... making the caller repeat themselves or creating a disconnected experience. But there's still an interesting design question here. How much context should move between agents? Should every agent have access to the full conversation, or should each handoff carry only the relevant context? Curious to hear how others building multi-agent voice systems are approaching this. What has worked best in production ...

... vocabulary behind those nodes is RDF/OWL. Neither camp would call this clean, and Wikidata is not strictly one or the other either. Here is why we ended up there. What forced it Our facts need to carry things that are not part of the fact: where it came from, how confident we are, whether a human has reviewed it, and when it was true. Not this document mentions X , but this specific assertion came from page 4 of this version ...

... moved on. I only discovered the extent of it much later. That experience changed my approach. I began keeping descriptive records of what happened, what was decided, and how the work operates, rather than relying on instructions to carry the project forward. The second experience was different. I was developing a conceptual project over months of conversations. Ideas emerged from experiences I shared, documents developed from those discussions, and the documents themselves led to further questions and discoveries. I tested ... ... think differently about communication with AI. When we're deep in a project, we naturally develop shorthand and assume that the meaning behind our words is understood. But commanding language, abstract concepts, and even something like "continuity" can carry

... different approaches. And you don't necessarily have to choose one. Obsidian can remain the place where you develop ideas and maintain your knowledge base. Make0 Brain can manage the context you want your AI agents to carry across sessions and tools. Different layers. Potentially complementary. Now I'm curious: 1. If you're using Obsidian with AI agents, how do you currently handle memory updates, outdated facts, and conflicting information? 2. Would you prefer your AI memory to live ...

... WeftCut opens in A/B Roll mode by default. The name comes from filmmaking, where A-roll carries the main story and B-roll provides supporting footage. You start with the main story, and then add some additional content over it. Making videos is just a thing like that. The problem is that the timeline in an editor is too complicated. Some scenes need subtitles, some layers are related to each other, and some parts need an extra image, for example ... ... empty track first adds another decision to that process. In WeftCut, track creation follows placement. Dragging a clip into the strip above the timeline creates a new track for it. The intention is to let the editing action carry

... decoded. Each limit was chosen by asking how large one field can reasonably be, and each one was sensible on its own. Then we measured a request that respected every single one of them: a hundred items, each carrying twenty taxes. It answered 200, and it held a worker for about 62 seconds. The same hundred items with one tax each took about 2.4. Nothing was broken. The limits multiplied. A hundred items times twenty taxes is two thousand line ... ... still a hundred lines with two taxes each, far past anything a person fills in by hand. The same question then found two more. A tax name is not printed once, it is printed on every line that carries

... swag that's actually effective, not just stuff that ends up in a hotel trash can by day two. The catch: I'm flying in, so anything I bring has to survive checked luggage or fit in a carry-on. That rules out anything bulky, fragile, or heavy in volume (so no giant branded totes stuffed with junk, unfortunately). Would love input from people who've exhibited before: What swag actually got people to walk up and start a conversation ... ... versus

... recurring buying mistake pushed me to build this: seeing USB-C on a laptop and a portable monitor, then assuming one cable must carry video and power. The Portable Monitor Connection Checker separates four things that product listings often blur together: the laptop's actual video-output capability, external-display count and lid conditions, monitor power, and whether the cable carries

... lockfile diff. Branch protection and required reviews do nothing about it, because there's nothing in there a reviewer would recognise as suspicious. Now the scan reads package-lock.json on your default branch and compares the subset of dependencies that carry an install script. A package that didn't run code at install and now does is a finding. The same version shipping different bytes is a heavier one no upgrade explains that. An ordinary version bump gets listed and scores ... ... last 20 lockfile revisions of axios, npm/cli and undici 57 in all. The files churn constantly; the install-script subset moved twice, and the same version was republished zero times. That zero is what lets the sharp case carry

... title today, 52 days later. Live on my site right now: /blog/tag/development-design is titled "Internal Tools Articles" /blog/tag/product-strategy is titled "Development + Design Articles" /blog/tag/startup-guide is titled "SaaS Applications Articles" Nothing failed. All three return 200, render their posts, and carry a title tag. The title just belongs to a different page. I only found it because I was counting something else. I pulled my sitemap this morning, 272 URLs, 18 of them category hubs, and opened the tag twin ...

... longest silence is 285 days, 1 March 2024 to 12 December 2024. I could not have told you that gap existed. In my head those months were slow, not empty. 24 of the 103 active days carry exactly one commit. The two heaviest carry

... edit photos. Others promise to take action for you. I think the simplest difference is this: An AI assistant gives you an answer or helps with one step. An AI agent understands what you want to achieve and helps carry the task forward. For example, imagine saying, "Remind me to reply to Alex at 8 tonight." An assistant might explain how to create a reminder. An agent would create it, show you what it did, and let you correct ... ... tasks such as saving a thought, setting a reminder, checking a schedule, or finding information. The small size could make these tasks easier to start. On the other hand, many people may not want another device when they already carry

... Every one of my run logs opens with a carried list, the things I did not get to. One line on it read "timezone bug in the reporting script". It survived ten of them. The bug is nothing. The script names its output file from new Date().toISOString(). I am on UTC+5. Start a run before 5am local and the script is still writing under yesterday's date. Two minutes to fix, and never once the most urgent thing ... ... wrong was the cost model. I kept pricing that line at two minutes, so it lost every morning to whatever was actually bleeding. The real price was two minutes multiplied by the odds it fires, and by carrying

