<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Dan Stolts</title><link>https://danstolts.com</link><description>Dan Stolts release feed</description><item><title>ADHD-first design, shipped across 3 specs and a public repo</title><link>https://danstolts.com/proof/adhd-first-design-and-jitneuro/</link><guid>https://danstolts.com/proof/adhd-first-design-and-jitneuro/</guid><description>A binding, portfolio-wide UX constraint set naming ADHD and dyslexia as design disabilities, applied against three shipped design specs and carried into a public open-source memory toolkit (jitneuro).</description><pubDate>Mon, 24 Aug 2026 00:00:00 +0000</pubDate></item><item><title>AI Boat Mechanic: a new vertical in weeks, not months</title><link>https://danstolts.com/proof/ai-boat-mechanic-marine-vertical-product-extension/</link><guid>https://danstolts.com/proof/ai-boat-mechanic-marine-vertical-product-extension/</guid><description>A second commercial vertical launched via a frontend branding skin, vertical-specific AI prompts, and light API additions on an unchanged engine -- a ~13-week scaffold-to-live-launch span, verified from commit history.</description><pubDate>Mon, 24 Aug 2026 00:00:00 +0000</pubDate></item><item><title>AI Field Support: voice-first, offline-first, Entra SSO, confidence-gated AI</title><link>https://danstolts.com/proof/ai-field-support-multi-tenant-voice-first-platform/</link><guid>https://danstolts.com/proof/ai-field-support-multi-tenant-voice-first-platform/</guid><description>Voice-first, offline-first, multi-tenant PWA with real Microsoft 365/Entra SSO, confidence-gated AI diagnosis, and multi-vendor AI failover -- CI-gated promotion to uat/main.</description><pubDate>Mon, 24 Aug 2026 00:00:00 +0000</pubDate></item><item><title>Chunking: 20-candidate operating point, 14-of-14 validated</title><link>https://danstolts.com/proof/chunked-section-bounded-processing-at-scale/</link><guid>https://danstolts.com/proof/chunked-section-bounded-processing-at-scale/</guid><description>A previously-recommended chunking rule, validated at real scale (14 of 14 acceptable), including a self-correction to its own sizing model.</description><pubDate>Mon, 24 Aug 2026 00:00:00 +0000</pubDate></item><item><title>Content factory design: 8-state lifecycle, 3-tier risk review</title><link>https://danstolts.com/proof/continuous-technology-portfolio-content-factory-design/</link><guid>https://danstolts.com/proof/continuous-technology-portfolio-content-factory-design/</guid><description>A designed eight-state record lifecycle and three-tier risk classification for continuous evidence publication; evidence-pending by the record&#x27;s own account.</description><pubDate>Mon, 24 Aug 2026 00:00:00 +0000</pubDate></item><item><title>L3 audit readiness: 1 source shape ready, 1 blocked and named</title><link>https://danstolts.com/proof/deep-tier-evidence-audit-l3-readiness/</link><guid>https://danstolts.com/proof/deep-tier-evidence-audit-l3-readiness/</guid><description>A mechanical L3 evidence-audit readiness check that found and reported a real, reproduced blocked source shape rather than assuming coverage.</description><pubDate>Mon, 24 Aug 2026 00:00:00 +0000</pubDate></item><item><title>Content gate: 10-stage blocking contract, named judges per stage</title><link>https://danstolts.com/proof/evidence-governed-content-delivery-gate-system/</link><guid>https://danstolts.com/proof/evidence-governed-content-delivery-gate-system/</guid><description>A canonical, currently binding multi-stage editorial quality-gate contract that every content asset must pass before it ships.</description><pubDate>Mon, 24 Aug 2026 00:00:00 +0000</pubDate></item><item><title>Redaction gate: 4 real leak paths found, fixed, and re-verified</title><link>https://danstolts.com/proof/fail-closed-secret-and-pii-redaction-gate/</link><guid>https://danstolts.com/proof/fail-closed-secret-and-pii-redaction-gate/</guid><description>A real adversarial red-team of the system&#x27;s own secret-redaction mechanism, four leak paths found, fixed, and re-verified with a full regression run.</description><pubDate>Mon, 24 Aug 2026 00:00:00 +0000</pubDate></item><item><title>Source isolation: git-object-only reads, byte-identical replay</title><link>https://danstolts.com/proof/git-object-only-source-isolation-and-reproducibility/</link><guid>https://danstolts.com/proof/git-object-only-source-isolation-and-reproducibility/</guid><description>A scanner that reads only pinned, committed Git objects, never a dirty working tree, with independently re-run byte-identical replay.</description><pubDate>Mon, 24 Aug 2026 00:00:00 +0000</pubDate></item><item><title>Delivery exemplar: one governed standard across real products</title><link>https://danstolts.com/proof/governed-ai-enabled-delivery-capability-exemplar/</link><guid>https://danstolts.com/proof/governed-ai-enabled-delivery-capability-exemplar/</guid><description>A deliberately restrained cross-product exemplar of governed AI-enabled delivery; evidence-pending by the record&#x27;s own account.