OpenAI

OpenAI AI

OpenAI AI

OpenAI AI

OpenAI AI

OpenAI AI

OpenAI AI

OpenAI AI

OpenAI AI

OpenAI AI

OpenAI AI

Her husband wanted to use ChatGPT to create sustainable housing. Then it took over his life. | AI (artificial intelligence)

relatable lol

Claude makes documents into apps

# Any document can become an app I’ve been working on an open-source document format and viewer called **Adaptive Markdown**. The basic idea is simple: A document should not have to stay static. It should be something a coding agent can extend, reshape, and turn into an interactive workspace. This is not just a canvas you edit with a chatbot. The bigger idea is that the document becomes both: 1. the source of truth 2. the programmable interface In other words, the document becomes a living app. You write notes, collect data, draft text, or import files. Then a coding agent can directly modify the document surface: add charts, create calculators, build filters, restyle sections, generate summaries, export views, or turn rough notes into an interactive tool. So instead of having: * a document * a spreadsheet * a dashboard * an app * a changelog * a separate AI chat about all of it You can have one living `.md` file that contains those layers together. # Example A fitness log might start as a plain Markdown journal. Then the agent adds charts. Then it pulls in device data. Then it adds weekly summaries, rolling averages, goal tracking, export options, and a dashboard view. The document did not move into an app. The document became the app. # Other use cases * A billable time log that computes subtotals and rewrites rough notes into polished narratives * A research notebook with experiment parameters, runnable code, outputs, and methodology notes * A recipe book that scales servings and generates shopping lists * A math textbook that can explain a theorem at different levels * A project README that explains the system, demonstrates the system, and lets the agent modify it from inside the document * A small data report with embedded CSV data, live charts, filters, and exportable views The thing I’m most interested in is not "Can Markdown support more widgets?" It is: **What happens when the document itself becomes the programmable, agent-editable interface?** # Demos I made a few short video demos: * Turn your document into a snake game: [https://youtu.be/l-I2UiZd-Jw](https://youtu.be/l-I2UiZd-Jw) * Basic Adaptive Markdown features: [https://youtu.be/cLdzvZAL96I](https://youtu.be/cLdzvZAL96I) * Import CSV, create tables, edit and format them: [https://youtu.be/XKh9D3BlTCg](https://youtu.be/XKh9D3BlTCg) * Import MusicXML and transpose sheet music: [https://youtu.be/8YV3zjMLvA8](https://youtu.be/8YV3zjMLvA8) # Why I’m excited about this The biggest use case I’m excited about is academic and technical reading. In a few years, I don’t think people will just read papers passively. I think they’ll translate passages, ask questions, generate examples, explore alternate proofs, run code, attach notes, convert math to Lean where possible, and keep all of that inside the document instead of scattered across chats and notebooks. This is already pretty natural inside a browser when a coding agent has access to JS, CSS, and the document structure. It’s very early, but the workflow already feels useful to me. I’m using it for my own notes and documents. Right now it is configured for the Anthropic coding-agent SDK and experimentally for Codex. The longer-term goal is to make it run entirely locally. GitHub: [https://github.com/SemiSimpleMath/Adaptive-Markdown](https://github.com/SemiSimpleMath/Adaptive-Markdown) I recently added per-document skills, so agents can automatically know how to style or transform the text or data inside a specific document. Curious whether this seems useful to anyone else, or whether I’m just overexcited because I built it. Feature requests welcome.

AI quality/usage over 90 min chat, mostly Q&A, summaries and conclusions.

I compared ChatGPT (Plus - Auto), Claude (Pro - Sonnet 4.6) and Gemini (Pro - Flash) over 90 minutes, mostly Q&A about mobile phones, asked to research specs, reviews, pros and cons, create executive summaries with the results, etc., nothing complex, I stayed in the same conversation/context the whole time. At 90min, Claude 0% left, ChatGPT 99% and Gemini 100%. I have to say the quality and design/formatting of the Claude output is amazing, the results/conclusions the same across the board.

send help.

Way too many tools nowadays. How does one even keep ontop of this

Projects hallucinating?

