{
  "site": {
    "name": "Joseph Magly / magly.net",
    "url": "https://magly.net",
    "description": "Portfolio, services, open source work, public keys, authored writing, and generated project updates."
  },
  "machineReadable": {
    "llms": "https://magly.net/llms.txt",
    "fullText": "https://magly.net/llms-full.txt",
    "contentJson": "https://magly.net/content.json",
    "contentIndex": "https://magly.net/content/index.json",
    "pagesJson": "https://magly.net/content/pages/",
    "rss": "https://magly.net/rss.xml",
    "sitemap": "https://magly.net/sitemap.xml",
    "fortemiCorpus": "https://magly.net/fortemi-corpus/manifest.json"
  },
  "pages": [
    {
      "id": "home",
      "path": "/",
      "url": "https://magly.net/",
      "title": "Joseph Magly - Enterprise Architect, Functional Safety & Bounded AI",
      "summary": "Portfolio and operating profile for Joseph Magly: founder of Integro Labs, CTO of Selfient, operator at BT6, and principal engineer for Roko Network.",
      "sections": [
        {
          "heading": "Current focus",
          "body": "Building the production stack for AI agents: AIWG for multi-agent workflows, Fortemi for self-hosted AI memory and retrieval, agentic-sandbox for isolated execution, and Carbonyl for trusted browser automation."
        },
        {
          "heading": "Positioning",
          "body": "Twenty-five years of building software for other people, with current work centered on systems that ship, get used, and stay maintainable. The through-line is enterprise architecture, functional safety, bounded AI, agentic operations, and practical delivery."
        },
        {
          "heading": "Current roles",
          "body": "Founder and CEO of Integro Labs. Chief Technology Officer of Selfient. Operator at BT6. Principal Engineer for Roko Network."
        }
      ],
      "json": "https://magly.net/content/pages/home.json"
    },
    {
      "id": "about",
      "path": "/about",
      "url": "https://magly.net/about",
      "title": "About Joseph Magly",
      "summary": "Engineering executive and hands-on builder in enterprise architecture, functional safety, Web3, industrial automation, cloud systems, and bounded AI.",
      "sections": [
        {
          "heading": "Background",
          "body": "Joseph Magly has led architecture and engineering work across education platforms, patent-processing systems, warehouse robotics, blockchain infrastructure, and agentic AI systems."
        },
        {
          "heading": "Safety lens",
          "body": "The bounded-AI and independent-monitor patterns used in current agent work are grounded in prior SIL3 / IEC 61508 functional-safety architecture for warehouse robotics."
        }
      ],
      "json": "https://magly.net/content/pages/about.json"
    },
    {
      "id": "work",
      "path": "/work",
      "url": "https://magly.net/work",
      "title": "Work",
      "summary": "Selected work across agentic systems, enterprise architecture, industrial automation, developer tools, and Web3 infrastructure.",
      "sections": [
        {
          "heading": "Agentic systems",
          "body": "AIWG, Fortemi, agentic-sandbox, Carbonyl, matric-eval, and related tools form the current Agentic OS stack."
        },
        {
          "heading": "Enterprise architecture",
          "body": "Microservices, event-driven architecture, service boundaries, cloud optimization, architecture review boards, and production delivery across large engineering organizations."
        },
        {
          "heading": "Industrial and safety systems",
          "body": "SIL3-rated robotics safety, PLC integration, human-zone protection, and independent runtime monitors."
        },
        {
          "heading": "Web3 and trust systems",
          "body": "Smart contracts, escrow architecture, identity, verifiable credentials, timing infrastructure, and on-chain / off-chain trust boundaries."
        }
      ],
      "json": "https://magly.net/content/pages/work.json"
    },
    {
      "id": "services",
      "path": "/services",
      "url": "https://magly.net/services",
      "title": "Services",
      "summary": "Consulting and build lanes for agentic platforms, executive technology leadership, AI governance, production AI memory, trusted browser automation, and safety/security/compliance review.",
      "sections": [
        {
          "heading": "Operating model",
          "body": "Engagements are scoped around practical outcomes: architecture, production agent builds, reviews, memory/retrieval deployments, and agent/browser automation that survives real use."
        },
        {
          "heading": "Proof base",
          "body": "The service catalog is anchored in public and production work: AIWG, Fortemi, agentic-sandbox, Carbonyl, matric-eval, Selfient, Roko Network, LearningMate, LitAgility, OneOf, and Symbotic."