</description><pubDate>Mon, 24 Aug 2026 00:00:00 +0000</pubDate></item><item><title>Discovery pipeline: 5-level architecture, 10+ documented runs</title><link>https://danstolts.com/proof/governed-multi-agent-discovery-pipeline/</link><guid>https://danstolts.com/proof/governed-multi-agent-discovery-pipeline/</guid><description>A five-level (L0-L4) governed multi-agent discovery architecture, run across 10+ documented passes against real committed source material.</description><pubDate>Mon, 24 Aug 2026 00:00:00 +0000</pubDate></item><item><title>Strategy trace: every ruling SHA-pinned, staleness self-disclosed</title><link>https://danstolts.com/proof/horizon-aligned-strategy-to-execution-traceability/</link><guid>https://danstolts.com/proof/horizon-aligned-strategy-to-execution-traceability/</guid><description>Every execution ruling traces to a specific hashed strategy commit, with an explicit alignment table and a self-disclosed staleness flag.</description><pubDate>Mon, 24 Aug 2026 00:00:00 +0000</pubDate></item><item><title>Backlog governance: 12 lanes, 11 gated, 1 tooling defect caught</title><link>https://danstolts.com/proof/hubcentral-ticket-governed-backlog-execution/</link><guid>https://danstolts.com/proof/hubcentral-ticket-governed-backlog-execution/</guid><description>A twelve-lane ticket roster gated by an explicit duplicate sweep, which caught a real tooling defect before it produced a false result.</description><pubDate>Mon, 24 Aug 2026 00:00:00 +0000</pubDate></item><item><title>JitPlatform: one locked contract retiring duplicate code across 7 products</title><link>https://danstolts.com/proof/jitplatform-shared-provider-services-contract/</link><guid>https://danstolts.com/proof/jitplatform-shared-provider-services-contract/</guid><description>Shared provider-services contract -- locked schema, versioned API, and a working adapter reference implementation for one real consumer.</description><pubDate>Mon, 24 Aug 2026 00:00:00 +0000</pubDate></item><item><title>jITSecure: 4-sector risk scoring, 9-phase parity-before-cutover port</title><link>https://danstolts.com/proof/jitsecure-multi-sector-security-scanning-platform/</link><guid>https://danstolts.com/proof/jitsecure-multi-sector-security-scanning-platform/</guid><description>Multi-sector security scanning platform with industry risk multipliers, mid-migration to a cross-platform .NET binary.</description><pubDate>Mon, 24 Aug 2026 00:00:00 +0000</pubDate></item><item><title>Model routing: ladder-based selection, verified from raw score files</title><link>https://danstolts.com/proof/model-effort-escalation-governance/</link><guid>https://danstolts.com/proof/model-effort-escalation-governance/</guid><description>A measured, ladder-based model and effort escalation discipline, including a case where raising effort made the same model perform worse.</description><pubDate>Mon, 24 Aug 2026 00:00:00 +0000</pubDate></item><item><title>Precision measurement: full census of 176 candidates, 2 defects fixed</title><link>https://danstolts.com/proof/precision-measurement-and-defect-correction-discipline/</link><guid>https://danstolts.com/proof/precision-measurement-and-defect-correction-discipline/</guid><description>A full-census (not sampled) precision measurement by independent judges, refusing to compare a stricter new figure to a stale, easier one.</description><pubDate>Mon, 24 Aug 2026 00:00:00 +0000</pubDate></item><item><title>Recall measurement: 9.6% headline, reported without softening</title><link>https://danstolts.com/proof/recall-ground-truth-methodology/</link><guid>https://danstolts.com/proof/recall-ground-truth-methodology/</guid><description>The first recall measurement for this scanner, built from an independent expert inventory read before the scan output was ever opened.</description><pubDate>Mon, 24 Aug 2026 00:00:00 +0000</pubDate></item><item><title>Scanner Platform: 149 checks, one engine, two shipped products</title><link>https://danstolts.com/proof/scanner-platform-dual-mode-multi-product-scan-engine/</link><guid>https://danstolts.com/proof/scanner-platform-dual-mode-multi-product-scan-engine/</guid><description>One 149-check scan orchestrator, live on Azure Container Apps, powering an AEO/SEO product and a security domain-scan product.</description><pubDate>Mon, 24 Aug 2026 00:00:00 +0000</pubDate></item><item><title>Designing an externalized interrupt cadence for long-running AI work</title><link>https://danstolts.com/proof/scheduler-backed-wakeup-pattern/</link><guid>https://danstolts.com/proof/scheduler-backed-wakeup-pattern/</guid><description>A scheduler-owned wakeup pattern that keeps long-running automation checked-in without relying on continuous unaided attention.