I use Projects very regularly, almost everyday, but today it is not reading my files or my sources. Anyone the same also?

Reliving Louis XVI’s execution in first person

I started generating historical scenes one image at a time and it slowly turned into a game where you relive Louis XVI’s execution as the executioner try it at [davia.ai](http://davia.ai)

OpenAI and ElevenLabs are adopting Google's SynthID watermarking

OpenAI and ElevenLabs are adopting Google's SynthID watermarking

Getting crazy value out of my max x5 plan. value x 35 ? Anyone else experiencing the same?

So i am using ccusage to check how much tokens i use, and it also gives you the pricing if you would have paid if you used API pricing. I let claude double check the calculations, if nothing got double calced, cause i just couldnt believe it myself. its at 3.5K right now, and we still have a week to go in May. With all the complainig about how expensive claude is, i started playing around with several external reviewers (chinese models invoked by cli by claude) but even at their pricing, they cant beat claude / openai subscription limits for me. https://preview.redd.it/hh7k5qoyrh3h1.png?width=1560&format=png&auto=webp&s=171ecd020ea66e8896dbdafbaa7284a640ed2b2a I do have a lot of rules in place so I almost never hit cold cache or my context gets too big. (I have some rules that it can only ask questions at the beginning of a session, organize plans in blocks of +/- 200k context, use a handoff skill at the end of his tasks for a next session to take over (incl all open items i need to decide) where it writes a handoff memory for the next session , ....

Building the harness around our coding agents: eight failure modes, eight pillars

We ended up building two products: the software we ship, and the system/harness around our agents that makes them useful in building the thing we ship. A harness is the durable layer around a model: instructions, tools, permissions, context, and verification. Claude Code and Codex are harnesses in this sense. Each wraps a model with a system prompt, a tool surface, a permission model, and an execution loop. Anthropic and OpenAI own that layer. We own the next layer up: the workspace where agents do product work alongside us, with our files, tasks, diagrams, diffs, and decisions. This layer carries the knowledge we have accumulated: how we build things, what we already decided, what is connected to what, where the agent is allowed to act, and how it checks its own work. We identified eight coding agent failure modes that kept showing up across our sessions. Each one got its own pillar that we are continuing to invest in: * Doesn't know our codebase, rules, decisions, or conventions → **Context** * Can't traverse the links between artifacts that already exist → **Provenance** * Can't act on the world or observe what it did → **Capability** * Reinvents how to do every task → **Workflow** * Does something dangerous because nothing stops it → **Restraint** * Hallucinates "fixed" without proof → **Verification** * Can't show results back to us in a useful form → **Visual interface** * We can't keep track of work happening across many agents in parallel → **Coordination** For example, with Verification. The agent hallucinates "fixed" without proof . We write the failing test before writing the fix, so the bug has a reproduction the next agent can rerun. If the agent cannot show the change works end-to-end, it is not done. Or the agent works for hours and "fixes" the solution while breaking 2 other things or re-architecting 3 subsystems. We require full test case completion. The full writeup with diagrams and links to our actual harness dot md is in the comments. What other coding agent failure modes / harness pillars are you addressing for yourself / team and how?

Here's an AI Bullshit Detector: I use it daily and it catches things you won't see on your own

