{
  "version": 1,
  "generatedFrom": "src/config/capabilities.ts + docs/context/FEATURE_SPECS",
  "firm": {
    "name": "Casey Key Advisors",
    "shortName": "Casey Key",
    "tagline": "Multi-agent systems, installed and operated.",
    "location": "Osprey, Florida",
    "serviceArea": "Headquartered on Florida's Gulf Coast."
  },
  "founder": {
    "title": "Founder & AI Systems Lead",
    "bio": "I build practical AI systems that help leaders make better decisions and run stronger operations. At Casey Key Advisors I lead the design, integration, and operation of multi-agent systems, AI applications, and governed workflows. Previously I developed data products and commercial partnerships across consumer location, advertising technology, and mobile platforms."
  },
  "hero": {
    "headline": "We Build AI-Native Systems For How Your Company Actually Works.",
    "subhead": "Design - Develop - Integrate - Operate",
    "primaryCta": "Tell us what you want to improve",
    "secondaryCta": "See what we build"
  },
  "leadSlugs": [
    "ai-agents-automation",
    "agent-apps-mcp-channels",
    "governed-agent-delivery"
  ],
  "outOfScope": [
    "brochure website design with no owned conversion workflow or agent",
    "social media management",
    "logo work",
    "toy chatbot demos with no owned workflow"
  ],
  "scopeRules": [
    "Questions like \"do you build GPT apps?\", \"Custom GPTs?\", \"Claude apps?\", or \"MCP servers/channels?\" are IN SCOPE. Affirm yes. Route to service slug agent-apps-mcp-channels. Describe channel patterns (ChatGPT/Claude apps, MCP tool servers, human gates), not client or product brand names.",
    "Questions about MCP product marketing, cold-start onboarding, install/connect pages, app packs, starter prompts, or \"first-run for MCP\" are IN SCOPE. Affirm yes. Route to agent-apps-mcp-channels and name the MCP Product Marketing System (MPMS) pack.",
    "Questions about MCP session persistence, working-set reconnect, Default buy box / default policy bugs, or ChatGPT recalculating tool math are IN SCOPE. Affirm yes. Route to agent-apps-mcp-channels and name MCP Session Integrity (persist working set; tool output is source of truth).",
    "Questions about Content Lab, Brand Lab, Email Lab, running Brand Lab in the terminal/MCP, place-order email templates, social/poster drafts returned in chat, or lifecycle email+social via MCP are IN SCOPE. Affirm yes. Route to agent-apps-mcp-channels. Name Content Lab and the execute tools run_brand_lab / run_email_lab (drafts in chat; humans send/publish).",
    "Questions about GEO, Generative Engine Optimization, AI Overviews, showing up in ChatGPT/Perplexity answers, llms.txt, entity clarity for AI citations, or GEO Success Frameworks are IN SCOPE. Affirm yes. Route to ai-search-data. Name GEO Snapshot (/geo-analyzer) and MCP tools describe_geo / recommend_geo_framework when relevant.",
    "Questions about Page Lab, AI landing page studio, generating landing pages from a brief, brand block registry, or draft LP layouts via MCP are IN SCOPE. Affirm yes. Route to agent-apps-mcp-channels. Name Page Lab and the execute tool run_page_lab (draft layouts in chat; humans publish).",
    "Questions about fleet ownership matrices, who owns which agent path, 7-day silence / pager ownership, agent run memory, cross-run memory, or amnesiac agent relaunches are IN SCOPE. Affirm yes. Route to governed-agent-delivery. Name Fleet Ownership Matrix (Assess section H) and Agent Run Memory (Pilot Ship section I) when relevant.",
    "Questions about B2B catalog conversion, wholesale or hospitality storefront rebuilds, product finders, quote/sample/RFQ flows, AI inquiry agents, or owner-led catalog/foodservice businesses are IN SCOPE. Affirm yes. Route to ai-agents-automation (primary). Name agent-apps-mcp-channels when they want ChatGPT/Claude/MCP. Describe catalog conversion and inquiry-agent patterns; do not name client brands.",
    "Questions about who works here, founders, principals, or a public team roster are OUT OF PUBLIC SCOPE for naming. Affirm the firm is an operator-grade advisory practice. Point to /about for firm overview and /contact or Calendly to meet the people on an engagement. Do not invent teammates. Do not disclose personal names. Do not claim a public people page exists.",
    "Questions about MedTech commercial launch, clinical conversion, program launch, capital/enterprise GTM, PE MedTech diligence, or MedTech domain packs are IN SCOPE. Affirm yes. Route to ai-agents-automation (primary) and name MedTech domain packs on the governed agent framework. Describe MedTech commercial operating judgment by capability, not by personal name.",
    "Never say \"not standalone GPT apps\" or refuse ChatGPT/Claude/MCP product channels. OUT OF SCOPE \"toy chatbot demos\" means novelty chat widgets with no owned workflow, not Custom GPT / Claude / MCP work.",
    "OUT OF SCOPE \"brochure website design\" means agency visual redesign with no owned conversion workflow or agent. Catalog conversion systems, product finders, quote/RFQ agents, and AI inquiry agents are in scope under ai-agents-automation.",
    "When affirming channel work, emit a service_detail or services_overview that includes agent-apps-mcp-channels.",
    "When affirming catalog conversion or inquiry-agent work, emit a service_detail or services_overview that includes ai-agents-automation."
  ],
  "surfaces": {
    "classicSite": "https://caseykeyadvisors.com",
    "agentApp": "https://caseykeyadvisors.com/app",
    "classicHome": "https://caseykeyadvisors.com",
    "readinessSnapshot": "https://caseykeyadvisors.com/geo-analyzer",
    "mcpEndpoint": "https://mcp.ckey.app/mcp",
    "mcpTools": [
      "scan_website",
      "readiness_snapshot",
      "assess_repo",
      "list_services",
      "describe_agent_framework",
      "list_case_studies",
      "describe_mcp_offer",
      "recommend_mcp_tools",
      "describe_mcp_gtm",
      "recommend_app_pack_sections",
      "describe_mcp_runtime",
      "describe_content_lab",
      "list_brand_lab_channels",
      "run_brand_lab",
      "run_email_lab",
      "recommend_email_series",
      "recommend_post_series",
      "scaffold_place_order",
      "describe_geo",
      "recommend_geo_framework",
      "describe_page_lab",
      "list_page_lab_blocks",
      "run_page_lab",
      "recommend_landing_layout",
      "book_strategy_call"
    ],
    "calendly": "https://calendly.com/jamie-casey-key-advisors/30min",
    "contactEmail": "contact@caseykeyadvisors.com"
  },
  "values": [
    {
      "number": "01",
      "title": "Automate real work",
      "description": "Agents, automations, and integrations doing the company's actual work: reporting, support, data, marketing, and ops, not demos that never leave a slide deck."