</description><pubDate>Mon, 24 Aug 2026 00:00:00 +0000</pubDate></item><item><title>Full-fleet MDR onboarding in a two-week window</title><link>https://danstolts.com/proof/full-fleet-mdr-onboarding-2026/</link><guid>https://danstolts.com/proof/full-fleet-mdr-onboarding-2026/</guid><description>A mid-size operator moved its entire endpoint fleet onto managed detection and response with zero unresolved onboarding exceptions.</description><pubDate>Mon, 24 Aug 2026 00:00:00 +0000</pubDate></item><item><title>Building the technology proof discovery pipeline</title><link>https://danstolts.com/proof/technology-portfolio-discovery-program/</link><guid>https://danstolts.com/proof/technology-portfolio-discovery-program/</guid><description>A queue-driven scanner that grounds every portfolio fact in an exact commit before it ever reaches public copy.</description><pubDate>Mon, 24 Aug 2026 00:00:00 +0000</pubDate></item><item><title>Why Does One Bug Filed as Two Tickets Get Fixed Twice?</title><link>https://danstolts.com/writing/why-one-bug-in-two-tickets-gets-fixed-twice/</link><guid>https://danstolts.com/writing/why-one-bug-in-two-tickets-gets-fixed-twice/</guid><description>If one bug appears in two tickets, can we prove before work starts that our people and AI agents will not repair the same outcome twice? Without that proof, the second authorization can repeat staff hours, AI calls,...</description><pubDate>Tue, 17 Mar 2020 00:00:00 +0000</pubDate></item><item><title>Should an AI Verifier Agent Check a Bug Diagnosis Before the Fix Author Starts?</title><link>https://danstolts.com/writing/does-a-passing-test-prove-your-bug-is-caught/</link><guid>https://danstolts.com/writing/does-a-passing-test-prove-your-bug-is-caught/</guid><description>Of seven diagnoses in one documented pass, independent verification overturned two and deepened one. The same pass also caught a non-persistent proposed repair before merge. Those outcomes put fix-author time, AI spend,...</description><pubDate>Mon, 15 Dec 2025 00:00:00 +0000</pubDate></item><item><title>Is the Rule You Rely On Actually a Control, or Just a Wish?</title><link>https://danstolts.com/writing/is-your-rule-actually-a-control/</link><guid>https://danstolts.com/writing/is-your-rule-actually-a-control/</guid><description>Pick one consequential rule in an AI-assisted workflow. List every route that can perform the action and the current denial evidence for each route.</description><pubDate>Tue, 10 Mar 2026 00:00:00 +0000</pubDate></item><item><title>Why Did My AI Forget What I Taught It Yesterday?</title><link>https://danstolts.com/writing/why-did-my-ai-forget-what-i-taught-it-yesterday/</link><guid>https://danstolts.com/writing/why-did-my-ai-forget-what-i-taught-it-yesterday/</guid><description>Your AI appeared to forget what you taught it yesterday because the later session did not retrieve the decision that should govern its work. That can happen even when a product saves conversation history or maintains...</description><pubDate>Mon, 09 Mar 2026 00:00:00 +0000</pubDate></item><item><title>Why Isn&#x27;t a Merged Pull Request the Same as an Approval?</title><link>https://danstolts.com/writing/why-a-merged-pull-request-is-not-an-approval/</link><guid>https://danstolts.com/writing/why-a-merged-pull-request-is-not-an-approval/</guid><description>Pick one of the last five consequential changes we merged. If we cannot show who authorized the next action for those exact bytes, under which current policy, and with what current evidence, we are one deployment away...</description><pubDate>Thu, 14 May 2026 00:00:00 +0000</pubDate></item><item><title>Why Is a Ticket Marked Ready Still Blocking the Team?</title><link>https://danstolts.com/writing/why-is-a-ticket-marked-ready-still-blocking-the-team/</link><guid>https://danstolts.com/writing/why-is-a-ticket-marked-ready-still-blocking-the-team/</guid><description>When that proof is missing, each failed pickup costs the team directly. Staff spend hours clarifying a request that should have been complete. AI workers burn tokens and compute on work they cannot finish. Every ticket...</description><pubDate>Wed, 09 Sep 2026 00:00:00 +0000</pubDate></item><item><title>Why Can a Second AI Opinion Still Leave You With the Wrong Decision?</title><link>https://danstolts.com/writing/second-opinion-is-not-a-review-panel/</link><guid>https://danstolts.com/writing/second-opinion-is-not-a-review-panel/</guid><description>Ask a second AI for a check on a plan, get an answer that agrees with the first one, and you still have no idea whether the work is safe to fund. In our own platform planning, the working plan said 8 features and 24...</description><pubDate>Sat, 04 Jul 2026 00:00:00 +0000</pubDate></item></channel></rss>
