Page 89 of 97 — MCP servers, Claude Code skills, prompt packs, and workflows.
Claude intakes property data (beds, baths, sqft, features, neighborhood, photos) and generates MLS-compliant listing descriptions: headline, long description (no forbidden fair-housing language), feature bullets, and the neighborhood-narrative paragraph. Writes in the agent's voice via style RAG, passes MLS word-count limits, optimizes for the "search shows 2 lines" preview.
Claude reviews daily pricing decisions: reads occupancy forecast, competitor rates (OTA scraping), events calendar, demand drivers, and suggests per-room-category rate adjustments optimizing RevPAR (not just occupancy). Flags over-discount patterns, surfaces upsell opportunities on slow days. Works alongside Duetto/IDeaS or replaces the spreadsheet-based decision-making in smaller properties.
Claude workflow that writes the three-email abandoned-cart recovery sequence for every cart, personalized to the specific products abandoned, customer purchase history, and discount strategy. Writes in the brand's voice via RAG from past campaigns. Typical recovery rate: 15-25% of abandoned cart revenue — pays for itself in week one of any DTC operator.
When a critical supplier signals disruption (delay, shortage, quality issue), Claude pulls the affected parts + BOM dependencies + inventory position + alternate suppliers + lead-time implications, then generates the response plan: internal escalation brief, customer-communication draft (only if production schedule will slip), alternate-sourcing RFQ, and the cost-impact estimate. Turns a week of scrambling into a day of execution.
Writes the multi-touch renewal reminder sequence personalized per client: 90-day educational touch, 60-day review proposal, 30-day actionable deadline, 7-day urgency with specific rate + coverage recommendations. Pulls the account's loss history, current coverage gaps, and cross-sell opportunities into the messaging. Moves renewal retention from 75% → 90%+.
AI operating system blueprint for insurance brokers, agencies, and carrier ops. Architecture: agency management (Applied Epic, AMS360) + policy + claims + underwriting + compliance + RAG over policy wordings, ISO forms, endorsements, state regulations + Claude MCP + n8n. 10 MCP tool schemas: policy review summarizer, claim intake triager, underwriting-risk assessor, renewal reminder sequencer, certificate-of-insurance generator, audit support drafter, regulatory filing assistant, cross-sell opportunity finder, loss-run analyzer, endorsement drafter. Every agency-management-system workflow becomes AI-accelerated without ripping out the system of record.
Reads foremen's field reports (photos, notes, timesheets), generates the structured daily report: progress vs. schedule, manpower on site, materials delivered, safety incidents, RFIs raised, weather impact, owner-ready summary. Auto-flags schedule-slip risks. Replaces the daily-report-format-everyone-hates with a 5-minute voice memo from the foreman.
Claude coordinates sub scheduling for construction projects: reads the build schedule, identifies sub handoff dependencies, sends the 48-hour heads-up, the 24-hour confirm, and the day-of directions text. Handles reschedules cascade-style (if framing slips, drywall shifts, electric shifts, etc.). Replaces the project manager's WhatsApp chaos with a coordinated system.
Real estate has 6-18 month sales cycles. This Claude workflow builds the multi-touch nurture sequence for buyer leads: home-search updates, market commentary, first-time-buyer education series, mortgage-rate trigger alerts. Personalized by price range + geography + life-stage. Auto-pauses when lead signals ready-to-buy (increased engagement, specific-property interest). Turns cold leads into warm-at-the-right-moment.
Writes guest-facing messages across channels (SMS, email, in-stay app, review-platform DM) for every touchpoint: pre-arrival, check-in, mid-stay upsell, check-out, post-stay review request, complaint recovery. Personalizes per guest preference/history. Compliant with TCPA/GDPR. Replaces the generic "Welcome to XYZ Hotel!" template that every guest already ignores.
Claude runs menu-engineering matrix on POS data: classifies each item as Star (high-margin + high-popularity), Plowhorse (low-margin + high-popularity), Puzzle (high-margin + low-popularity), or Dog (low-margin + low-popularity). Recommends specific actions per item — promote, re-engineer cost, reposition on menu, retire. Surfaces the hidden margin-eating items most GMs never catch.
AI operating system blueprint for GCs, subs, and specialty trade businesses (HVAC, plumbing, electrical, roofing). Architecture: project management (Procore, Buildertrend, JobNimbus) + estimating (PlanSwift, Stack) + field reporting + CRM + scheduling + Claude MCP + n8n. 10 MCP tool schemas: bid/estimate generator from scope, subcontractor schedule coordinator, RFI drafter, change-order generator, safety toolbox-talk writer, daily field report synthesizer, invoice/payapp generator, punchlist tracker, customer update sender, lien waiver manager. Replaces the WhatsApp-and-Excel coordination layer that every contractor pretends is working.