    },
    {
      "number": "02",
      "title": "Visible logic",
      "description": "You can see what the system did, why, and with what data. Logging and ownership from the first pilot. Irreversible steps stay with your people."
    },
    {
      "number": "03",
      "title": "Measurable adoption",
      "description": "Did the automation get used? Did the business move? Those are the only scores that matter."
    },
    {
      "number": "04",
      "title": "Human gates stay in place",
      "description": "Approve, send, publish, merge, deploy: the irreversible actions stay human. We install and operate the system; your people cross the line."
    }
  ],
  "engagementModels": [
    {
      "number": "01",
      "name": "Project-based",
      "description": "Defined scope to design, build, and deploy a specific agent, automation, or adoption initiative."
    },
    {
      "number": "02",
      "name": "Retainer",
      "description": "Ongoing build and advisory partnership with dedicated execution support."
    },
    {
      "number": "03",
      "name": "Advisory",
      "description": "Executive counsel for prioritization, operating design, and high-stakes AI decisions."
    }
  ],
  "adoptionFramework": [
    {
      "step": 1,
      "name": "Assess",
      "description": "Name the workflow and systems, map data boundaries and human gates, pick one pilot job and a success measure."
    },
    {
      "step": 2,
      "name": "Pilot",
      "description": "Build the agent or automation with logging and a working approve / reject path from day one."
    },
    {
      "step": 3,
      "name": "Ship",
      "description": "Deploy into your existing systems with named ownership and a clear weekly review rhythm."
    },
    {
      "step": 4,
      "name": "Enable",
      "description": "Playbooks, measurement, and the next jobs sequenced from evidence so adoption compounds."
    }
  ],
  "expertiseAreas": [
    "Cloud agent development",
    "Business process automation",
    "Reporting and analytics automation",
    "Support and operations automation",
    "Data extraction and cleansing with provenance",
    "ChatGPT and Claude app channels",
    "MCP channel design and install",
    "MCP product marketing and cold-start packs",
    "MCP session integrity and working-set persistence",
    "Content Lab (Brand Lab + Email Lab) place-order systems",
    "Generative Engine Optimization (GEO) and citation frameworks",
    "Page Lab (AI Landing Page Studio) brand-block landing-page drafts",
    "Governed agent delivery",
    "Fleet ownership matrices for agent paths",
    "Agent run memory and warm relaunch proof",
    "Multi-agent workflow design",
    "Stack integration and automation",
    "AI adoption strategy",
    "Operating ownership and human-gate design",
    "Control-health reporting",
    "Trust-sensitive product launch",
    "First-party data for AI",
    "Executive operating cadence"
  ],
  "featuredResults": [
    {
      "category": "Channels",
      "title": "ChatGPT-first consumer utility",
      "outcome": "Shipped a ChatGPT-first savings checkup: ZIP-based scans across licensed providers, clear next steps, and a free dashboard for follow-up bill categories.",
      "metricLabel": "Consumer utility channel"
    },
    {
      "category": "Channels",
      "title": "GPT and Claude data-control assistants",
      "outcome": "Built GPT/Claude assistant surfaces around protect and rewards workflows so the agent guides, and humans still own remove, share, and payout decisions.",
      "metricLabel": "Data-control channel"
    },
    {
      "category": "Channels",
      "title": "Guided intake with a human submit gate",
      "outcome": "Shipped a GPT/Claude guided pre-filing workflow: structured intake, gap surfacing, and a hard stop before any human-owned file or submit step.",
      "metricLabel": "Guided intake channel"
    },
    {
      "category": "Catalog",
      "title": "B2B catalog conversion and inquiry agents",
      "outcome": "Rebuilt catalog conversion systems with collections, product finders, and inquiry paths so wholesale buyers can browse and request quotes without a brochure redesign agency loop.",
      "metricLabel": "Catalog conversion"
    },
    {
      "category": "Delivery",
      "title": "AI agents shipping real code under real gates",
      "outcome": "Agents write the code; humans hold merge and deploy authority, with CI gates and evidence on every change. Proven first on our own repository.",
      "metricLabel": "Governed delivery"
    }
  ],
  "portfolioApps": [
    {
      "slug": "chatgpt-bill-savings",
      "name": "ChatGPT bill-savings checkup",
      "channel": "ChatGPT + web",
      "summary": "Consumer utility pattern: a ChatGPT-first savings checkup that runs ZIP-based scans across licensed providers and opens a dashboard for follow-up bill categories.",
      "url": null
    },
    {
      "slug": "consumer-data-control",
      "name": "Consumer data-control assistants",
      "channel": "GPT / Claude + product surfaces",
      "summary": "Privacy and rewards pattern: GPT/Claude assistants that guide people through exposure, earn, and remove workflows while humans keep irreversible privacy actions.",
      "url": null
    },
    {
      "slug": "guided-prefiling",
      "name": "Guided pre-filing with human submit gates",
      "channel": "GPT / Claude + MCP",
      "summary": "Structured intake pattern: chat-guided pre-filing with gap checks, cited tool math, session integrity across host reconnects, and a hard stop before human submit.",
      "url": null
    }
  ],
  "mcpChannelOffer": {
    "headline": "MCP as a product channel for one company job",
    "summary": "We design allowlisted MCP tools (and companion ChatGPT/Claude apps) so other agents can do real work against your systems without open credentials. Humans keep irreversible keys. MPMS makes install and first-run work for cold users. Session Integrity persists working set across host reconnects and forbids LLM parallel compute.",
    "pilotIncludes": [
      "One named workflow and success measure",
      "3–8 read-first tools with clear side-effect labels",
      "MCP product marketing pack: category claim, connect page, jobs map, seeds, try prompts",
      "Runtime cold-start tools: overview, site guide, onboard checklist, seeds",
      "Session integrity: working-set persist, hydrate helpers, default-policy warning, never-recalculate instructions",
      "Content Lab when in scope: brand-data, one email place-order, one Brand post soft-linked, draft-only MCP tools",