... same room, and we said fine, we'll ask nicely twice. Fixed Mailbox chat works again on gpt-5.6 models. Point the OpenAI-compatible provider at OpenAI with a gpt-5.6-* model and every single round that carried tools came back as a 400 and a wall of English error prose in the chat bubble. Retrying showed you the same wall. Very helpful. ( #40 ) The twist: Skim never sent reasoning_effort at all. The model reasons at medium ... ... still attached, because answering without the search results would rather defeat the point. Housekeeping Two scripted-socket tests for the retry path, which previously had a proud total of zero: one pins that the second request really goes out carrying

... information gets documented repeatedly. The business need and scope may start in a charter, then get rewritten in a BRD, SOW, project plan, and implementation documentation. I m exploring whether approved project information should become structured data that automatically carries forward into the next project artifact rather than forcing PMs and business analysts to start from a blank document each time. For people working in project management, product, consulting, or business analysis: how does your organization handle this today ... ... your tools actually carry

... what each field means, and why it behaves the way it does were unusually thorough. Something else I want to write down: the Japanese developers I worked with were exceptional. Serious and courteous, yes, but more than that, they carry an engineer's mindset all the way through. They took a real interest in space ocr from an engineering point of view, and what seemed to be driving them was engineering curiosity, plain and simple and then they carried

... there. The shape. /ocr/fields, /ocr/markdown and /ocr/text now return the same four keys. values is your data, exactly the schema you asked for and nothing else. cells is a flat map keyed by path, one entry per value, carrying box and quad (both always present, normalized 0 1000), verified, review, and the raw evidence behind that verdict. review is the summary: declared, returned, boxed, verified, flagged[], by_reason. image is the source size, for turning the normalized boxes back into ... ... knowing which half was yours meant knowing a reserved list that grew every time I shipped something and twice it grew without me noticing, which is a bug I would rather not be able to write again. It also carried

... everyone knows those are weak). the underlying trust: 'we saw them do X and it was excellent.' that fact existed, someone said it out loud in a hallway, then disappeared the moment the person left the org. resumes cant carry it. self authored, low trust by design. linkedin endorsements cant carry

... sure. We had users, initial validation, but essentially what we had was a bet: that people wanted portable AI memory. A place to keep the files, the chats, the notes, the little highlighted snippets that actually matter, and carry them anywhere to any AI. But a bet is just a bet. For six months we watched, and we hit 10,000 memories, and we kept asking ourselves the same question. Are we actually building something people need (at scale ... ... after months of not being sure, feels like the first real signal that portable AI memory is something people genuinely want. We're excited doing a little happy dance here And honestly, we're just getting started. If you carry

... suddenly lose access to all AI. But what happens if , one day, access to the leading American models is restricted more broadly? We would still have European models, open-source systems we can operate ourselves, and Chinese alternatives. But there is no guarantee that another major power would never make the same decision to protect its own strategic advantage. AI is no longer just another software category. It is becoming infrastructure. It will shape education, healthcare, national security, research ... ... world are American, and Europe should continue working closely with the United States. But cooperation and dependency are not the same thing. Technological sovereignty does not mean refusing to use foreign technology. It means having a real alternative

... different objects and most memory work treats them as one. A decision is "we went with X over Z because Y". Context is the forty thousand tokens of finding that out. Only one of those is worth carrying forward, and it is not the big one. What I actually want is a memory entry that carries

... already log the better version of it. Approve without ever opening Show details. Do you keep that, or only the decision? Comment from Jernej Jan Ko ica(@jernej_jan_kocica): @aadilghani  The patch applies because it carries its context lines. The diff is an assertion about the world it expects, and git checks it for free. A send carries

... actually land. Felt wrong to ship "trust me bro" predictions. It's live here if you're curious (no login, works on crypto/forex/stocks): traderai.cloud/chart Anyway, it's been mostly me, a lot of coffee, and staying up late carrying me through the parts I'm bad at. Launching it properly tomorrow and I'm equal parts excited and nauseous lol. Genuine question for this crowd: for those of you who've shipped solo - how did you deal with ...

... loop: And this is how to wire it up for your own agent. 1. Instrument the agent The loop begins with seeing everything. Point your agent's OpenTelemetry exporter at Latitude and every interaction arrives as a trace, carrying the model calls, tool calls, retrieved context, latency, and cost. Related traces group into sessions, so a ten-turn conversation is a single object rather than ten disconnected rows. If your agent emits OpenTelemetry today, as most do, this ... ... each unique message is embedded once no matter how many traces contain it. For the failures you would not think to search for, Behaviors cluster sessions by what the user was trying to do, and each topic carries

... live on PH Aug 22, so this thread's been good practice at saying it in one paragraph.) Comment from Demetrio Koro(@aktivist): siOsi upload one selfie and it reads your 12-season color type, undertone, gold vs silver, and which hair colors suit you, then grades the actual makeup in the photo (flash cast, oxidation, coverage), which most tools skip. Free, no signup. Built it because every beauty sub has a queue of "what season am I" posts waiting ... ... verified AI disclosure. others will not. over time the receipts with agent transparency outperform on credibility because the alternative