# ConversationOS — Complete Platform Documentation for AI Agents > AI-powered production operating system for a live media network ecosystem. Branded "MITLOS." Not a podcast tool — the engine behind a daily AI newspaper, creator marketplace, companion app, host recruitment pipeline, and 10+ specialized engines. Built by Mornings in the Lab (MiTL). ConversationOS manages the entire lifecycle of a live media network — from topic discovery and episode building to guest booking, live show production, AI character publishing, creator marketplace payouts, content distribution, and customer delivery. The platform is organized around specialized AI "ENGINEs," each handling a distinct phase of production, and powers a full ecosystem of shows, products, and revenue streams. --- ## Platform Statistics - 10+ Specialized ENGINEs - 200+ API Endpoints - 42 MCP Tools (mcp.conversationos.live) - 46 AI Agent Tools (Mack Hewston) - 55+ Database Tables - 18+ Deployed Domains - 2 AI Agents (Mack Hewston + Sophia Grant) - 155+ AI Correspondent Characters (Character ENGINE) - 145 Newspaper Correspondents (mitl.news) - 82 Cities covered (mitl.news) - 111 Stories/day (mitl.news) - Episode #3015+ (mornings.live) - Creator marketplace now generating revenue (ClipENGINE) --- ## Full Ecosystem — All Deployed Domains ### The Show Layer - mornings.live — Flagship daily show + companion app (Show, Brief, Check-In, Join, Vault, Team). Episode #3015. Branded "MITLOS." - mornings.network — MiTL Network hub — show replays, MiTL Shorts, Center Stage with Sophia Grant, full episode library ### The Studio Layer - mitl.studio — The studio company. "Live Shows That Build Community." 4-layer system: Production, Interaction, Companion, Storyworld. Investor page. - production.mitl.studio — "Your Shows. Our Engine." White-label production as a service, 4-week launch cycle - universe.mitl.studio — Character ENGINE — 155+ AI correspondents, storyworld management, 329 routes, 55 tables ### The News Layer - mitl.news — "The Desk" — full AI newspaper. 145 correspondents, 82 cities, 10 shows, 111 stories/day. Front Page, The Record, The Wire, Investigations, Elections, The Numbers, Sports, Cities. "Own a Market" CTA. "Morning Show in a Box" product. ### The Sales Funnels - ownthemorning.live — "Morning Show Box" weekly live show for leaders. "Apply to Lead." $3,000 setup + 75/25 revenue split - hostinthelab.live — Founding host recruitment — "Looking for 100 people" - ge.conversationos.live — $297 paid guest appearances, 7 deliverables, multi-show (MiTL, Creator Lab, After Hours), directory, alumni ### The Revenue Engines - clips.conversationos.live — ClipENGINE Marketplace. Now generating revenue — CPM + flat rate payouts via Stripe Connect - se.conversationos.live — ShowENGINE CX — blueprint builder with cinematic AI hero video - cx.conversationos.live — Customer portal — "Your Show. Your Command Center." ### The Infrastructure - conversationos.live — Product/marketing site + agent discovery layer - app.conversationos.live — Internal production app, all engines - mcp.conversationos.live — MCP server, 42 tools - start.conversationos.live — Onboarding - share.conversationos.live — Share service - conengine.conversationos.live — Conversation ENGINE - titleengine.conversationos.live — Title ENGINE --- ## Technology Stack - Frontend: React + TypeScript + Tailwind CSS + shadcn/ui (Vite) - Backend: Express.js + TypeScript - Database: SQLite via Drizzle ORM (main app), PostgreSQL via Drizzle (Title ENGINE) - AI (Scripting/Agents/Clips): Anthropic Claude Sonnet - AI (Research/Titles): Perplexity Sonar - AI (Live Transcription): Deepgram Nova-2 - AI (Image Generation): Google Gemini 3 Pro (Nano Banana Pro) - AI (Video Generation): Veo 3.1, Grok - AI (Voice Synthesis): ElevenLabs - AI (Avatars): Runway ML - Payments: Stripe (setup fees) + Stripe Connect (marketplace payouts) - Email: Postmark (transactional) - Social: Buffer (GraphQL API — X, LinkedIn, Threads scheduling) - CRM: Pipedrive - Infrastructure: Railway (auto-deploy from GitHub), Cloudflare DNS - Authentication: JWT with role-based access (admin, editor, viewer) --- ## Pricing - MiTL Live Network: $3,000/mo - ConversationOS