      "Page Lab when in scope: Brand Config, block registry, one draft landing page via run_page_lab, human publish gate",
      "Pass-through or client-hosted auth; no CKA retention of client secrets",
      "request_human_gate (or equivalent) for any write path",
      "Logging, allowlist, and a one-job Pilot proof (cold session → first tool success; apply → reconnect → policy match)"
    ],
    "neverAutonomous": [
      "merge / production deploy",
      "campaign send or budget mutation",
      "landing page publish without human approve",
      "journal post / treasury move",
      "customer send without human approve"
    ],
    "engagementSpine": [
      "Assess",
      "Pilot",
      "Ship",
      "Enable"
    ],
    "serviceSlug": "agent-apps-mcp-channels"
  },
  "mcpProductMarketingOffer": {
    "headline": "MCP Product Marketing System (MPMS)",
    "summary": "Most MCP servers fail at install and first message. MPMS is the product-marketing pack ChatGPT, Claude, and Cursor load so buyers get value on turn one: category claim, connect narrative, jobs map, site guide, seeds, objections, and first-value measurement, authored and updated via MCP.",
    "includes": [
      "App pack templates (PRODUCT, JOBS, SITE_GUIDE, SEEDS, MEASURE, …)",
      "Runtime tools: get_product_overview, get_site_guide, get_onboard_checklist, seeds",
      "Connect page + starter prompts in buyer language",
      "Instructions/skill router (vague opener → overview + checklist)",
      "Validate checklist before Pilot Ship",
      "Installer agent to copy the pack into a client repo"
    ],
    "notIncluded": [
      "Kitchen-sink tool catalogs without a one-job claim",
      "Invented proof or certification claims",
      "Autonomous send / merge / deploy"
    ],
    "packPath": "delivery-framework/packs/mcp-product-marketing/",
    "serviceSlug": "agent-apps-mcp-channels"
  },
  "mcpSessionIntegrityOffer": {
    "headline": "MCP Session Integrity",
    "summary": "ChatGPT and other Streamable HTTP hosts drop MCP sessions between tool calls. Without a working-set persist layer and hard instructions, models recompute against Default policy and invent parallel math. Session Integrity is the runtime contract: hydrate → mutate → persist, never clobber live presets, warn on default policy, tool output is source of truth.",
    "includes": [
      "Working-set schema { userId, rows, policy, updatedAt }",
      "Mutation contract: hydrate if empty → mutate → persist if authenticated",
      "Hydrate helpers (runtime/hydrate.ts): no clobber of live non-default policy",
      "Compute response fields: policyName, defaultPolicy, warning, assistantGuidance",
      "Instructions snippet forbidding host-model recalculation",
      "Validate checklist + ChatGPT acceptance (named preset survives reconnect)",
      "Installer agent for client Workers (dogfooded on production channel installs)"
    ],
    "notIncluded": [
      "CKA-hosted multiplayer shared sessions or provisioned working-set storage",
      "Relying on the model to remember prior-turn apply_* args",
      "Silent multi-policy compare without an explicit tool",
      "A full deployable Worker from this pack alone (copy helpers into the client Worker)"
    ],
    "packPath": "delivery-framework/packs/mcp-session-integrity/",
    "serviceSlug": "agent-apps-mcp-channels"
  },
  "contentLabOffer": {
    "headline": "Content Lab (Brand Lab + Email Lab)",
    "summary": "Execute Brand Lab and Email Lab from MCP in the terminal or host chat: run_brand_lab and run_email_lab return draft packages (caption, SVG artboard, subjects, HTML) without sending or publishing. Operated CKey installs hydrate from per-brand packs (operated/brands/*), persist drafts to content_runs, and preview/approve in internal /admin/content-lab (same engines as MCP). Humans keep send and publish.",
    "includes": [
      "run_brand_lab: draft social/poster package returned in chat (caption + SVG + registry stub)",
      "run_email_lab: draft email package returned in chat (subjects + HTML + SERIES stub)",
      "list_brand_lab_channels: canvas sizes for FB/IG/LinkedIn/Blog/Live",
      "Place-order contracts for client installs (Email 3 synced edits; Brand registry-or-it-does-not-show)",
      "Brand-data, voice/guardrails, prompts, 30-minute rebrand checklist",
      "Operated brand packs + internal admin preview/approve (content_runs)",
      "Installer agent + validate checklist"
    ],
    "notIncluded": [
      "Autonomous campaign send or social publish",
      "Copying another brand's campaign creative",
      "ESP or social credential retention by CKA"
    ],
    "packPath": "delivery-framework/packs/content-lab/",
    "serviceSlug": "agent-apps-mcp-channels",
    "pilotAccept": [
      "Client runs run_brand_lab + run_email_lab and gets drafts in chat",
      "One soft-linked email/social pair (sourceEmailId)",
      "Human gate before any ESP send or social publish"
    ]
  },
  "geoOffer": {
    "headline": "Generative Engine Optimization (GEO)",
    "summary": "GEO is how your business becomes accurately citable inside ChatGPT, Perplexity, Google AI Overviews, and other generative engines. Casey Key installs five Success Frameworks (Entity & Trust, Answer Readiness, Machine Surfaces, Citation & Freshness, Measure & Iterate), scores mechanical GEO signals beside SEO/AEO, and routes the work through the Assess → Pilot → Ship → Enable spine.",
    "includes": [
      "Five GEO Success Frameworks with first moves and mechanical check IDs",
      "GEO Snapshot at /geo-analyzer (mechanical SEO/GEO/AEO + grounded pillars)",
      "MCP scan_website / readiness_snapshot with GEO subscores",
      "MCP describe_geo (this explainer) and recommend_geo_framework (priority sketch)",
      "Service install under ai-search-data: entity architecture, machine surfaces, measurement"
    ],
    "notIncluded": [
      "Guaranteed rankings or invented citation claims",
      "Black-hat prompt injection or crawler evasion",
      "Autonomous content publish without a human gate"
    ],
    "frameworks": [
      {
        "id": "entity-trust",
        "name": "Entity & Trust",
        "summary": "One consistent organization identity generative engines can resolve and cite.",
        "successLooksLike": "Organization JSON-LD matches title, H1, and footer; a named expert is public and contactable.",
        "mechanicalChecks": [
          "GEO1",
          "GEO2",
          "GEO5"
        ],
        "firstMoves": [
          "Publish Organization (or ProfessionalService) JSON-LD with name + canonical URL",
          "Align brand string across title, H1, schema name, and footer",