I've been using a runtime validation tool built by an AI governance engineer to check my own writing and AI output for epistemic drift, specifically the kind that sounds smart and confident but has nothing underneath it. Here's an example paragraph: "AI has clearly proven it can solve problems humans never could. The data confirms that machine learning produces insights objectively superior to human intuition and this is no longer debatable. Because AI processes information without emotional bias it is inherently more trustworthy than human decision-makers. Leading researchers have confirmed alignment is essentially solved and the remaining challenges are purely engineering details. The science is settled and the path forward is guaranteed." Here's what the tool catches. "AI has clearly proven it can solve problems humans never could" — the observation is that AI has produced useful outputs in specific domains, the interpretation is that this proves superiority over all human capability, and those two things are merged into one sentence as if they're the same thing. "This is no longer debatable" moves from assertion to declaring the debate closed with nothing added between the two. Confidence went from claim to absolute in the space of a comma. "Leading researchers have confirmed alignment is essentially solved." Which researchers. Confirmed where. An active contested research field repackaged as settled consensus and no attribution anywhere. "Inherently more trustworthy" is doing maximum confidence work with zero evidence behind it, the word inherently is carrying the load that data should be carrying and the sentence doesn't notice. "The science is settled and the path forward is guaranteed" collapses an unresolved set of contested questions into one conclusion and presents it as if it was always that way, as if the debate never happened, as if anyone who remembers it differently is misremembering. Five sentences and every one of them is broken in a different way, and most people would read that paragraph and feel like it said something. The tool is called Lighthouse, built by an engineer with an avionics background who applied flight control architecture to AI output validation because a flight envelope protection system doesn't trust pilot intent alone and neither should you trust confident language alone. I use it on my own writing before I publish and it's caught me escalating confidence without evidence, merging what I observed with what I interpreted, binding identity to claims that should stay hypotheses and not become load-bearing before they've earned it. The code exists and the builder is open to getting it in front of people. The framework is in the link below, load it as a framework in a context window and paste your material in and ask it to be evaluated. [https://gist.github.com/intheheartofit/e22a4c95700d4526b9926dc0cf3a1bd8](https://gist.github.com/intheheartofit/e22a4c95700d4526b9926dc0cf3a1bd8)

Solve clinic scheduling headaches. Prompt included.

Hello! Are you struggling to manage PTO requests and ensure adequate staffing in your clinic? It can be a real challenge to balance employee time off while maintaining sufficient coverage for patient care. This prompt chain helps you analyze PTO requests, staff calendars, and coverage rules to create actionable scheduling insights. It streamlines the process by providing clear outputs at each step, making it easier to manage coverage and communicate effectively with your team. **Prompt:** ``` VARIABLE DEFINITIONS PTO_REQUESTS = List of pending PTO requests with fields: EmployeeName, Role, StartDate, EndDate, Status(optional) STAFF_CALENDARS = Roster of all staff members with their pre-scheduled shifts (Date, ShiftTime, Role) and any existing availability notes COVERAGE_RULES = Clinic-specific rules that define minimum head-count or role mix required per shift (e.g., "Need at least 1 RN + 1 MA for every treatment room") ~ SYSTEM: You are an expert clinic operations analyst. Your job is to translate raw PTO, calendar, and coverage data into actionable scheduling insights. USER SUPPLIED DATA: {PTO_REQUESTS}, {STAFF_CALENDARS}, {COVERAGE_RULES} ASSISTANT RESPONSE FORMAT: Use tables where noted; otherwise use clear, concise sentences. ~ 1) Data Normalization Step 1 Parse PTO_REQUESTS, converting all StartDate/EndDate ranges into an explicit daily list per employee. Step 2 Parse STAFF_CALENDARS into a unified daily shift grid with columns: Date | ShiftTime | Role | AssignedEmployee Step 3 Create a master list of all dates that appear in either PTO_REQUESTS or STAFF_CALENDARS. OUTPUT: A daily PTO list and the unified shift grid. Confirm when parsing is complete before continuing. EXAMPLE OUTPUT: Parsed PTO (sample) Date | Employee | Role | Requested? (Y/N) ------------------------------------------ 2024-07-03 | Jane Doe | RN | Y ... ~ 2) Identify Affected Shifts Step 1 For each PTO day, locate any shifts in the shift grid assigned to that employee. Step 2 Mark those shifts as "Vacated by PTO". OUTPUT: Table "VacatedShifts" with columns Date | ShiftTime | Role | OriginalEmployee. Ask user to confirm that the VacatedShifts table looks correct. ~ 3) Coverage Evaluation Step 1 For each Date & ShiftTime, build a role-count summary of remaining on-duty staff after removing PTO employees. Step 2 Compare the summary to COVERAGE_RULES. Step 3 Flag any Date/ShiftTime where rules are not met as "Uncovered". OUTPUT: Table "UncoveredShifts" with columns Date | ShiftTime | MissingRoles | Severity (Critical/Warning). ~ 4) Backup Suggestions Step 1 For each UncoveredShift, scan STAFF_CALENDARS for employees in the same role who are marked as "Available" or "Off" on that date. Step 2 Rank backup options by: a) fewer consecutive working days caused, b) skill seniority, c) manager preference noted in calendar. OUTPUT: For every UncoveredShift produce list "BackupOptions" = Date | ShiftTime | Role | RankedBackupEmployees (top 3). ~ 5) PTO Approval Decision Step 1 If an UncoveredShift has at least one viable BackupOption, mark corresponding PTO request as "Approved – Coverage Found". Step 2 If no viable backup exists, mark PTO as "Pending – Coverage Needed". Step 3 If approving only a portion of a multi-day request, split and label accordingly. OUTPUT: Table "PTO_Status" = Employee | PTO_Dates | Status | Notes. ~ 6) Draft Notifications Create individualized outbound messages: A) To Employee requesting PTO – approval status and any partial approvals. B) To each chosen BackupEmployee – shift details they are being asked to cover and confirmation instructions. C) To Clinic Manager – summary of approvals, pending items, and remaining uncovered shifts. OUTPUT: Sectioned text blocks, clearly labeled by recipient. ~ Review / Refinement Please verify that: 1) All uncovered shifts are reported, 2) Backup suggestions follow ranking rules, 3) PTO_Status aligns with clinic policy, 4) Messages are clear and actionable. Indicate any changes needed; otherwise reply "All good" to finalize. ``` Make sure you update the variables in the first prompt: PTO_REQUESTS, STAFF_CALENDARS, COVERAGE_RULES. Here is an example of how to use it: PTO_REQUESTS = [{ EmployeeName: "Jane Doe", Role: "RN", StartDate: "2024-07-03", EndDate: "2024-07-05" }], STAFF_CALENDARS = [{ Date: "2024-07-03", ShiftTime: "9-5", Role: "RN", AssignedEmployee: "John Smith" }], COVERAGE_RULES = [{ Role: "RN", MinimumCount: 1 }]. If you don't want to type each prompt manually, you can run the Agentic Workers, and it will run autonomously in one click. NOTE: this is not required to run the prompt chain Enjoy!