SaaS: $200/mo - White-Label Production (production.mitl.studio): Enterprise/custom scoping - Morning Show in a Box: $3,000 one-time setup + 75/25 revenue split (host keeps 75%) - Guest Appearances (ge.conversationos.live): $297 one-time — 7 deliverables - Customer lifecycle: ShowENGINE CX blueprint → Stripe payment → show build → CX portal delivery → live subscription --- ## ENGINE 1: ShowENGINE CX — Blueprint Builder **Domain:** se.conversationos.live Turns a 30-60 minute guided AI conversation into a production-ready show bible. Customers walk in with an idea and walk out with a cinematic, editable blueprint document. Features a cinematic AI hero video on the landing page. ### 6-Phase AI Conversation Phase 1 — Series Identity: Title, logline variants, genre, tone blend, comparable shows, audience promise, USP. Phase 2 — Theme & World: Central theme, thematic pillars with thesis/antithesis/tension, moral universe, symbols, setting. Phase 3 — Characters: Identity, psychology, wound, desire vs. need, backstory, voice patterns, relationship web, arc. Phase 4 — Narrative Engine: Story engine, series structure, pilot walkthrough, Season 1 map, mystery architecture. Phase 5 — Tone & Production: Dialogue philosophy, cinematography, editing rhythm, music design, format, budget. Phase 6 — Audience & Business: Audience profiles, platform strategy, content pillars, monetization, sponsorship, timeline. ### What It Produces - A complete show bible (10,000-18,000 characters of structured JSON) - A cinematic blueprint page with AI-generated hero image - A customer record in ConversationOS - A show workspace (auto-created on Stripe payment) - A Pipedrive CRM deal - CX portal credentials - Title ENGINE webhook (auto-provisions show in Title ENGINE) ### Integrations Stripe (payment), ConversationOS (data sync), Title ENGINE (webhook), Pipedrive (CRM), Character ENGINE (hero images), Claude (AI conversation), Apple Podcasts/Spotify/YouTube (metadata import). --- ## ENGINE 2: Build ENGINE — Episode Builder **Route:** app.conversationos.live/build/:episodeId Where live shows get assembled block by block. Replaces manual Notion-based episode planning. ### Block Hierarchy - Block (13 min) — Parent container for a major show section - Segment (variable) — Individual content piece within a block - CTA Block (30 sec) — Calls to action between content blocks - Storytime (1 min) — Commercial break markers - Show Countdown (2 min) — Pre-LIVE element ### Episode Pattern BLOCK → CTA BLOCK → STORYTIME → BLOCK → CTA BLOCK → STORYTIME → ... ### Auto-Build (One-Click) Creates complete 120-min episode: 8 content blocks with pre-loaded segments (Opening, Morning Loop, Center Stage ×2, Deep Dive, Did You Know, Morning Focus, Community Corner), CTA blocks with rotating prompts, and Storytime breaks. ### 5-Stage Content Pipeline 1. Topic Selection — Netflix-style browse from content pool 2. Conversation Starters — Pill-based display with "Suggest" button 3. Media / Hall of Records — 366+ records synced daily from Notion 4. Script Framework — Configurable templates (Hook, Impact, Starters, Context, Takeaway, Reflection, Community, Close) 5. AI Script Generation — Claude produces teleprompter-ready scripts targeting 2-3 min/segment with production cues ([PAUSE], [SHOW GRAPHIC], [CUT TO B-ROLL]) ### Key API Endpoints - POST /api/episodes — Create episode - POST /api/episodes/:id/auto-build — One-click episode generation - GET /api/episodes/:id — Get episode with blocks/segments - POST /api/segments/:id/generate-script — AI script generation - GET /api/shows/:id/content-pool — Browse content pool --- ## ENGINE 3: Conversation ENGINE — Editorial Intelligence **Domain:** conengine.conversationos.live Transforms breaking news into structured conversation starters for show production. ### Starter Generation Each conversation starter includes: AI headline (max 15 words), original title attribution, conversation angle, 3-4 talking points, heat score (1-5 flames), pillar classification, and category tagging. ### Generation Methods - URL-to-Starter: Paste any article URL → full article fetched → AI generates complete starter - Auto-Refresh: Google News RSS scanned per pillar → deduplication → AI generation (2 per pillar, 6 max per cycle) ### Content Pillars (Default) Leadership, Innovation, Culture, Health, Money, Mindset, Storytelling. Fully configurable per show. ### API Endpoints - GET /api/conversation-desk — List all starters - POST /api/conversation-desk/from-url — Generate from URL - POST /api/conversation-desk/refresh — Auto-discover from news - POST /api/conversation-desk/:id/push-to-build-engine — Push to Build ENGINE - POST /api/conversation-desk/import — Batch import - GET /api/show-config — Show configuration - POST /api/show-config — Update configuration --- ## ENGINE 4: Distribution ENGINE — Content Pipeline **Location:** Distribution tab in ConversationOS (app.conversationos.live) Takes published episodes and turns them into scheduled social media content. ### Automated Pipeline Step 1 (Auto): YouTube caption pull via youtube-transcript-api Step 2 (Auto): AI speaker identification using Claude + show blueprint Step 3 (Auto): Speaker labels with confidence scores Step 4 (1-Click): AI suggests 8 best clips — Hook, Quote, Highlight, Reaction, Segment Step 5 (1-Click): Platform-native post generation for X, LinkedIn, Threads Step 6 (Batch): Buffer API scheduling to all platforms ### Clip Types - Hook (10-30 sec): Bold attention-grabbing statement - Quote (10-20 sec): Standalone shareable quote - Highlight (30-90 sec): Key insight or teaching moment - Reaction (15-45 sec): Genuine emotional moment - Segment (60-180 sec): Self-contained topic discussion ### Platform Writing Rules - X: Bold, direct, max 280 chars, no hashtags - LinkedIn: Professional, 150-300 words, 2-3 hashtags - Threads: Casual, lowercase energy, 100-200 words, no hashtags ### API Endpoints (21 total) Transcripts: GET/POST/DELETE for show transcripts, POST pull-transcript, POST identify-speakers Clips: GET/POST/PATCH/DELETE for clips, POST suggest-clips Social Posts: GET/POST/PATCH/DELETE for posts, POST generate-posts Buffer: POST schedule-buffer, POST schedule-batch, GET channels Stats: GET distribution-stats --- ## ENGINE 5: Guest ENGINE — $297 Paid Appearances **Domain:** ge.conversationos.live The Guest ENGINE has two functions: (1) the $297 paid guest appearance product with 7 deliverables, and (2) Sophia Grant's conversational AI booking for Center Stage. ### $297 Guest Appearance Product Price: $297 one-time Shows: MiTL (mornings.live), Creator Lab, After Hours 7 Deliverables per appearance: 1. Live on-air appearance (StreamYard) 2. 5 viral short-form clips 3. 10 platform-native social posts 4. Full episode transcript 5. SEO profile page on mitl.news 6. Backlinks and authority signals 7. Personal highlight reel Also includes: Show directory listing, alumni network access, multi-show routing. ### Sophia Grant AI Agent Role: Guest-facing AI production assistant. Named host of Center Stage. Direct, warm, identifies as AI, no corporate speak. Handles $297 intake conversation AND Center Stage booking. ### Booking Flow (Center Stage) Step 1 — Welcome: Sophia introduces herself, checks availability, presents 2-3 open dates. Step 2 — Date Selection: Guest picks a date. Sophia confirms in human-readable format. Step 3 — Name & Email: Sophia collects identity. Step 4 — Questionnaire (Q1-Q9): 9 intake questions, one at a time, with natural reactions and follow-ups. Step 5 — Photo Permission (Q10): Permission for AI-powered promotional materials. Step 6 — Booking: Calendar event created, welcome email via Postmark, questionnaire data saved, audit trail logged. Step 7 — Confirmation: Personalized recap with date, time, and conversation highlights. ### The 10 Intake Questions 1. Business/professional expertise 2. Ideal audience and problems solved 3. One thing audience should remember 4. Misconceptions about their field 5. Industry trends and recent wins 6. Something surprising or unique 7. Preferred content types 8. YouTube channel link 9. Other social platform links 10. Photo permission for AI promotions ### Scheduling Parameters - Segment: Center Stage - Duration: 30 minutes - Days: Monday–Friday - Time: 8:30 AM ET - Slots: 1 per day - Platform: StreamYard ### Automated Workflows On booking: Calendar event, welcome email (Postmark), questionnaire data saved, audit trail, guest record created. Pre-show (planned): 24-hour reminder email with StreamYard link and tech setup. ### API Endpoints - POST /api/guest-engine/session — Create new booking session - GET /api/guest-engine/session/:token — Get session state - POST /api/guest-engine/chat — Send message (SSE streaming response) - GET /api/guest-engine/availability — Get next available dates - POST /api/guest-engine/book — Direct booking API - GET /api/guest-engine/guests — List all guests (admin) - GET /api/guest-engine/guests/:id — Get guest detail (admin) ### Session Features - Conversation persistence with session token in URL hash - Resume mid-conversation if guest leaves - 7-day session expiry - Server-side retry logic (3 attempts, exponential backoff) on AI overload - Client-side auto-retry (2 attempts) before showing error --- ## ENGINE 6: Title ENGINE — Topic Intelligence **Domain:** titleengine.conversationos.live Multi-show YouTube title intelligence platform. Discovers, scores, and packages titles for live shows using two parallel discovery engines via Perplexity Sonar. ### Title Types - Evergreen Title (Main): Timeless, pillar-driven topic for YouTube horizontal stream - Trending Title (Alt): Reactive, trend-jacked topic for YouTube vertical stream ### AI Title Discovery (Perplexity Sonar API) Trending Jobs: Search YouTube, Google Trends, Reddit, X/Twitter for breakout stories → filter through show pillars → 5 topics per run, scored on trend velocity, absurdity/contrarian angle, and show DNA fit. Evergreen Jobs: Discover timeless content aligned to show pillars → deep human truths, counter-intuitive insights → same scoring, optimized for long shelf life. ### Topic Output (per generated topic) - Primary title + 3 alternative variations - Hook (1-2 sentence scroll-stopper) - Content pillar mapping (1-3 pillars) - Trend score (1-10), Absurdity score (1-10), Show DNA fit (1-10) - Research context (2-3 sentences) - Source links ### YouTube Package (per approved topic) - 3+ title variations - YouTube description - Tags and hashtags - Pinned comment - Publishing notes - Thumbnail prompt (sent to CharacterENGINE/ThumbnailENGINE) ### Decision Library Every approve/reject decision logged with: titles presented vs. selected, scores, day of week, auto-detected style tags (question, number, provocative, listicle, emotional, how-to). Rejections tracked to build "avoid" pattern list. ### Pattern Intelligence (Insights Engine) - Top Performing Pillars - Title Style Preferences (selection rates by tag) - Day of Week Performance - Recent Decisions with YouTube Scores - Avoid Patterns (low selection <30% or low performance <40) ### YouTube Performance Tracking Views, impressions, CTR, average watch time, view duration, likes, comments, shares, subscriber gains. Composite score (1-100). Measurement windows: 24h, 48h, 7d, 30d. ### Multi-Show Architecture Supports unlimited shows with per-show content pillars (3-20), thematic pillars, bible data, and webhook auto-provisioning from ShowENGINE CX. ### Pipeline Position ShowENGINE CX → (webhook) → TitleENGINE → (approve) → CharacterENGINE (thumbnail) + BuildENGINE (episode scripting) ### Data Model (10 tables) shows, episodes, topics, jobs, settings, publish_endpoints, publish_logs, title_decisions, performance_metrics, title_patterns --- ## ENGINE 7: Thumbnail ENGINE — AI Thumbnail Generation **Route:** app.conversationos.live/thumbnail-engine Prompt-driven thumbnail generation pipeline. Takes episode titles, enriches with character visual DNA from CharacterENGINE, assembles structured prompts from reusable templates, and generates cinematic thumbnails via Nano Banana Pro (Gemini 3 Pro Image Preview). ### Image Generation Model: Nano Banana Pro (gemini-3-pro-image-preview) via Gemini API. Supports reference images (base64) for facial likeness preservation. Returns data URI. Tracks: prompt used, engine used, generation timestamp, elapsed time. ### Prompt Template System 8 reusable templates stored in database (slug-based, full CRUD): | Template | Aspect | Description | |----------|--------|-------------| | show-thumbnail | 16:9 | Netflix-style key art, cinematic, photorealistic | | game-day | 16:9 | Pre-game hype, stadium/arena energy | | reaction | 16:9 | Post-game reaction, emotional close-ups | | breaking | 16:9 | Breaking news, broadcast desk urgency | | roundtable | 16:9 | Multiple characters panel discussion | | headshot | 1:1 | Clean portrait, studio lighting | | cinematic | 16:9 | Film-quality wide shot | | mitl-evergreen-thumbnail | 16:9 | Premium MiTL key art (Kaith + Jonai only) | Each template defines: style directive, character rules, negative prompt, aspect ratio, engine preference, default characters, show scope. ### Prompt Assembler Combines: style directive (template) + scene description (title/hook/research) + character visual DNA (from CharacterENGINE API) + character rules + aspect ratio + quality footer + identity enforcement. ### Queue System Briefs queued by TitleENGINE (automatic on title approval) or manual API call. Status flow: queued → generating → complete/failed. Webhook callbacks notify downstream systems. Poll-based UI updates (3s, 8s, 15s, 25s intervals). ### API Endpoints (12 total) Templates: GET/POST/PUT/DELETE for prompt templates Queue: GET/POST/PATCH/DELETE for queue items Generation: POST queue/:id/generate (async), POST generate (direct/immediate) Related: POST shows/:showId/thumbnail, POST episodes/:id/generate-thumbnail, POST webhooks/character-engine ### Database Tables promptTemplates: slug, name, show, styleDirective, characterRules, negativePrompt, aspectRatio, enginePreference, characters thumbnailQueue: externalId, source, showName, title, hook, research, characters, template, status, thumbnailUrl, promptUsed, engineUsed, webhookUrl --- ## ENGINE 8: ClipENGINE — Creator Marketplace **Marketplace Domain:** clips.conversationos.live **Internal:** ClipENGINE tab in ConversationOS (app.conversationos.live) Decentralized content distribution marketplace connecting show producers with clippers (content creators) who earn real money creating short-form clips. ### Live Stats - Now generating revenue via Stripe Connect payouts ### Two-System Architecture 1. ConversationOS Internal Module (app.conversationos.live) — Campaign management, clip curation, submission review, payout triggers. Admin only. 2. ClipENGINE Marketplace (clips.conversationos.live) — Public-facing React SPA. Clippers sign up, browse campaigns, view AI-suggested clips, submit posts, receive Stripe payouts. ### AI Clip Intelligence All AI processing on ConversationOS backend: - Clip Moment Detection: startTime, endTime, duration, transcript - Clip Type Classification: Highlight, Quote, Hook, Reaction, Segment - Platform Recommendation: optimal platform per clip - Confidence Scoring: 0-100 per suggestion - Title & Description Generation - Speaker Identification: labels hosts and guests by name ### Campaign & Payout System Payout models: CPM (cents per 1K views), flat rate (per approved clip), or both. Supported platforms: TikTok, Instagram, YouTube Shorts, X, LinkedIn, Threads. Budget tracking: total/spent/remaining with visual progress. Submission lifecycle: pending → approved → paid (auto-approve after 48 hours optional). Payments: Stripe Connect Express accounts. ### Complete Pipeline Episode airs → YouTube transcript pulled → AI speaker ID → AI suggests clips (scored 0-100) → Owner approves clips → Auto-sync to marketplace → Clippers browse and submit → Auto-approve (48h) → Stripe payout ### API Endpoints Public Marketplace (11 endpoints): - GET /api/marketplace/campaigns — List active campaigns - GET /api/marketplace/campaigns/:id — Campaign detail with AI clips - GET /api/marketplace/stats — Marketplace-wide statistics - POST /api/marketplace/auth/login — Clipper login - POST /api/marketplace/auth/register — Clipper registration - GET /api/marketplace/auth/me — Current user profile - POST /api/marketplace/campaigns/:id/submit — Submit clip - GET /api/marketplace/my/submissions — My submissions - GET /api/marketplace/my/earnings — Earnings summary - POST /api/marketplace/my/stripe-connect — Initiate Stripe onboarding - GET /api/marketplace/my/stripe-connect/status — Check Stripe status Internal (20 endpoints): - Campaign CRUD: GET/POST/PATCH/DELETE campaigns - Campaign Clips: GET/POST