        }
      ],
      "json": "https://magly.net/content/pages/services.json"
    },
    {
      "id": "opensource",
      "path": "/opensource",
      "url": "https://magly.net/opensource",
      "title": "Open Source",
      "summary": "Public repositories and ecosystems around AIWG, Fortemi, Carbonyl, Pagenary, MCP tooling, model evaluation, and systems utilities.",
      "sections": [
        {
          "heading": "Agentic framework and runtime",
          "body": "AIWG provides skills, agents, rules, and provider deployment tooling. agentic-sandbox provides sandbox and VM tooling for safer, repeatable agent execution."
        },
        {
          "heading": "AI knowledge and semantic search",
          "body": "Fortemi is a self-hosted semantic-memory and retrieval system. fortemi-react provides React and browser-native components. HotM is knowledge and semantic-search tooling in the Fortemi ecosystem."
        },
        {
          "heading": "AI tooling and automation",
          "body": "Carbonyl and carbonyl-agent support real-browser agent work. mcp-hound exposes Hound code search over MCP. matric-eval supports model and workflow evaluation."
        },
        {
          "heading": "Apps and utilities",
          "body": "Public work also includes Pagenary, markdown-editor, transcribe-cli, kintsugi-usb, ubuntu-setup, yubi-toolkit, and roko-token-extractor."
        }
      ],
      "json": "https://magly.net/content/pages/opensource.json"
    },
    {
      "id": "fortemi",
      "path": "/fortemi",
      "url": "https://magly.net/fortemi",
      "title": "Fortemi Browser Demo",
      "summary": "In-browser knowledge-base demo powered by Fortemi components, showing semantic memory, retrieval, note browsing, metadata, and graph-oriented knowledge surfaces.",
      "sections": [
        {
          "heading": "Purpose",
          "body": "Fortemi is a data layer for structured and unstructured knowledge: one normalized store with hybrid retrieval, provenance, metadata, citations, and graph organization."
        },
        {
          "heading": "Agent access",
          "body": "The site also exposes the Fortemi corpus under /fortemi-corpus so agents and crawlers can fetch manifest, shard, summaries, notes, topic graph, and citation graph files directly without running the browser app."
        }
      ],
      "json": "https://magly.net/content/pages/fortemi.json"
    },
    {
      "id": "cv",
      "path": "/cv",
      "url": "https://magly.net/cv",
      "title": "CV",
      "summary": "Role timeline covering BT6, Selfient, Roko Network, Integro Labs, LearningMate, OneOf, LitAgility, Symbotic, and earlier engineering roles.",
      "sections": [
        {
          "heading": "Current roles",
          "body": "BT6 Operator; Selfient Chief Technology Officer; Roko Network Principal Engineer; Integro Labs Founder and CEO."
        },
        {
          "heading": "Prior highlights",
          "body": "LearningMate Principal Architect; OneOf VP Engineering; LitAgility CTO; Symbotic Software Architect / Principal Software Engineer with SIL3 functional-safety work."
        }
      ],
      "json": "https://magly.net/content/pages/cv.json"
    },
    {
      "id": "now",
      "path": "/now",
      "url": "https://magly.net/now",
      "title": "Now",
      "summary": "Current public operating focus: shipping agentic infrastructure, AI memory and retrieval, browser automation, evaluation, and bounded autonomous operations.",
      "sections": [
        {
          "heading": "Current direction",
          "body": "Driving toward verifiable, fully autonomous operations by combining bounded agents, retrieval-backed memory, isolated execution, browser automation, and evaluation loops."
        }
      ],
      "json": "https://magly.net/content/pages/now.json"
    },
    {
      "id": "contact",
      "path": "/contact",
      "url": "https://magly.net/contact",
      "title": "Contact",
      "summary": "Contact route for Joseph Magly and engagement inquiries.",
      "sections": [
        {
          "heading": "Email",
          "body": "joe@magly.net"
        },
        {
          "heading": "Profiles",
          "body": "LinkedIn: https://www.linkedin.com/in/josephmagly. GitHub: https://github.com/jmagly. Gitea: https://git.integrolabs.net/roctinam."
