulators and Real devices before session creation and maintains it in device pool. Dynamically allocates a free device from device pool while creating driver session. Updates the device pool with a new device during test execution. Allocates random ports for parallel execution.

One of those factors is globalization which has helped lift hundreds and millions out of poverty, most notably in China and India, but which, along with automation has also ended entire economies, accelerated global inequality, and left millions of others feeling betrayed and angry at existing political institutions.

Disappearing message apps take something humans are bad at (remembering to do a specific task at a specific time) and hand that job to a computer, which is really good at that.

In its simplest sense, automation is about each individual part performing the same repetitive steps over and over again

Oh and I automated my entire crypto portfolio.

The script is in batch with some portions of powershell. The base code is fairly simple and most of it came from Googling ".bat transfer files" followed by ".bat how to only transfer certain file types" etc. The trick was making it work with my office, knowing where to scan for new files, knowing where not to scan due to lag (seriously, if you have a folder with 200,000 .txt files that crap will severally slow down your scans. Better to move it manually and then change the script to omit that folder from future searches)

It essentially scans the on-site drive for any new files, generates hash values for them, transfers them to the Cloud, then generates hash values again for fidelity (in court you have to prove digital evidence hasn't been tampered with).

How To Become A Task Automation Hero Using Python [Guide With Examples]

/!\ If you see "Complete the captcha then press enter", switch to the Chromium window and do the captcha. ¨¨ Press Enter when you're done, the page will reload and you should see the source code of the URL you request

This code permit a manual bypass of the CloudFare captcha for FanFiction.net

Side note: When I flagged yours as a dupe during review, the review system slapped me in the face and seriously accused me of not paying attention, a ridiculous claim by itself since locating a (potential) dupe requires quite a lot of attention.

Trust Tokens is a new API to help combat fraud and distinguish bots from real humans

Trust Token is a new API to help combat fraud and distinguish bots from real humans

This poses a few problems for automation. In some environments, there may be no graphical display available, or it may be desirable to not have the browser appear at all when being controlled.

Aqara Magic Cube ZHA (51 actions)

If everyone is marked as Extended Away, activate vacation mode to automatically turn off/on lights to make it look like someone is home

Best Home Security System Apps to Learn From

Yes, autoexpect is a good tool, but it is used just to automatically create TCL-expect scripts, by watching for user. So it’s can be equal to writing expect-scripts by hand.