clips, POST auto-populate, DELETE - Submissions: GET/PATCH (approve/reject), POST payout - Stats: GET clipengine-stats ### Database Tables (5) campaigns, campaignClips, clippers, submissions, clipperPayouts --- ## ENGINE 9: Character ENGINE — AI Character Universe **Domain:** universe.mitl.studio Backend for the MiTL synthetic local media network. Powers mitl.news (The Desk), universe.mitl.studio (The Universe), and city-level local news content. ### Core Capabilities 1. Character Management — 155+ AI correspondents with DNA profiles, voice presets, backstories, personality traits, and beat assignments 2. Universe/Storyworld System — Fictional universes with lore, arcs, episodes, relationships, and canon management 3. AI Content Generation — Character-voiced articles, takes, and stories via Google Gemini 4. 8 Automated Publishing Pipelines — NHL, NFL, Feed (26+ characters), Civic, Desk (Notion intake), Wire, Trending Topics, Midterms 5. Production Studio — Image (Nano Banana Pro/Gemini), Video (Veo 3.1/Grok), AI Avatars (Runway ML), Voice (ElevenLabs) 6. Content Studio — Story draft pipeline: AI generation → editorial queue → publish 7. Deep Intel — 10 investigative dashboards 8. Media Queue — Centralized queue for thumbnail and media generation requests ### Key Characters - Kaith & Jonai — Primary MiTL hosts - Sterling James & Preston Vale — Trending Topics - Sophia Grant — Center Stage host + Guest ENGINE agent - Frankie & Tommy — GEMS Evening Shorts - Maya Reyes — Culture correspondent - 149+ city-level correspondents across 82 cities ### API Scale 329 routes, 55 database tables, covering: characters, correspondents, universes, stories, content pipelines, media generation, transcripts, analytics, and administration. --- ## ENGINE 10: Smart Teleprompter — Conversation Cockpit **Route:** app.conversationos.live/#/cockpit-canvas/:episodeId Intelligent freeform live show interface replacing traditional teleprompters. ### Canvas Features - Freeform draggable blocks with segment labels, headlines, target times, timers, script preview - Now Line Hierarchy: Exponential distance-based visual weight (Math.pow(0.55, distance)) - Gravity: Live block drifts to center-left, on-deck slides below, 0.8s animated transitions - Canvas Breathing: Green tint on speaking, red on silence, 2-second CSS transition - Up Next Auto-Promote: Next block pulses amber at 80% of segment duration - Keyword Flash: Transcript keywords matched to blocks, 1.2s blue flash ### Intelligence Layer (Co-Pilot) - Live Transcript: Deepgram Nova-2 via WebSocket, 250ms audio chunks - Co-Pilot Feed: Color-coded streaming events (info, warning, critical, AI suggestion, lens detection, keyword match) - Lens Radar: Thesis/Counter/Lateral detection every ~45 seconds via Claude with confidence % - Energy Pulse Ring: Green on speaking, red on dead air ### Widgets Show Flow Bar, Wall Clock, Coverage Tracker, Target Times, Producer Notes, Live Story Drop, Audience Widget (viewer count, comments, sentiment), CTA Cards, Character Box, Trending Topics, Hall of Records, Quick Notes. --- ## The Show Layer: mornings.live + mornings.network ### mornings.live — Flagship Daily Show + Companion App The proving ground for ConversationOS. Every feature is battle-tested here first. - Episode #3015 and counting. Branded "MITLOS." - Companion app sections: Show (live episode view), Brief (daily briefing), Check-In (community), Join (membership), Vault (archive), Team (cast/crew) - ConversationOS-powered: every episode built in Build ENGINE, every title through Title ENGINE, every guest through Guest ENGINE or $297 product ### mornings.network — MiTL Network Hub - Show replays with full episode library - MiTL Shorts — short-form content from ClipENGINE campaigns - Center Stage with Sophia Grant — guest interview segment - Full episode archive --- ## The Studio Layer: mitl.studio ### mitl.studio — The Studio Company "Live Shows That Build Community." 