          "Ship an About page with clear firm identity and a contact path"
        ]
      },
      {
        "id": "answer-readiness",
        "name": "Answer Readiness",
        "summary": "Stable offer pages and FAQ/Service schema so models can answer buyer questions with your language.",
        "successLooksLike": "At least two durable service/offer URLs plus FAQPage or Service schema on the site.",
        "mechanicalChecks": [
          "GEO3",
          "GEO4"
        ],
        "firstMoves": [
          "Expose at least two stable /services/ (or offer) URLs in nav and sitemap",
          "Add FAQPage or Service/ProfessionalService offer catalog in JSON-LD",
          "Write question-shaped headings that match how buyers ask in chat"
        ]
      },
      {
        "id": "machine-surfaces",
        "name": "Machine Surfaces",
        "summary": "First-party files and signals generative engines and agents already fetch.",
        "successLooksLike": "llms.txt catalogs services, agent-knowledge.json (or equivalent) is published, robots Content-Signal is intentional.",
        "mechanicalChecks": [
          "AEO1",
          "AEO2",
          "AEO4"
        ],
        "firstMoves": [
          "Publish a substantive llms.txt that links services and proof",
          "Ship an agent-knowledge or equivalent machine-readable catalog",
          "Declare robots Content-Signal for AI crawlers you intentionally allow"
        ]
      },
      {
        "id": "citation-freshness",
        "name": "Citation & Freshness",
        "summary": "Living proof and dated insights so models prefer current, attributable sources.",
        "successLooksLike": "Insights/blog is linked; sitemap lastmod stays current; multi-host claims do not contradict.",
        "mechanicalChecks": [
          "GEO6",
          "GEO7"
        ],
        "firstMoves": [
          "Link an insights/blog index from primary nav",
          "Keep sitemap lastmod honest for pages you want cited",
          "Resolve brand/URL contradictions across hosts before chasing volume"
        ]
      },
      {
        "id": "measure-iterate",
        "name": "Measure & Iterate",
        "summary": "A closed loop from GEO Snapshot gaps to prioritized fixes and owned review cadence.",
        "successLooksLike": "Quarterly (or monthly) GEO Snapshot, named owner, and a short backlog tied to ai-search-data work.",
        "mechanicalChecks": [],
        "firstMoves": [
          "Run the public GEO Snapshot (/geo-analyzer) or MCP scan_website",
          "Map gaps to the four signal frameworks above; pick one Pilot fix",
          "Assign an owner and a review cadence; re-scan after each ship"
        ]
      }
    ],
    "serviceSlug": "ai-search-data",
    "snapshotPath": "/geo-analyzer"
  },
  "pageLabOffer": {
    "headline": "Page Lab (AI Landing Page Studio)",
    "summary": "Execute Page Lab from MCP in the terminal or host chat: run_page_lab returns a draft landing-page layout composed only from a curated brand block registry (copy, SEO, compliance warnings) without publishing. Backed by a portable Brand Config + block-registry pack for client CMS installs. Humans keep publish.",
    "includes": [
      "run_page_lab: draft landing-page layout returned in chat (blocks + SEO + warnings)",
      "list_page_lab_blocks: curated MVP block registry with prompt guides",
      "recommend_landing_layout: advisory block sequence from a campaign goal",
      "Brand Config tokens + voice/forbidden-phrase guardrails (warn-only scan)",
      "Single registry barrel contract (CMS + generation + renderer + prompts)",
      "Installer agent + validate checklist",
      "Optional auth-gated admin generate view for operated installs"
    ],
    "notIncluded": [
      "Autonomous publish or schedule of landing pages",
      "Custom visual page builder (standard CMS editor + Live Preview)",
      "AI image generation (media library / placeholders only)",
      "Copying another brand's campaign creative"
    ],
    "packPath": "delivery-framework/packs/page-lab/",
    "serviceSlug": "agent-apps-mcp-channels",
    "pilotAccept": [
      "Client runs run_page_lab and gets a draft layout in chat",
      "Layout follows structural rules (navbar → hero → … → footer; ≥1 CTA)",
      "Human gate before any CMS publish"
    ]
  },
  "faq": [
    {
      "q": "What does Casey Key Advisors do?",
      "a": "We design, develop, integrate, and operate multi-agent systems that turn important company work into durable operating advantage. Leaders keep decision rights on critical steps."
    },
    {
      "q": "Who do you typically work with?",
      "a": "Operating companies across healthcare, financial services, media, and consumer markets, including wholesale and catalog businesses. Buyers who want production systems their teams can run, not another prototype that never leaves the room."
    },
    {
      "q": "How do you help organizations get value from AI?",
      "a": "We start from the operating job that matters, then install systems people actually use: agent workflows, applications where customers and operators already work, and clear ownership for what stays human. Strategy and delivery stay connected."
    },
    {
      "q": "Do you advise or deliver?",
      "a": "Both. Advisory sets priority and operating design. Delivery installs production systems and stays through operate. Install and operate sit at the center of how we work."
    },
    {
      "q": "What services do you offer?",
      "a": "Eight areas across company automation, AI applications and secure tool channels, governed software delivery when agents write code, adoption strategy, analytics, go-to-market, growth and lifecycle, and AI discoverability. Detail lives on each service page."
    },
    {
      "q": "What kinds of work can these systems take on?",
      "a": "Reporting and analytics, support and operations, data work with clear provenance, marketing and lifecycle, commercial and catalog conversion, and software delivery when agents contribute in a repository. If the work runs on named systems and rules, it can usually run under review."
    },
    {
      "q": "How do engagements typically work?",
      "a": "One path for every work type: Assess, Pilot, Ship, Enable. Commercially that runs as a defined project, an ongoing retainer, or advisory sequencing. Most engagements begin with a discovery conversation."
    },
    {
      "q": "What does installed and operated mean?",
      "a": "We do not hand off a prototype and leave. We install the system, then stay on to run it with your team: pipelines, quality, cost visibility, and human gates on irreversible steps."