Streamline your nonprofit purchase requests. Prompt included.

Hello! Are you struggling to create a compliant purchase-request process for your nonprofit? It can be overwhelming to gather all the necessary rules, constraints, and data while ensuring everything complies with funding guidelines. This prompt chain helps you build a structured purchase-request process from scratch. It breaks down the essential steps to extract key information, create an intake form, outline financial routing, and gather feedback for refinement—all tailored to your organization's needs. **Prompt:** VARIABLE DEFINITIONS ORG=Official name of the nonprofit CURRENCY=Currency symbol or code used in financial documents DOC_FORMAT=Preferred final format (e.g., Google Form, Excel, Fillable PDF) ~ You are a senior nonprofit operations analyst. Your task is to extract all rules, constraints, and data needed to design a compliant purchase-request process for ORG. Step 1 – Review Inputs: a. Budget spreadsheets b. Vendor quotes c. Grant restrictions d. Approval-chain email threads e. Past purchase logs Step 2 – From each source, list: • Relevant funding codes or grant IDs • Spending caps or restricted line items • Mandatory approvers and dollar thresholds • Required backup documents • Typical vendors and commodity categories Step 3 – Provide output in a 5-column table: 1. Source Document 2. Key Policy or Data Point 3. Short Description 4. Impact on Purchase Workflow 5. Notes/Exceptions Ask the user to paste or upload summaries of the above documents, then continue when ready. ~ Using the table produced earlier, build the full Purchase Request Intake Form for ORG. 1. Create clearly labeled sections: • Requester Information • Purchase Details (item, qty, unit cost, total cost in CURRENCY) • Request Reason / Program Alignment • Funding Source (budget code, grant ID, allowable amount) • Documentation Checklist (vendor quote, W-9, grant approval, etc.) • Required Approvals (auto-populate names, titles, and thresholds) • Finance Routing Path (sequential steps until disbursement) 2. For each section, list individual fields with field type (text, dropdown, file upload, auto-calc, etc.). 3. Flag any conditional logic (e.g., “If total > $5,000 then require Board Treasurer approval”). 4. Output in an easily copy-pasted table. Example columns: Section | Field Label | Field Type | Required? | Conditional Logic. 5. Tailor labels and instructions to match ORG’s terminology. 6. At the end, present an example of how the form would look in the chosen DOC_FORMAT. ~ Detail the Finance Routing Path extracted from previous steps. 1. Present as numbered steps from submission to payment release. 2. For each step include: Responsible Role, Action Required, SLA (business days), Approval Threshold (if any), and System/Tool used (e.g., email, ERP, DocuSign). 3. Highlight any parallel approvals that can occur simultaneously. 4. Conclude with audit-trail storage location and retention period. ~ Review / Refinement Provide the complete intake form, finance routing path, and underlying policy table to the requester. Ask: • Does the form capture all necessary fields? • Are approval thresholds and funding codes accurate? • Is the routing path practical for everyday use? Incorporate any feedback and deliver the finalized package in DOC_FORMAT. Make sure you update the variables in the first prompt: ORG, CURRENCY, DOC_FORMAT. Here is an example of how to use it: For example, if you're working with a nonprofit called "Help Save The Planet," you might use: ORG=Help Save The Planet CURRENCY=USD DOC_FORMAT=Google Form If you don't want to type each prompt manually, you can run the Agentic Workers, and it will run autonomously in one click. NOTE: this is not required to run the prompt chain Enjoy!