        }
      ],
      "json": "https://magly.net/content/pages/contact.json"
    },
    {
      "id": "keys",
      "path": "/keys",
      "url": "https://magly.net/keys",
      "title": "Public Keys",
      "summary": "Public signing keys and verification material for release artifacts.",
      "sections": [
        {
          "heading": "Carbonyl release key",
          "body": "The Carbonyl release signing key is available as direct text and armored key material under /keys/carbonyl-release.asc.txt and /keys/carbonyl-release.asc."
        }
      ],
      "json": "https://magly.net/content/pages/keys.json"
    },
    {
      "id": "inquire-agentic-platform-architecture",
      "path": "/inquire/agentic-platform-architecture",
      "url": "https://magly.net/inquire/agentic-platform-architecture",
      "title": "Agentic platform architecture",
      "summary": "Plan the agent workforce, build pipelines on AIWG, integrate MCP, and ship to production.",
      "sections": [
        {
          "heading": "Engagement shape",
          "body": "Starts with the workforce: which agents and roles to run, how they hand off, where humans stay in the loop, and what guardrails bound the agent surface."
        },
        {
          "heading": "Stack",
          "body": "Built on AIWG, Fortemi for self-hosted hybrid-retrieval memory, matric-eval for benchmarking, agentic-sandbox for isolation, and Matric runtime components where appropriate."
        }
      ],
      "json": "https://magly.net/content/pages/inquire-agentic-platform-architecture.json"
    },
    {
      "id": "inquire-technology-leadership",
      "path": "/inquire/technology-leadership",
      "url": "https://magly.net/inquire/technology-leadership",
      "title": "Executive technology leadership / enterprise AI architecture",
      "summary": "Board- and founder-level technical leadership for organizations turning AI ambition into operating capability.",
      "sections": [
        {
          "heading": "Engagement shape",
          "body": "Align the business bet, set the architecture, unblock the team, and ship the migration or capability."
        },
        {
          "heading": "Proof base",
          "body": "Current and recent technical leadership includes Selfient, Roko Network, LearningMate, Integro Labs, OneOf, and LitAgility."
        }
      ],
      "json": "https://magly.net/content/pages/inquire-technology-leadership.json"
    },
    {
      "id": "inquire-agent-governance-evaluation",
      "path": "/inquire/agent-governance-evaluation",
      "url": "https://magly.net/inquire/agent-governance-evaluation",
      "title": "Agent governance and evaluation",
      "summary": "Bounded-AI patterns, eval harnesses, and accountability stacks for agentic systems.",
      "sections": [
        {
          "heading": "Engagement shape",
          "body": "Threat model the agent surface, design evaluation rubrics, run benchmark and behavior validation loops, and turn findings into practical controls."
        },
        {
          "heading": "Proof base",
          "body": "Uses patterns from SIL3 independent safety monitors and current eval work across frontier models and agentic systems."
        }
      ],
      "json": "https://magly.net/content/pages/inquire-agent-governance-evaluation.json"
    },
    {
      "id": "inquire-safety-security-compliance-review",
      "path": "/inquire/safety-security-compliance-review",
      "url": "https://magly.net/inquire/safety-security-compliance-review",
      "title": "Safety, security and compliance review",
      "summary": "Independent review for systems where AI behavior, security boundaries, and compliance evidence all matter.",
      "sections": [
        {
          "heading": "Engagement shape",
          "body": "Threat model the agent surface, map safety controls to runtime guardrails, identify compliance evidence gaps, and produce an executive-readable risk register."
        },
        {
          "heading": "Proof base",
          "body": "Grounded in SIL3 / IEC 61508 functional-safety architecture, frontier-model red-team and evaluation methodology, and current security architecture around identity, key custody, smart contracts, and trust boundaries."
        }
      ],
      "json": "https://magly.net/content/pages/inquire-safety-security-compliance-review.json"
    },
    {
      "id": "inquire-production-ai-memory",
      "path": "/inquire/production-ai-memory",
      "url": "https://magly.net/inquire/production-ai-memory",
      "title": "Production AI memory and retrieval",
      "summary": "Hybrid retrieval, automatic knowledge graphs, and multimodal ingestion for self-hosted AI memory.",
      "sections": [
        {
          "heading": "Engagement shape",
          "body": "Deploy Fortemi against the client corpus and sovereignty requirements: BM25 plus dense retrieval plus reciprocal-rank fusion, automatic graph construction, multimodal adapters, and MCP agent access."