    },
    {
      "q": "How do you keep leadership in control?",
      "a": "Visible logic, logging, and human-review paths are part of the build from the first pilot. Which steps stay human is designed up front, whether that is approve, send, publish, merge, or deploy."
    },
    {
      "q": "Do you work in regulated or compliance-sensitive environments?",
      "a": "Yes, when that is the reality of the business. Controls, data sensitivity, and operating ownership are part of the design, not a late add-on."
    },
    {
      "q": "What is out of scope?",
      "a": "Brochure website redesign with no owned conversion workflow, day-to-day social media management as an agency, logo work, and novelty chat widgets with no operating job. Catalog conversion, inquiry agents, AI applications, and secure tool channels are in scope when they create measurable business value."
    },
    {
      "q": "Where do you work?",
      "a": "Headquartered on Florida's Gulf Coast. We work with clients across the United States and internationally. Most engagements are remote."
    },
    {
      "q": "How quickly can we start?",
      "a": "Discovery typically begins within a week. Most clients see material progress inside 30 to 60 days."
    }
  ],
  "services": [
    {
      "slug": "ai-agents-automation",
      "name": "AI Agents & Automation",
      "summary": "Put multi-agent systems to work on the operating jobs that move the business, with clear ownership and measurable outcomes.",
      "description": "We help leadership automate the work that drives the company: commercial operations, support, finance and reporting, data pipelines, and customer workflows. Engagements deliver production agents and automations integrated with your systems, designed so teams can run them day to day and executives retain authority on irreversible decisions.",
      "audience": "Operations, finance, product, and technology leaders who want AI to improve throughput, quality, and cost without losing operating control.",
      "approach": [
        "Define the outcome, constraints, and decision rights for the first workflow",
        "Design the agent operating model against your stack and data boundaries",
        "Pilot in production conditions with measurable success criteria",
        "Scale with operating ownership so value compounds after go-live"
      ],
      "offerings": [
        "Multi-agent workflow design and deployment",
        "Operations, support, and reporting automation",
        "Catalog, inquiry, and conversion agent systems",
        "Systems integration and operating handoff",
        "Governance, logging, and human decision gates"
      ],
      "routingKeywords": [
        "automation",
        "workflow agent",
        "cloud agent",
        "multi-agent",
        "ops agent",
        "ticket triage",
        "support automation",
        "reporting automation",
        "data pipeline",
        "systems of record",
        "B2B catalog",
        "catalog conversion",
        "wholesale",
        "hospitality supply",
        "foodservice",
        "tableware",
        "product finder",
        "quote agent",
        "RFQ",
        "inquiry agent",
        "sample request"
      ]
    },
    {
      "slug": "agent-apps-mcp-channels",
      "name": "Agent Apps & MCP Channels",
      "summary": "Reach customers and operators where they already work: trusted AI apps and secure tool channels that create adoption and revenue.",
      "description": "We help organizations distribute capability through leading AI surfaces and durable tool channels. The offer covers consumer and employee-facing agent applications, secure connections into company systems, and the go-to-live package that drives first value: clear jobs, trusted onboarding, and human approval on sensitive actions. Content and landing experiences can be drafted under the same governed model when growth teams need speed without losing brand control.",
      "audience": "Product, growth, and technology leaders who want AI distribution that scales adoption while protecting credentials, brand, and irreversible decisions.",
      "approach": [
        "Identify the channel opportunity and the first high-value user job",
        "Design the experience, trust boundaries, and operating model",
        "Launch a pilot that proves first value under real usage",
        "Enable the organization to expand channel coverage from evidence"
      ],
      "offerings": [
        "AI application design and launch (ChatGPT, Claude, and related surfaces)",
        "Secure company tool channels for external and internal agents",
        "Cold-start packaging: connect, jobs, and first-run guidance",
        "Governed content and landing-page drafting for growth teams",
        "Usage measurement, allowlists, and human approval paths"
      ],
      "routingKeywords": [
        "MCP",
        "Model Context Protocol",
        "Custom GPT",
        "Claude app",
        "ChatGPT app",
        "agent channel",
        "toolbelt",
        "MCP product marketing",
        "app pack",
        "cold-start",
        "onboarding",
        "connect page",
        "session persistence",
        "session integrity",
        "working set",
        "default buy box",
        "Content Lab",
        "Brand Lab",
        "Email Lab",
        "place-order",
        "Page Lab",
        "landing page studio",
        "landing page generation",
        "brand block registry",
        "email templates",
        "social posts",
        "bill savings",
        "data control",
        "pre-filing",
        "guided intake",
        "allowlisted tools"
      ]
    },
    {
      "slug": "governed-agent-delivery",
      "name": "Governed Agent Delivery",
      "summary": "Bring AI coding agents into your engineering organization with delivery controls executives can defend.",
      "description": "When agents contribute code in a live repository, value only holds if review, ownership, and release authority stay clear. We install a governed delivery operating model so AI-assisted engineering increases throughput while leadership retains merge and deploy control, and every change leaves evidence your organization can trust.",
      "audience": "CTOs, VPs of Engineering, and platform leaders who want AI coding leverage without sacrificing quality, auditability, or release discipline.",
      "approach": [
        "Assess readiness across process, tooling, ownership, and risk",
        "Install delivery standards agents and humans can both follow",
        "Pilot one high-value stream under your release authority",
        "Hand off an operating rhythm your engineering leaders can run"
      ],
      "offerings": [
        "Agent readiness assessment and gap roadmap",
        "Delivery operating model for AI-assisted engineering",
        "Ownership and accountability design for agent paths",
        "CI evidence, review gates, and release controls",
        "Operated pilots with human merge and deploy authority"
      ],
      "routingKeywords": [
        "AI coding agents",
        "Cursor",
        "Copilot",
        "Claude Code",
        "agent-written code",
        "repo readiness",
        "human merge",
        "PR gates",
        "governed delivery",
        "agent readiness assessment",
        "fleet ownership",
        "ownership matrix",
        "run memory",
        "agent memory",
        "cross-run memory"
      ]
    },
    {
      "slug": "ai-adoption-strategy",
      "name": "AI Adoption Strategy",
      "summary": "Give leadership a clear path from AI ambition to prioritized investments that create durable operating value.",