AI operating system blueprint for independent hotels, restaurant groups, and travel agencies. Architecture: PMS (Opera, Cloudbeds, Mews) + POS (Toast, Square) + booking engines + review platforms + dynamic pricing + guest messaging + Claude MCP + n8n. 10 MCP tool schemas: hotel revenue management pricing reviewer, restaurant menu engineering analyzer, guest message drafter (multi-channel), review response drafter, booking-abandonment recovery, housekeeping coordinator, F&B waste tracker, staff scheduling optimizer, OTA rate parity monitor, group-block negotiator. RevPAR and check-average lift without adding headcount.
Claude-powered cohort analysis: groups customers by first-purchase month + acquisition channel + first-product, tracks revenue + frequency + LTV over time, identifies which cohorts are compounding vs. churning, and generates the narrative executive summary. Answers the "are we building a business or just selling boxes?" question most DTC brands never honestly answer.
Contractor intakes scope sheet + site conditions + material costs + labor rates, Claude generates the full bid with line-item takeoff, labor allocation by trade, material pricing with markup, contingency, timeline, and payment schedule. Cross-references against last 10 similar jobs via RAG to sanity-check the number. Cuts a 6-hour estimating job to 45 minutes.
Claude agent that reads velocity + inbound POs + seasonality + lead-time per SKU per warehouse, predicts stockout dates, and generates the reorder recommendation with suggested quantities by location. Escalates high-impact SKUs (high velocity, low days-of-cover) for same-day action. Prevents the "sold out at Peak" and "overstocked at Warehouse-B" pattern that drains DTC margin.
AI operating system blueprint for independent real estate agents, teams, and brokerages. Architecture: MLS + CRM (Follow Up Boss, KVCore, CINC) + transaction management (Dotloop, SkySlope) + marketing stack (BombBomb, ListingLeads) + review/reputation + Claude MCP + n8n. 10 MCP tool schemas: listing description generator, buyer nurture sequence builder, seller listing package creator, market comp analyzer, open-house follow-up, contract timeline keeper, transaction doc checker, lender coordinator, review request automator, referral-network nurturer. Turns a solo agent into a solo-agent-with-assistant, and a team into 2x-output team.
Reads the production schedule + equipment availability + workforce schedule + maintenance windows + current WIP, outputs the balanced daily/weekly plan with bottleneck identification + suggested load-shifts. Surfaces "we'll miss the customer promise on line 3 by Thursday" two weeks in advance, not after the fact.
Claude ingests an insurance policy (P&C, life, health, etc.), generates the plain-English client-facing summary: what's covered, key exclusions, sub-limits, deductibles, renewal terms. Flags notable changes from prior year. Compliant with state disclosure requirements. Turns a 90-page policy into a 2-page document the client actually reads.
Intakes product data (images, specs, category, brand-voice rules) and generates: short description, long description, 5 bullet benefits, SEO title, meta description, Google Shopping feed attributes. Generates per-variant copy (color/size/material) without duplication penalties. Writes in the brand's actual voice via style-guide RAG, not generic AI-ese.
Pulls OEE (Overall Equipment Effectiveness) components — Availability, Performance, Quality — from MES, diagnoses the driver of every movement week-over-week, and writes the executive-readable narrative: what moved, why, what we're doing about it. Replaces the "here's the OEE dashboard, figure it out yourself" pattern that loses plant executives at the chart.
Scores voice-agent interactions before escalating to human: intent-captured? sentiment OK? resolution complete? next-step clear? If any fails, auto-hands-off to a live agent with the full context packet. Prevents the classic "AI agent kept the angry customer in a loop for 12 minutes" failure mode.
Pulls MES / ERP production data at shift change, generates the structured handoff report: production vs. target, quality issues, downtime events, safety incidents, maintenance-in-progress, open action items. Replaces the 20-minute verbal hand-off that always loses detail; every incoming shift supervisor walks in with the full context in 2 minutes.
Claude agent that ingests every new support ticket, classifies it by product area + urgency + intent, routes to the right tier or team, suggests the KB article that answers it, and pre-drafts the first reply for human review. Converts the "pile of unsorted tickets" into a prioritized queue in seconds, not hours.
Claude reasoning layer on top of predictive-maintenance sensor data (MindSphere / ThingWorx / Azure IoT). When a model fires an anomaly alert, Claude pulls the asset's maintenance history, similar past incidents, parts inventory, and tech availability; then recommends: immediate action, scheduled maintenance window, or monitor. Converts raw alert noise into a prioritized action queue the ops team can trust.
