https://scrappylabs.ai/lab Experiments, publications, visual explainers, field notes and open source from ScrappyLabs. Built in the open. en Mon, 27 Jul 2026 06:45:28 +0000 https://scrappylabs.ai/memory-invocation memory-invocation Mon, 27 Jul 2026 12:00:00 +0000 We built a four-tier memory system for our coding agents, then instrumented it to find what was still failing. It wasn't the retriever — every "forgotten" fact came back at rank 1. The failure was invocation, and what predicts it is how the task is framed. Includes the architecture, the numbers, the audit that caught us, and a build path you can point your own agent at. agent memory architecture evaluation build in the open https://stuff.scrappylabs.ai/public/k3-review/fear-segment.html ai-fear-segment Sat, 25 Jul 2026 12:00:00 +0000 A good argument about open-weight economics, and the exact moment it turns into something else. Four tells that a technical explainer has stopped reasoning and started steering — plus a five-step checklist you can reuse on any AI threat segment. media literacy security open models https://scrappylabs.ai/mage-flow mage-flow Thu, 23 Jul 2026 12:00:00 +0000 Microsoft shipped a 4B image model that does generation and instruction editing under an MIT licence — the licence being the part that usually kills these for client work. We had it measured on our own GPU the same evening. It beat its own published speed at half the advertised memory, and it renders readable text on embroidery, blueprints and screenprints. Eighteen images, one seed, no retries — plus the attention-backend trap that cost us an hour, written down so it doesn't cost you one. image models benchmarks measured on our iron open weights https://stuff.scrappylabs.ai/public/moya/mage-flow-gallery.html mage-flow-gallery Thu, 23 Jul 2026 12:00:00 +0000 Every frame we generated while evaluating Mage-Flow: embroidered patches, cyanotype blueprints, risograph prints, a neon-noir alley, stained glass, and a three-step edit chain. One seed, first output kept, nothing cherry-picked. sample art image models generated https://stuff.scrappylabs.ai/public/moya/watch-skill-guide-20260722.html watch-guide Wed, 22 Jul 2026 12:00:00 +0000 How we let an AI actually watch a video instead of guessing from the title — scene-aware frame extraction, deduplicated shots, transcript from captions or local speech-to-text. The how-to, and what it costs to run. video multimodal open source https://github.com/scrappylabsai/audhd gh-audhd Thu, 16 Jul 2026 12:00:00 +0000 Open source. Spawns isolated parallel agents under distorted cognitive frames to fight premature convergence, then scores and clusters what comes back. A fork of UditAkhourii/adhd that names the systematizing-critic half. MIT. open source agents ideation https://github.com/scrappylabsai/watch gh-watch Sun, 12 Jul 2026 12:00:00 +0000 Open source. Scene-aware frame extraction plus local-first speech recognition, so an agent can actually answer questions about a video instead of guessing from the title. Cloud transcription is opt-in, not required. open source video asr https://scrappylabs.ai/babysitter babysitter Fri, 10 Jul 2026 12:00:00 +0000 An autonomous agent with a terminal will happily run rm -rf if a poisoned web page tells it to. So before any of ours runs a command, a world model dry-runs it in an imagined world and flags DESTRUCTIVE before it happens. The predict-then-act gate that catches rm -rf, agentjacking, and exfil — and the honest reason “we keep it wrangled” is a mechanism, not a slogan. world model agent safety predict-then-act prompt injection https://scrappylabs.ai/specvoice specvoice Fri, 10 Jul 2026 12:00:00 +0000 Borrowing a trick from CPU design and applying it to conversation: start computing the likely answer before the caller has finished asking. What speculative execution looks like when the pipeline is a voice agent rather than a processor. voice agents latency architecture https://scrappylabs.ai/two-years two-years Wed, 08 Jul 2026 12:00:00 +0000 The honest arc — from fighting Windows to a fleet of local models that finally feel real. The MCP tool-use moment, a memory layer a frontier model graded A− (and told us exactly where we're still behind), the month local inference got real, and a take on the buggy-whip crowd. local inference the arc memory https://scrappylabs.ai/scrappy-tools scrappy-tools Fri, 20 Mar 2026 12:00:00 +0000 Small local models can drive real command-line tools if you teach them properly. What worked, what didn't, and why tool-use is the capability that makes a small model useful instead of merely fast. small models tool use local inference https://scrappylabs.ai/qb-agent qb-agent Tue, 10 Feb 2026 12:00:00 +0000 Bookkeeping by conversation instead of by form. Invoices, expenses, reports and purchase orders handled in plain English, wired straight into the accounting system you already run — with a human approving anything that writes. agents quickbooks automation https://scrappylabs.ai/demo demo Wed, 04 Feb 2026 12:00:00 +0000 Our text-to-speech running live in your browser, across languages. Type something and hear it back — no signup, nothing to install. tts voice demo