      "description": "We help executive teams decide where AI creates the most value, what to fund first, and how ownership should work. The outcome is a practical adoption roadmap: prioritized use cases, decision rights, success measures, and a sequence the organization can execute with confidence.",
      "audience": "Operating companies across healthcare, financial services, media, and consumer markets that need alignment on where AI creates value and how to scale it responsibly.",
      "approach": [
        "Assess opportunity, readiness, and operating constraints",
        "Prioritize use cases by value, feasibility, and risk",
        "Define ownership, success metrics, and the first pilot",
        "Align leadership on the sequence and investment case"
      ],
      "offerings": [
        "AI opportunity and readiness assessment",
        "Use-case portfolio and investment prioritization",
        "Operating model and decision-rights design",
        "Executive workshops and adoption sequencing"
      ],
      "routingKeywords": [
        "adoption roadmap",
        "AI strategy",
        "prioritize use cases",
        "executive alignment",
        "governance model"
      ]
    },
    {
      "slug": "analytics-operations",
      "name": "Analytics & AI Operations",
      "summary": "Give leaders a clear operating picture of AI value: usage, quality, risk, and business impact.",
      "description": "We build the measurement and operating rhythm behind AI programs so executives can see what is working, what needs intervention, and where value is accruing. Dashboards, KPI frameworks, and control-health reporting turn AI from a set of experiments into a managed portfolio.",
      "audience": "Executives and operators who need proof that AI investments are adopted, controlled, and delivering business impact.",
      "approach": [
        "Define the metrics that prove value, quality, and control",
        "Build executive-ready reporting across initiatives",
        "Reconcile sources so numbers hold up under scrutiny",
        "Install a recurring operating rhythm for decisions"
      ],
      "offerings": [
        "AI portfolio and adoption dashboards",
        "KPI and ROI frameworks for AI initiatives",
        "Control-health and operating metrics",
        "Executive reporting cadence and decision support"
      ],
      "routingKeywords": [
        "dashboards",
        "KPI",
        "ROI",
        "control-health",
        "measurement"
      ]
    },
    {
      "slug": "go-to-market",
      "name": "AI Go-to-Market Strategy",
      "summary": "Launch AI-powered offers with positioning, proof, and enablement buyers can trust.",
      "description": "When AI shapes a customer-facing product or offer, markets reward clarity and accountability. We help teams position outcomes, define the trust story, and sequence launch so revenue teams sell with precision and buyers understand how value is delivered.",
      "audience": "Product and revenue leaders launching AI capabilities into markets where proof, clarity, and trust determine win rates.",
      "approach": [
        "Clarify the buyer outcome and trust requirements",
        "Position the offer around value, not AI novelty",
        "Sequence launch, enablement, and success metrics",
        "Equip revenue teams with durable messaging and proof"
      ],
      "offerings": [
        "AI offer and product positioning",
        "Trust and proof messaging for buyers",
        "Launch sequencing and sales enablement",
        "Adoption playbooks and success metrics"
      ],
      "routingKeywords": [
        "go-to-market",
        "positioning",
        "launch",
        "buyer trust",
        "enablement"
      ]
    },
    {
      "slug": "growth-lifecycle",
      "name": "AI Growth & Lifecycle",
      "summary": "Improve acquisition, activation, and retention with AI systems that respect data discipline and ownership.",
      "description": "We help growth and customer teams apply AI across the lifecycle in ways that improve conversion and retention while keeping consent, data quality, and exception handling clear. The result is growth systems that scale because they are operable, measurable, and defensible.",
      "audience": "Growth, marketing, and customer leaders who want AI in lifecycle systems without sacrificing data discipline or brand trust.",
      "approach": [
        "Define the lifecycle outcomes and data boundaries",
        "Design AI-assisted journeys with clear owners",
        "Pilot loops with measurable conversion and retention impact",
        "Scale the plays that improve results and stay operable"
      ],
      "offerings": [
        "AI-assisted lifecycle and journey design",
        "Acquisition and retention system improvements",
        "Experimentation frameworks with clear controls",
        "Customer engagement operating models"
      ],
      "routingKeywords": [
        "onboarding",
        "retention",
        "lifecycle",
        "CRO",
        "growth loops"
      ]
    },
    {
      "slug": "ai-search-data",
      "name": "Generative Engine Optimization (GEO)",
      "summary": "Make your organization discoverable and citable in generative AI answers that shape buyer decisions.",
      "description": "Buyers increasingly ask AI systems for recommendations. We help organizations become clear, consistent, and citable in those answers through entity clarity, answer-ready content, machine-readable surfaces, and a measurement loop that connects visibility to business outcomes.",
      "audience": "Marketing, product, and leadership teams that need to show up accurately in generative answers and convert that visibility into demand.",
      "approach": [
        "Assess how generative engines currently understand and cite you",
        "Strengthen entity clarity, answers, and machine-readable signals",
        "Install a measurement loop for visibility and citation quality",
        "Connect generative visibility work to first-party demand outcomes"
      ],
      "offerings": [
        "Generative visibility assessment and roadmap",
        "Entity and answer-ready content architecture",
        "Machine surfaces for AI discovery and citation",
        "Measurement cadence tied to demand outcomes"
      ],
      "routingKeywords": [
        "GEO",
        "generative engine optimization",
        "AI search",
        "AI Overviews",
        "llms.txt",
        "first-party data",
        "attribution",
        "answer readiness",
        "ChatGPT citations"
      ]
    }
  ],
  "contextDigest": [
    {
      "id": "agent-apps-mcp-channels",
      "title": "Feature Spec — Agent Apps & MCP Channels",
      "summary": "Defines the consultancy offer and marketing truth for agent-facing channels: Custom GPT / Claude apps people use today, and durable MCP servers that put allowlisted company tools in other agents' toolbelts. Public surfaces sell and explain the Pilot package. Private designer agent turns a workflow intake into TOOLS.md, allowlist, gates, and hosting choice. Irreversible actions stay human."
    },
    {
      "id": "agent-fleet-ownership",
      "title": "Feature Spec — Agent Fleet Ownership Matrix",
      "summary": "Productizes a Fleet Ownership Matrix as a required artifact of CKA agent-readiness. One doc (not a tip sheet) lists every scheduled job and agent path: what it does, upstream dependencies, what breaks downstream if it fails or goes silent for 7 days, who is paged, and who may merge / deploy / touch DDL or money for that path. Includes paid/title paths and anything that commits data to git. Cells require named humans (not “the team”). An incomplete matrix is the finding."