        },
        {
          "heading": "Proof base",
          "body": "Fortemi is the reference implementation and runs as a self-hosted Rust/PostgreSQL memory layer with browser-native demo surfaces."
        }
      ],
      "json": "https://magly.net/content/pages/inquire-production-ai-memory.json"
    },
    {
      "id": "inquire-trusted-browser-automation",
      "path": "/inquire/trusted-browser-automation",
      "url": "https://magly.net/inquire/trusted-browser-automation",
      "title": "Trusted browser automation for agents",
      "summary": "Real-browser automation purpose-built for LLM agents: no display, real fingerprint, real session state.",
      "sections": [
        {
          "heading": "Engagement shape",
          "body": "Integrate Carbonyl runtime, carbonyl-agent, and fleet management into agent pipelines for scraping, testing, monitoring, red-team operations, and session-bearing browser work."
        },
        {
          "heading": "Proof base",
          "body": "Carbonyl is an actively maintained Chromium-stable terminal browser stack with an agent-facing SDK and fleet-oriented automation layer."
        }
      ],
      "json": "https://magly.net/content/pages/inquire-trusted-browser-automation.json"
    },
    {
      "id": "engage-agent-teardown",
      "path": "/engage/agent-teardown",
      "url": "https://magly.net/engage/agent-teardown",
      "title": "Agent Teardown",
      "summary": "$1,000 engagement. Bring the workflow eating the most time. Within 48 hours you get a working proof and a build plan you can act on, with the fee credited toward a full Build Sprint.",
      "sections": [
        {
          "heading": "Offer",
          "body": "Bring the workflow eating the most time. Within 48 hours you get a working proof and a build plan you can act on, with the fee credited toward a full Build Sprint."
        },
        {
          "heading": "Access",
          "body": "Primary route: https://magly.net/engage/agent-teardown. Legacy /hire/agent-teardown URLs redirect here."
        }
      ],
      "json": "https://magly.net/content/pages/engage-agent-teardown.json"
    },
    {
      "id": "engage-build-sprint",
      "path": "/engage/build-sprint",
      "url": "https://magly.net/engage/build-sprint",
      "title": "Build Sprint",
      "summary": "$8,000 engagement. One production-grade agent or automation, tested, documented, handed off, and owned by the client. Built on the open AIWG stack.",
      "sections": [
        {
          "heading": "Offer",
          "body": "One production-grade agent or automation, tested, documented, handed off, and owned by the client. Built on the open AIWG stack."
        },
        {
          "heading": "Access",
          "body": "Primary route: https://magly.net/engage/build-sprint. Legacy /hire/build-sprint URLs redirect here."
        }
      ],
      "json": "https://magly.net/content/pages/engage-build-sprint.json"
    },
    {
      "id": "engage-architecture-review",
      "path": "/engage/architecture-review",
      "url": "https://magly.net/engage/architecture-review",
      "title": "Architecture Review",
      "summary": "$5,000 engagement. A fixed-scope review of a system or AI architecture, with threat modeling, an executive-readable risk register, a prioritized remediation plan, and a findings walkthrough.",
      "sections": [
        {
          "heading": "Offer",
          "body": "A fixed-scope review of a system or AI architecture, with threat modeling, an executive-readable risk register, a prioritized remediation plan, and a findings walkthrough."
        },
        {
          "heading": "Access",
          "body": "Primary route: https://magly.net/engage/architecture-review. Legacy /hire/architecture-review URLs redirect here."
        }
      ],
      "json": "https://magly.net/content/pages/engage-architecture-review.json"
    },
    {
      "id": "engage-vibe-code",
      "path": "/engage/vibe-code",
      "url": "https://magly.net/engage/vibe-code",
      "title": "Vibe Code GTM Readiness",
      "summary": "$1,000 engagement. A generative or vibe-coded system review that returns a go-to-market readiness plan, executable issue deck, and hardening priorities.",
      "sections": [
        {
          "heading": "Offer",
          "body": "A generative or vibe-coded system review that returns a go-to-market readiness plan, executable issue deck, and hardening priorities."
        },
        {
          "heading": "Access",
          "body": "Primary route: https://magly.net/engage/vibe-code. Legacy /hire/vibe-code URLs redirect here."