Claude workflow that pulls new reviews across platforms (Google, Shopify, Amazon, Yelp, Trustpilot), synthesizes them into themes, quantifies sentiment trends, drafts platform-specific responses (public for positive, empathetic-with-resolution for negative), and flags the 1-2 complaints that signal a systemic product/service issue. Closes the post-purchase reputation loop.
Watches resolved tickets, detects the ones representing novel/recurring issues, and auto-drafts a knowledge-base article with the resolution path. Reviewer approves → publishes to Zendesk/Intercom Help Center → future tickets on the same issue get auto-resolved. Turns support volume into a compounding-content machine instead of a Sisyphean pile.
Watches SKU-level margin, velocity, and competitor pricing; recommends price changes with per-SKU reasoning (e.g., "raise 4% — inventory low, competitor out of stock, margin below target"). Runs weekly against full catalog, escalates high-impact changes for approval, auto-applies approved moves via the platform API.
Claude-powered SDR that qualifies inbound leads (ICP fit scoring via CRM + firmographic enrichment), triggers personalized outbound sequences for high-fit prospects, handles meeting booking via calendar MCP, and escalates ready-to-buy signals to human AEs. Ships the ICP rubric template, sequence library (cold outbound, re-engagement, post-demo follow-up), CRM sync schema (Salesforce/HubSpot/Attio), and the weekly performance dashboard — connects to the full pipeline funnel.
Full finance back-office Claude workflow: invoice OCR + GL coding suggestion + approval routing (AP), invoice generation + sending + reminder sequences + dunning (AR), employee expense receipt OCR + policy-violation flagging + approval routing. Reconciles against bank feeds, flags variances, generates month-end close checklists. Runs on top of QuickBooks/Xero/NetSuite/Ramp/Brex.
Full deployment stack for replacing tier-1 customer support with a Claude-powered agent. Ships the KB-ingest pipeline (Zendesk/Intercom/Gorgias article extraction → Qdrant), the agent persona template, escalation rules with scope boundaries (when to handoff to human), approval gates for refunds/exceptions, daily QA sampling rubric, and the analytics dashboard for containment-rate / CSAT / first-response-time. Realistic deployment window: 3 weeks to 60% containment on common issue types.
End-to-end recruiting Claude workflow: job description drafter (calibrated to role + comp band), LinkedIn/Apollo sourcing agent, resume screening against rubric, automated candidate outreach sequences, interview scheduler (calendar MCP), post-interview scorecard drafter, and the rejection-letter generator with regulatory compliance. Ships with EEOC-aware language gates and the audit log every TA team needs.
Turns your scattered Slack/Notion/Drive/Confluence into a Claude-queryable knowledge base. Ships the connectors (source → chunker → embedder → Qdrant), permission-aware retrieval (respects source-system ACLs), Slack slash-command Q&A bot, stale-content detector, and the quarterly-refresh automation. New hires onboard in days, not months; every team answer is one query away.
Claude agent that takes natural-language requests ("give me weekly new-customer LTV by acquisition channel") and returns actual dashboards: writes the SQL against your warehouse, validates the output, renders charts, writes the narrative summary, and schedules the recurring refresh. Connects to Snowflake/BigQuery/Postgres/Supabase. Replaces the "ask data team → wait 3 days" pattern with a 5-minute self-serve loop for business stakeholders.
One-brief → multi-format content pipeline: intake brief (topic, angle, audience, CTA) → Claude generates the blog post + Twitter/X thread + LinkedIn post + 3 email variants + ad copy + podcast script outline, all voice-consistent via brand-guidelines RAG. Auto-publishes to CMS/Buffer/Ghost via MCP, schedules across channels, captures performance data into the analytics loop for continuous prompt tuning.
End-to-end document pipeline: PDF/image ingestion → Claude vision OCR → structured data extraction with schema validation → routing to downstream system (CRM / ERP / billing / filing cabinet). Handles invoices, contracts, loan docs, medical forms, claim documents, tax filings. Includes the rejection queue for low-confidence extractions, human-in-the-loop approval UI, and the audit trail every regulated industry requires.
Claude-powered inbox: auto-categorizes incoming mail (sales / support / vendor / personal / spam), drafts replies for the top 80% of recurring patterns (calendar coordination, FAQ answers, intro emails), applies your written voice via few-shot examples, surfaces VIP threads with context from prior exchanges, and handles the "I'll get back to you" promises that usually fall through the cracks. Turns a 2-hour-daily inbox into a 15-minute review queue.