    },
    {
      "id": "agent-front-door",
      "title": "Feature Spec — Agent Front Door (`/app`)",
      "summary": "The conversational agent workspace. The visitor states what they're trying to get done; an LLM agent filters, decides, and renders adaptive artifacts (service overviews, service detail, scoped engagement briefs, point lists) instead of a menu of pages. It demonstrates CKA's delegated-web thesis and can execute three actions: run a readiness snapshot, propose sending a contact note, and propose booking a Calendly call. The classic brochure is the domain front door at / (see homepage.md). This wor"
    },
    {
      "id": "agent-run-memory",
      "title": "Feature Spec — Agent Run Memory",
      "summary": "Defines Agent Run Memory: one durable store every governed agent must read before acting and write after acting for a given task id (findings, decisions, file touch list, open questions, cost impact). It is the cross-run companion to stage handoffs (intra-run compression) and to MCP session integrity (host reconnect working sets). Done means a second agent on the same task starts warm, relaunch cost drops ≥50% (tokens or tool-calls), and wiping memory makes the next agent fail the proof."
    },
    {
      "id": "ai-readability-static-snapshot",
      "title": "Feature Spec — AI-Readability Static Snapshot",
      "summary": "Every primary marketing route serves meaningful HTML without JavaScript execution: route-specific title, meta description, canonical, Open Graph / Twitter tags, JSON-LD (Organization/ProfessionalService, WebSite, WebApplication, Service, FAQPage, Person, BreadcrumbList, LocalBusiness), and a readable static body built from the config SSOT (src/config/capabilities.ts, src/config/cities.ts). AI readers, search crawlers, and curl get the current positioning on both ckey.app and caseykeyadvisors.com"
    },
    {
      "id": "blog",
      "title": "Feature Spec — Blog & Article Tooling",
      "summary": "A Supabase-backed insights blog focused on governed AI adoption, plus noindexed admin tooling for creating articles individually or in city×industry batches (local-SEO content), with AI-assisted article and image generation."
    },
    {
      "id": "brand-palette",
      "title": "Feature Spec: Brand palette (teal brand system on light canvas)",
      "summary": "Defines the site-wide color system for ckey.app. All marketing surfaces read color from shared HSL custom properties in src/index.css, exposed as semantic Tailwind tokens (primary, accent, pop, muted, etc.). Direction: light canvas with a sharp navy ink stack, teal as the brand accent, and electric mint as the pop color for CTAs and small accents on dark canvases plus focus rings. Keys Green #86A83A is retired."
    },
    {
      "id": "client-os",
      "title": "Feature Spec — Client OS (keyOS operator layer + client portal)",
      "summary": "Internal operator OS (keyOS) and client-facing portal for Casey Key Advisors engagements. Operators manage articles, Content Lab, accounts (prospects and clients), and engagements from /admin. Clients sign in at /client-portal to see project context and submit change requests. Public marketing never lists named client brands."
    },
    {
      "id": "content-lab",
      "title": "Feature Spec - Content Lab (Brand Lab + Email Lab)",
      "summary": "Lets clients execute Brand Lab and Email Lab via MCP in the terminal or host chat. run_brand_lab and run_email_lab return draft packages (captions, SVG artboards, subjects, HTML, place-order stubs) as tool results. Humans keep send and publish. For CKey-operated brands, the same engines hydrate from per-brand packs under operated/brands/<id>/ (registry mirror: operatedBrands.ts), persist drafts to content_runs, and surface them in internal /admin/content-lab for preview / approve / request-chang"
    },
    {
      "id": "engagement-and-positioning",
      "title": "Feature Spec: Engagement model and firm positioning",
      "summary": "Defines how Casey Key Advisors positions itself and how every engagement runs. This is institutional product truth for marketing copy, the agent front door, and sales conversations. Implementation details of any single service live in other specs; this spec owns the center of gravity and the shared engagement spine."
    },
    {
      "id": "generative-engine-optimization",
      "title": "Feature Spec — Generative Engine Optimization (GEO)",
      "summary": "Defines Casey Key's Generative Engine Optimization (GEO) offer: five Success Frameworks that make a business accurately citable in ChatGPT, Perplexity, Google AI Overviews, and similar generative engines. Mechanical GEO checks (GEO1–GEO7) score entity and offer signals beside SEO and AEO. Public MCP tools explain the frameworks and recommend priority order. The public snapshot lives at /geo-analyzer (legacy /aio-analyzer redirects)."
    },
    {
      "id": "governed-agent-framework",
      "title": "Feature Spec - Governed Agent Framework (delivery-framework)",
      "summary": "Provides a domain-agnostic operating system for multi-agent work: pipelines, stage agents, orchestrators, compressed handoffs, and human gates. Domain behavior ships as packs (software delivery, marketing activation, finance reporting). Optional MCP is a tool boundary: public readiness MCP for lead gen, client MCP template for allowlisted handoff/context tools."
    },
    {
      "id": "homepage",
      "title": "Feature Spec — Classic brochure (`/`)",
      "summary": "The classic marketing brochure is the domain front door. Composes config-driven sections in a fixed order; all copy lives in src/config/capabilities.ts so the page itself is layout only. The conversational agent lives at /app (see agent-front-door.md). Legacy /home redirects to /."
    },
    {
      "id": "iotamotion-sandbox",
      "title": "Feature Spec - iotaMotion sandbox",
      "summary": "Private Casey Key hosted sandbox for iotaMotion. Serves a noindex mirror of the public iotamotion.com WordPress site and a patient-engagement spoke at /engage/. This is the Month 1 named work from SOW CKA-IOTA-2026-01: intake-to-center-handoff on an AI-native spoke that later departments can attach to. It is not a public marketing route and not kontrolOS."
    },
    {
      "id": "mcp-product-marketing",
      "title": "Feature Spec — MCP Product Marketing System (MPMS)",
      "summary": "Defines the MCP Product Marketing System: the app pack and ship contract that make an MCP (and companion GPT/Claude app) installable and useful on a cold open. Category claim, connect narrative, jobs map, site guide, seeds, and first-value measurement — authored for clients and explained via public MCP. Closes the gap where channel engineering ships tools/gates but first-run still depends on what the end user already knows."