        }
      ],
      "json": "https://magly.net/content/pages/engage-vibe-code.json"
    },
    {
      "id": "engage-agentic-research-report",
      "path": "/engage/agentic-research-report",
      "url": "https://magly.net/engage/agentic-research-report",
      "title": "Agentic Research Report",
      "summary": "$1,500 engagement. A scoped, sourced research report with multi-model research, clean synthesis, confidence notes, conflicts, recommended next moves, and an auditable source trail.",
      "sections": [
        {
          "heading": "Offer",
          "body": "A scoped, sourced research report with multi-model research, clean synthesis, confidence notes, conflicts, recommended next moves, and an auditable source trail."
        },
        {
          "heading": "Access",
          "body": "Primary route: https://magly.net/engage/agentic-research-report. Legacy /hire/agentic-research-report URLs redirect here."
        }
      ],
      "json": "https://magly.net/content/pages/engage-agentic-research-report.json"
    },
    {
      "id": "engage-agentic-codebase-audit",
      "path": "/engage/agentic-codebase-audit",
      "url": "https://magly.net/engage/agentic-codebase-audit",
      "title": "Agentic Codebase Audit",
      "summary": "$1,000 engagement. A bounded codebase pass with multi-pass model review, ranked findings, and an executable issue list.",
      "sections": [
        {
          "heading": "Offer",
          "body": "A bounded codebase pass with multi-pass model review, ranked findings, and an executable issue list."
        },
        {
          "heading": "Access",
          "body": "Primary route: https://magly.net/engage/agentic-codebase-audit. Legacy /hire/agentic-codebase-audit URLs redirect here."
        }
      ],
      "json": "https://magly.net/content/pages/engage-agentic-codebase-audit.json"
    },
    {
      "id": "engage-agentic-docs-pack",
      "path": "/engage/agentic-docs-pack",
      "url": "https://magly.net/engage/agentic-docs-pack",
      "title": "Agentic Docs Pack",
      "summary": "$1,000 engagement. Turn scattered project knowledge into useful onboarding, architecture, runbook, handoff, or focused Markdown documentation.",
      "sections": [
        {
          "heading": "Offer",
          "body": "Turn scattered project knowledge into useful onboarding, architecture, runbook, handoff, or focused Markdown documentation."
        },
        {
          "heading": "Access",
          "body": "Primary route: https://magly.net/engage/agentic-docs-pack. Legacy /hire/agentic-docs-pack URLs redirect here."
        }
      ],
      "json": "https://magly.net/content/pages/engage-agentic-docs-pack.json"
    },
    {
      "id": "engage-agentic-data-extraction",
      "path": "/engage/agentic-data-extraction",
      "url": "https://magly.net/engage/agentic-data-extraction",
      "title": "Agentic Data Extraction",
      "summary": "$2,000 engagement. Extract, normalize, and hand back structured data from a bounded source set with quality notes and repeatability guidance.",
      "sections": [
        {
          "heading": "Offer",
          "body": "Extract, normalize, and hand back structured data from a bounded source set with quality notes and repeatability guidance."
        },
        {
          "heading": "Access",
          "body": "Primary route: https://magly.net/engage/agentic-data-extraction. Legacy /hire/agentic-data-extraction URLs redirect here."
        }
      ],
      "json": "https://magly.net/content/pages/engage-agentic-data-extraction.json"
    },
    {
      "id": "engage-agentic-migration-pass",
      "path": "/engage/agentic-migration-pass",
      "url": "https://magly.net/engage/agentic-migration-pass",
      "title": "Agentic Migration Pass",
      "summary": "$4,000 engagement. Move a bounded system or corpus with transfer work, comparison, verification notes, and a next-pass map.",
      "sections": [
        {
          "heading": "Offer",
          "body": "Move a bounded system or corpus with transfer work, comparison, verification notes, and a next-pass map."
        },
        {
          "heading": "Access",
          "body": "Primary route: https://magly.net/engage/agentic-migration-pass. Legacy /hire/agentic-migration-pass URLs redirect here."
        }
      ],
      "json": "https://magly.net/content/pages/engage-agentic-migration-pass.json"
    },
    {
      "id": "engage-quick-win",
      "path": "/engage/quick-win",
      "url": "https://magly.net/engage/quick-win",
      "title": "Quick Win",
      "summary": "$500 engagement. A small expert intervention for one concrete problem, decision, review point, or stuck technical issue.",
      "sections": [
        {
          "heading": "Offer",
          "body": "A small expert intervention for one concrete problem, decision, review point, or stuck technical issue."