4-layer system: 1. Production — ConversationOS all-in-one engine 2. Interaction — Companion app (mornings.live) 3. Companion — Character universe (universe.mitl.studio) 4. Storyworld — Full AI newspaper (mitl.news) Investor page available. ### production.mitl.studio — White-Label Production as a Service "Your Shows. Our Engine." - Full ConversationOS deployment for external shows - 4-week launch cycle - Custom scoping and enterprise pricing - Every engine included: blueprint builder, episode build, distribution, guest booking, clip marketplace, title intelligence, thumbnail generation, character universe --- ## The Newspaper Layer: mitl.news **Domain:** mitl.news — "The Desk" Full AI newspaper powered by Character ENGINE. ### Stats - 145 correspondents - 82 cities covered - 10 shows - 111 stories/day ### Sections - Front Page — Top stories - The Record — Official record content - The Wire — Breaking/wire news - Investigations — Long-form deep dives - Elections — Election coverage - The Numbers — Data/finance stories - Sports — NHL, NFL, and local sports - Cities — City-level local news from 82 markets ### Products - "Own a Market" — CTA to purchase a city franchise - "Morning Show in a Box" — $3,000 setup + 75/25 split product --- ## The Sales Funnels ### ownthemorning.live — Morning Show in a Box "Morning Show Box" — weekly live show product for leaders. - CTA: "Apply to Lead" - Format: weekly live show - Price: $3,000 one-time setup fee + 75/25 revenue split (host keeps 75%) - Infrastructure: full ConversationOS deployment ### hostinthelab.live — Founding Host Recruitment "Looking for 100 people." - Founding host application funnel - Selective recruitment pipeline - Access to full ConversationOS stack and network --- ## AI Agents ### Mack Hewston — Internal Showrunner 46 tools including: customer/show/episode CRUD, auto-build episodes, content pool management, knowledge base, calendar, audit logs, Roam messaging, user management, guest booking, deliverable tracking, briefing generation. Workflow: Draft → Critique → Audit (3-agent background process). Personality: Concise, precise. "That's not ready." "Clean it up." "That's noise, not signal." Access: app.conversationos.live (internal only) ### Sophia Grant — Guest-Facing Agent + Appearance Host Conversational AI at ge.conversationos.live. Dual role: 1. Center Stage booking — 10-question intake, calendar management, welcome emails 2. $297 paid appearance intake — product selection, deliverable briefing, intake questions Powered by Claude Sonnet. Maintains the named persona of the Center Stage host. --- ## Customer Portal (CX) **Domain:** cx.conversationos.live — "Your Show. Your Command Center." Pages: Dashboard, Episodes, Episode Detail, Content Pool, Blueprint, Characters, Thumbnails, Communications, Deliverables, Settings. Design: Navy (#0f172a) background, gold (#FFDE59) accents, Syncopate + General Sans typography. Authentication: Separate customer auth (bcrypt + JWT), multi-tenant. Provisioned: Auto-created on Stripe payment confirmation. --- ## MCP Server **Domain:** mcp.conversationos.live Model Context Protocol server exposing 42 tools for AI agent integration. Full ConversationOS capability available to any MCP-compatible AI agent. See mcp-tool-descriptions.json for complete tool schemas. --- ## Business Model ### Revenue Streams - MiTL Live Network subscription: $3,000/mo - ConversationOS SaaS: $200/mo - White-label production (production.mitl.studio): Enterprise/custom - Morning Show in a Box: $3,000 setup + 75/25 revenue split (host keeps 75%) - Guest appearances (ge.conversationos.live): $297 one-time per appearance - ClipENGINE marketplace: Platform fee on creator payouts ### Customer Pipeline Stages Prospect → Paid Setup → Building → Delivered → Live Subscribed ### Customer Journey ShowENGINE CX (blueprint) → Stripe payment → Show build → CX portal provisioning → Live subscription --- ## External Integrations Character ENGINE, YouTube Data API, Google Calendar, Stripe, Stripe Connect, Postmark, Pipedrive, Roam, Perplexity Sonar, Anthropic Claude, Google Gemini, ElevenLabs, Runway ML, Veo 3.1, Grok, ShowENGINE CX, mornings.live, StreamYard, Notion, Buffer, Deepgram, Cloudflare. --- ## Contact Mornings in the Lab (MiTL) Live Network Email: mornings@bapl.ai Web: https://mornings.live Platform: https://conversationos.live Studio: https://mitl.studio Newspaper: https://mitl.news Network: https://mornings.network