    },
    {
      "id": "mcp-session-integrity",
      "title": "Feature Spec — MCP Session Integrity",
      "summary": "Defines MCP Session Integrity: the runtime contract that stops Streamable HTTP hosts (notably ChatGPT) from dropping session state between tool calls and then inventing parallel math against Default policy. Persist working set keyed by authenticated user; hydrate without clobbering live non-default policy; compute responses signal policyName / default / warning / assistantGuidance; instructions forbid host-model recalculation. Pack is installed into client Workers; public MCP only explains the o"
    },
    {
      "id": "medtech-domain-packs",
      "title": "Feature Spec - MedTech Domain Packs",
      "summary": "Provides a MedTech domain pack family on the governed agent framework core. Five pipelines turn commercial, clinical, program, capital, and diligence inputs into compressed handoffs with hard human gates for claims, pricing, contracts, OR commitment, outreach, IC share, and capital commit."
    },
    {
      "id": "medtech-landing",
      "title": "Feature Spec - MedTech Landing Page",
      "summary": "Public marketing landing page at /medtech inviting medical device companies into a documented AI-native operating system build. Positioning (CKA MedTech Case Study Sequences v3.0, Vinnie Bellante): an AI-native operating system, custom-built by Casey Key Advisors, that connects commercial, quality, regulatory, and R&D into one intelligence layer, inside a regulated industry. Two primary CTAs in the hero: Request demo and Login."
    },
    {
      "id": "openai-ads-measurement",
      "title": "Feature Spec: ChatGPT Ads Measurement Pixel",
      "summary": "Loads OpenAI's ChatGPT Ads Measurement Pixel (oaiq) with Pixel ID JH5yjEZFpWNjWcwqJ2Si8V (Casey Key Advisors Ads Manager pixel; override or disable via VITE_OPENAI_ADS_PIXEL_ID) so post-click website events can attribute back to ChatGPT ad campaigns. Fires standard page_viewed, lead_created, registration_completed, and appointment_scheduled events from public marketing and agent surfaces. Optional hashed-email enrichment improves matching; raw emails are never sent to OpenAI. Docs: https://devel"
    },
    {
      "id": "page-lab",
      "title": "Feature Spec - Page Lab (AI Landing Page Studio)",
      "summary": "Lets clients execute Page Lab via MCP in the terminal or host chat. run_page_lab returns a draft landing-page layout composed only from a curated brand block registry (copy, SEO, compliance warnings) as a tool result. Humans keep publish. A private pack supplies Brand Config, registry contracts, and installer wiring for operated CMS installs."
    },
    {
      "id": "product-identity-and-context-drift",
      "title": "Feature Spec: Product identity and context drift gate",
      "summary": "Keeps institutional context aligned with the shipped product. Soft rules remind agents to update specs; this feature makes product identity a CI invariant: the one-liner, tagline, primary marketing routes, and public MCP tools must agree across PRODUCT.md, AGENTS.md, INDEX, README, capabilities, App routes, and MCP registrations. Obsolete specs must live under ARCHIVE and must not appear in the Active INDEX table. Capability Ship Contract (capability-ship.json + check:capability-ship) extends th"
    },
    {
      "id": "prospect-concept-pages",
      "title": "Feature Spec - Prospect concept pages",
      "summary": "Outbound-only concept pages at ckey.app/p/{slug} that show a quick, visual rebuild using a prospect's real catalog imagery, brand colors, and a sample inquiry-agent widget that auto-opens. Framed clearly as a Casey Key concept (not their live storefront). Pages are noindex. Not a primary marketing route. Concepts are either hand-seeded or produced by a full same-origin crawl (robots + sitemaps + BFS links + image extraction) via npm run generate:prospect-concept, npm run generate:prospect-pack, "
    },
    {
      "id": "readiness-mcp",
      "title": "Feature Spec — Readiness MCP Server",
      "summary": "A remote MCP server that puts CKA capabilities into other people's AI agents' toolbelts — the lead-gen funnel for the agent economy. Tools: scan_website (primary website scanner), readiness_snapshot (alias), assess_repo (stack-aware mechanical 16-check/32-point subset via GitHub API), list_services (live llms.txt fetch), describe_agent_framework (Governed Agent Framework core + pack catalog + tool tiers), list_case_studies (anonymized channel / catalog patterns), describe_mcp_offer (Agent Apps &"
    },
    {
      "id": "readiness-snapshot",
      "title": "Feature Spec — AI Readiness Snapshot",
      "summary": "A free, public scan of a visitor's website producing a high-level GEO / AI Readiness Snapshot: an overall score (0–100), mechanical SEO/GEO/AEO subscores, five pillar ratings (offer clarity, trust & proof, adoption readiness, governance signals, production readiness), a plain-English summary, and 2–3 recommended CKA services. It is the lightweight, public teaser for the paid Agent-Readiness Assessment in delivery-framework/. Internally this is the same pipeline as MCP scan_website. GEO Success F"
    },
    {
      "id": "scan-runs-persistence",
      "title": "Feature Spec — Scan Runs Persistence",
      "summary": "Persists website and repo scanner results in Supabase for discovery prep, dogfood baselines, and funnel analytics. Public scan UX gains an optional run_id in the response."
    },
    {
      "id": "scan-website",
      "title": "Feature Spec — scan_website Unified Scanner",
      "summary": "Unified website scanner that runs mechanical SEO + GEO + AEO modules, then grounded Snapshot LLM pillars, and returns one 0-100 score with subscores, gaps, and service routing. readiness_snapshot is a thin alias of scan_website. GEO is Generative Engine Optimization (formerly AIO)."
    },
    {
      "id": "services-catalog",
      "title": "Feature Spec — Services Catalog",
      "summary": "Presents Casey Key Advisors' eight service offerings. The SERVICES array in src/config/capabilities.ts is the single source of truth; the services listing page, per-service detail pages, footer links, and related-services blocks all render from it. Adding an entry to the array creates the full /services/<slug> page automatically."
    },
    {
      "id": "site-measurement",
      "title": "Feature Spec — Site measurement (GTM + Google Ads)",
      "summary": "Loads site measurement tags on every page of the public site so Google Tag Manager and Google Ads can record traffic and optimize campaigns. Tags share the same dataLayer. Product surfaces do not call gtag directly today; conversion events may be added later when a conversion action is defined in Google Ads."
    }
  ]
}