        },
        {
          "heading": "Access",
          "body": "Primary route: https://magly.net/engage/quick-win. Legacy /hire/quick-win URLs redirect here."
        }
      ],
      "json": "https://magly.net/content/pages/engage-quick-win.json"
    },
    {
      "id": "engage-build-engagement",
      "path": "/engage/build-engagement",
      "url": "https://magly.net/engage/build-engagement",
      "title": "Build Engagement",
      "summary": "$20,000 engagement. A larger expert-led build with scoped outcome ownership, architecture, hands-on execution, checkpoints, and clean handoff.",
      "sections": [
        {
          "heading": "Offer",
          "body": "A larger expert-led build with scoped outcome ownership, architecture, hands-on execution, checkpoints, and clean handoff."
        },
        {
          "heading": "Access",
          "body": "Primary route: https://magly.net/engage/build-engagement. Legacy /hire/build-engagement URLs redirect here."
        }
      ],
      "json": "https://magly.net/content/pages/engage-build-engagement.json"
    }
  ],
  "updates": [
    {
      "slug": "2026-06-the-brakes-come-first",
      "title": "June 2026 Portfolio Update: The brakes come first.",
      "date": "2026-06-30",
      "tags": [
        "monthly",
        "bounded-ai",
        "functional-safety",
        "agentic-ai"
      ],
      "summary": "A generated June 2026 portfolio update from project history and release notes across the open agent stack: brakes, monitors, isolation, memory, and publishing surfaces.",
      "image": "https://magly.net/assets/updates/2026-06/brakes-come-first-hero.png",
      "url": "https://magly.net/updates/2026-06-the-brakes-come-first/",
      "html": "https://magly.net/updates/2026-06-the-brakes-come-first.html",
      "markdown": "https://magly.net/updates/2026-06-the-brakes-come-first.md",
      "json": "https://magly.net/content/2026-06-the-brakes-come-first.json",
      "body": "# June 2026 Portfolio Update: The brakes come first.\n\n![Functional-safety themed release artwork for the June 2026 portfolio update.](/assets/updates/2026-06/brakes-come-first-hero.png)\n\nJune was a build month.\n\nAIWG shipped a cockpit. The sandbox hardened its VM path and its verified connections. Fortemi moved deeper into streaming memory and inbound events. Pagenary learned how to publish blogs. Carbonyl stayed in the maintenance lane. Fortemi React kept the browser-side memory surface moving.\n\nJune also put the stack in public, human rooms. We attended a16z's Boston and New York Tech Week events, met a long list of founders, operators, investors, and product teams, and left grateful for the hosts and organizers who made those rooms useful. The quiet proof point for us was that the agent stack came along as working infrastructure: our AI tools acted as personal assistants across the week, helping keep schedules straight, notes usable, and attention pointed toward the right rooms and the right conversations. The events were the story. The tools did what good tools should do: stay mostly out of the way while making the work sharper.\n\n![A visual roundup of the June 2026 open agent stack release campaign.](/assets/updates/2026-06/portfolio-roundup-stack.png)\n\nThat sounds like a stack update, and it is. But the through-line is older than the stack.\n\nThe brakes come first.\n\nI learned that in industrial systems before I had the words for bounded AI. A robot fleet can be fast, adaptive, and useful, but none of that matters if the system cannot stop cleanly when its confidence, context, or operating envelope changes. The safety layer is not decoration around the primary system. It is the reason the primary system is allowed to move.\n\nThat is still how I think about agents.\n\nAn agent that can run tools, write files, open terminals, call APIs, and coordinate other agents is not a chatbot anymore. It is an actuator. Once software can act, the important question shifts from \"can it do the task?\" to \"what bounds the task while it is doing it?\"\n\nIn June, most of the useful work landed around those bounds.\n\n![June 2026 at a glance across AIWG, Fortemi, Pagenary, Carbonyl, agentic-sandbox, and fortemi-react.](/assets/updates/2026-06/june-at-a-glance.png)\n\nAIWG got better at showing what agents are doing and giving operators a place to steer them. That matters because opaque automation is not autonomy; it is just unsupervised state change. The Cockpit work is not a prettier console. It is an operator surface. You should be able to see a session, understand the posture it is running under, and interrupt or redirect it without spelunking through logs.