The cross-industry automation blueprint: 10 workflows that apply to every business with enterprise-grade ROI — customer support agents, AI SDR, document intelligence, meeting assistant + CRM auto-update, content factory, internal RAG knowledge base, recruiting automator, finance ops (AP/AR/expense), email triage, data analysis agent. Ships the 10-workflow reference architecture (Claude MCP + n8n + Qdrant + existing SaaS integrations), 60+ MCP tool schemas, 15+ n8n templates, deployment sequencing (quickest-ROI-first rollout order), and the org-change-management playbook. Use as-is or combine with a vertical OS flagship (WealthOps/MedOps/ShopOps/etc.) for vertical-specific depth.
Post-meeting Claude workflow: pulls Zoom/Meet/Teams transcripts, extracts action items + commitments + next steps, auto-drafts follow-up email, updates the CRM opportunity record (stage, next activity, notes), creates calendar holds for committed follow-ups, and surfaces any risk signals for sales ops review. Every meeting becomes structured pipeline data, not buried-in-transcript vapor.
Cross-industry AI operating system blueprint for customer-service ops. Architecture: helpdesk (Zendesk/Intercom/Gorgias) + voice platform (Twilio/Dialpad) + knowledge base in Qdrant + Claude MCP + n8n triggers. 9 MCP tool schemas: inbound ticket triager, first-reply drafter, voice-agent handoff scorer, KB article auto-generator, refund/exception approver with rules gate, SLA breach escalator, customer-health scorer, churn-risk flagger, QA call reviewer. Biggest ROI lever in the entire AI stack — every hour of tier-1 CSR time automated at 90% quality.
AI operating system blueprint for DTC brands, Shopify stores, and multi-channel retailers. Architecture: Shopify/WooCommerce/Amazon + Klaviyo + Gorgias/Zendesk + inventory management + demand forecasting model + dynamic pricing engine + Claude MCP + n8n. 10 MCP tool schemas: customer support automator, dynamic pricing reviewer, inventory reorder predictor, product description generator, review sentiment synthesizer, abandoned-cart recovery writer, returns dispositioner, supplier PO assistant, Meta/Google ad copy generator, cohort analyzer. Turns a 1-person-brand into a team-of-five-equivalent.
Claude reviews contracts against firm playbook (or attorney-defined rules): flags deviations by severity, suggests specific redlines with redline-explanation comments, identifies missing clauses, surfaces governing-law mismatches, and outputs a negotiation memo ranking issues. 4-hour senior-associate review compressed to 30 minutes of partner review.
AI operating system blueprint for energy producers, utilities, and grid operators. Architecture: SCADA/OT data integration + demand forecasting model + anomaly detection + outage coordination + regulatory filing assistant + Claude MCP + n8n. 8 MCP tool schemas: demand forecaster explainer, outage coordinator, maintenance window scheduler, regulatory filing drafter (FERC/NERC), equipment monitoring summarizer, DR program communicator, renewable-curtailment explainer, customer-outage messenger. Operational decision intelligence on top of hardened SCADA, not replacing it.
Claude workflow for management consultants: intakes project context + interview notes + data findings, outputs client-ready deliverable stack — executive summary memo, detailed analysis slides with McKinsey-style pyramid structure, appendix workbooks, and a speaking notes doc for the partner. Scales a 3-person team's output 3x without hiring.
Ingests the patient chart + requested procedure/medication + payer rules, generates the prior auth request packet with medical necessity letter, CPT/ICD justification, supporting chart excerpts, and the payer-specific form. Handles the #1 time-sink in every practice — PA requests. Turns a 45-minute staff task into 5 minutes of review + fax.
AI operating system blueprint for discrete and process manufacturers. Architecture: MES / ERP integration + predictive-maintenance model layer + computer-vision QC inspection + supply-chain planner + Claude MCP for the reasoning layer + n8n for triggers. 8 MCP tool schemas: predictive-maintenance alert triager, QC anomaly explainer, supply-disruption responder, shift handoff reporter, OEE narrative generator, supplier scorecard updater, EHS incident drafter, capacity planner. Physical operations stay physical; the intelligence layer is AI-native.
HIPAA-aware AI operating system blueprint for clinics, medical groups, and life-sciences teams. Architecture: EHR bridge (Epic / athenahealth / eClinicalWorks via FHIR) + patient intake automation + AI scribe for transcription + medical-coding assistant (ICD-10, CPT, HCC) + RAG over clinical guidelines + Claude MCP with PHI-redaction boundary + n8n. Ships 10 MCP tool schemas (patient intake, AI scribe, prior auth, claims billing, denial appeals, refill handler, referral router, outcomes tracker, patient education, claim-denial trend analyzer), HIPAA data-handling design, and deployment guide with BAAs considered.
Ingests the denial letter + patient chart + policy language, identifies the denial reason code, and drafts the appeal letter with medical-necessity argument + specific policy citations + required documentation checklist. Targets the $262B/year that payers withhold via denials — most practices capture <40% of recoverable denials due to manual-appeal friction.