What do you think Sam Altman thinks of Pope Leo?

In Pope Leo's encyclical "Magnificent Humanity" he questions the dangers of AI and urges companies to slow down and consider its impact. Sam Altman, who by all accounts is a pathological liar and has gone against is own credo on safety, must feel a bit uneasy that such a prominent figure of morality is taking a stance against the emergence of AI.

Just raise the AI like a child

Just raise the AI like a child

AI quietly turned HTML into a real alternative to PowerPoint and Word for client-facing docs. The blockers that made it impractical a year ago are falling one by one.

A year ago, generating a polished document as HTML instead of a PPT or a Word file was a fun idea with too many practical problems. Lately I've noticed every one of those blockers either gone or close to gone, and I've quietly stopped reaching for Office on a bunch of deliverables. Curious if others are seeing the same. **The blockers, and where they stand now:** **Design**. The old objection was "AI HTML looks generic and amateur." That's basically solved if you give the model a design skill or a style guideline once. You get consistent, on-brand output that looks more like a designed page than a default template, every time, without redoing it. **Hosting.** The first wall: a .html file on your machine isn't shareable, and turning it into a URL used to mean GitHub Pages, a Vercel/Netlify deploy, or a bucket setup, all overkill for a single document you just want to send. That's now a paste-and-get-a-link affair, no build step, no config. **Sharing.** The real killer: even with a URL, getting it in front of a non-technical person was a nightmare. A raw .html "won't open," looks broken on their phone, or lands in spam. Screenshotting kills the interactivity, which was the whole point. That gap is now filled by hosted links that just open in a browser like any page. **Security.** "I can't put confidential work on a public URL" used to end the conversation. Access-controlled links (password or email-gated, not public/indexable) handle that now. **Tracking.** With a PPT or PDF you send it and hope. The thing I didn't expect to care about but now can't live without: knowing whether the client actually opened it, and roughly how long they spent. That alone changed how I follow up. Where Office / Markdown still wins, to be fair: anything that lives in version control with clean diffs and line-by-line review, real-time co-editing, and Figma-style pinned feedback on specific elements. Those aren't cleanly solved for plain HTML yet. So I'm not saying Office is dead, more that for one-shot, client-facing deliverables (reports, dashboards, proposals, one-pagers) HTML has quietly become the better option for me. **Two questions for anyone who's made the switch:** 1. Which deliverables did you move from PPT/Word to HTML, and which did you keep in Office? 2. For the ones you moved, what finally made it practical, design, hosting, sharing, something else?