\n\n![AIWG cockpit mock showing sessions, approval gates, stop controls, context posture, and a live terminal.](/assets/updates/2026-06/aiwg-cockpit-mock.png)\n\n[Full AIWG June report](https://docs.aiwg.io/#blog/2026-6-aiwg-june-2026-report)\n\nThe sandbox work is the same argument at the runtime layer. Verified connections, VM enrollment, gateway SSH, and more reliable lifecycle handling are not glamorous features, but they are the difference between \"an agent ran somewhere\" and \"an agent ran in a place with walls.\" If the agent is going to execute code, then the execution environment has to be designed as part of the product.\n\n![Agentic Sandbox transport diagram covering mTLS transport, vsock enrollment, gateway SSH lease, and signed images.](/assets/updates/2026-06/agentic-sandbox-transport.png)\n\n[Full agentic-sandbox June report](https://docs.aiwg.io/agentic-sandbox/#blog/2026-06-agentic-sandbox)\n\nMemory has the same shape. Fortemi's June work around streaming chat, signed incoming webhooks, resumable ingest, redaction, audit logging, and fail-closed startup is not just about making an assistant remember more. Memory is a control surface. What comes in, what is trusted, what is retained, what is redacted, and what can be replayed all become part of the safety case.\n\n![Fortemi ingest pipeline diagram showing streaming, back-pressure, resumable upload, signed webhooks, redaction, and audit logging.](/assets/updates/2026-06/fortemi-ingest.png)\n\n[Full Fortemi June report](https://docs.fortemi.com/server/#posts/2026-06-fortemi)\n\nThe browser build matters because a lot of agent work ends up needing to be inspectable by people. Fortemi React kept pushing the local, visible memory interface: graph layout, tool intent, static-file backends, and browser-side interaction patterns. A good agent stack needs headless power, but it also needs surfaces where a human can read the state without becoming an archaeologist.\n\n![Fortemi React browser-side memory mock with semantic search, note graph, local loading, and tool intent.](/assets/updates/2026-06/fortemi-react-search.png)\n\n[Full fortemi-react June report](https://docs.fortemi.com/react/#posts/2026-06-fortemi-react)\n\nPagenary's blog work looks like publishing infrastructure, but I think of it as part of the same loop. Systems that cannot explain themselves drift. Monthly reports, docs, maps, search, and simple public pages force a stack to say what changed and why. The blog layout family is useful because it turns the record into a habit instead of a one-off artifact.\n\n![Pagenary build pipeline diagram from Markdown to static pages, visual effects, Fortemi search, and deployment.](/assets/updates/2026-06/pagenary-build.png)\n\n[Full Pagenary June report](https://pagenary.com/updates/post/?id=posts%2F2026-06-pagenary)\n\nCarbonyl was quieter. That is fine. Some months are about moving the boundary; some months are about keeping the edge from drifting. Maintenance and QA are part of the brakes too.\n\n![Carbonyl pipeline diagram showing input, headless Chromium, terminal rendering, and structured extraction.](/assets/updates/2026-06/carbonyl-pipeline.png)\n\n[Full Carbonyl June report](https://github.com/jmagly/carbonyl/blob/main/docs/blog/2026-6-carbonyl-june-2026-report.md)\n\nThe mistake I see in AI systems is treating safety as a later pass. First make it powerful, then make it safe. First give it tools, then figure out policy. First let it roam, then add monitoring.\n\nThat order is backwards.\n\nThe brake is not the opposite of speed. It is what lets speed be used.\n\nThe same pattern shows up whether the machine is a warehouse robot, a VM-hosted coding agent, a memory service ingesting events, or a publishing system carrying the changelog. The useful work is not only the model, the tool call, or the generated patch. The useful work is the envelope around it: isolation, provenance, operator visibility, redaction, verification, and the ability to stop.\n\n![System map of the open agent stack connecting orchestration, sandboxed execution, memory, browser surfaces, publishing, and terminal automation.](/assets/updates/2026-06/agentic-os-stack.png)\n\nThat is the direction for the stack after June.\n\nMore capability, but only where the operating envelope is getting clearer at the same time. More autonomy, but with better independent monitors. More memory, but with explicit intake and audit boundaries. More publishing, because if the system cannot explain what changed, it is not ready to be trusted.\n\nThat is the long direction: verifiable, fully autonomous operations where the system can act, show its work, prove its bounds, and stop cleanly when those bounds are no longer true.\n\nThe brakes come first."
    }
  ]
}