Keep-customer swarm · Review Response Agent · Review-response-drafting skill · Build pillar · Published August 20, 2026

How to build per-location AI review-response drafting at multi-location scale

This guide explains how to architect the review-response-drafting skill on the review-response agent end-to-end at multi-location AI-review-response-generator scale: per-portfolio per-banner per-location per-canonical-review-source-pointer + per-canonical-class-differentiated-drafting-engine-spec + per-canonical-multi-LLM-ensemble-spec + per-canonical-rep-approval-spec + per-canonical-auto-publish-threshold-spec + per-canonical-multi-arm-bandit-spec + per-canonical-compliance-gate-spec + per-canonical-audit-trail + per-portfolio-audit-trail. This skill is the artifact-production counterpart to the cs-agent-assist co-pilot (sibling build-pillar on the same review-response agent) — drafting ships publishable artifacts; the co-pilot ships in-context suggestions.

Engage the review-response agent See the 32-agent swarm

What you will build

Per-portfolio per-banner per-location per-canonical-review-source-pointer across 40+ review sources (same surface set as the cs-agent-assist build-pillar) — GBP + Yelp + Facebook Page reviews + Apple Maps Ratings + Bing Places + TripAdvisor + OpenTable + Resy + Tock + Healthgrades + Zocdoc + RateMDs + Vitals + WebMD + Avvo + Lawyers.com + Martindale + Cars.com + DealerRater + Edmunds + KBB + AutoTrader + BBB + Trustpilot + SiteJabber + Glassdoor + Indeed employer + Angi + HomeAdvisor + Thumbtack + Houzz + Porch + Nextdoor + Google Play Store + Apple App Store + Amazon product reviews + Yotpo + Loox + Stamped + Okendo + Judge.me + Reviews.io Shopify integrations.
Per-canonical-class-differentiated-drafting-engine-spec — per-5-star-praise (low-friction acknowledgment + thank-you + detail-specific callback + invitation-to-return + cross-sell opportunity + loyalty-program mention when applicable + no promotional overreach) + per-4-star-praise-with-suggestion (acknowledgment + thank-you + brief response to suggestion + invitation-to-return + no defensive explanation) + per-3-star-neutral (acknowledgment + brief clarification + offline resolution CTA + no over-explanation) + per-2-star-concern (deeper empathy + accountability + offline resolution CTA with specific channel and rep name + no defensive language + no blame of customer + no public detail disclosure) + per-1-star-crisis (immediate empathy + accountability without admission of legal liability + executive escalation path + offline resolution CTA with direct line + no PHI + no PII + no account detail + crisis escalation via PagerDuty/Opsgenie/VictorOps/xMatters/Better Stack + PR firm handoff + executive team paging) + per-multi-star-nuanced-mixed-sentiment (per-aspect-by-aspect response + positive acknowledgment + concern acknowledgment + offline resolution for concern + thank-you for balanced feedback) + per-fake-review-detection (skip draft + FTC fake-review rule of 2024 flag + platform removal request + fraud investigation handoff) + brand-voice spec + claims allowlist + forbidden phrase library + trademark-aware substitution + 14-language drafting (Spanish + Mandarin + Vietnamese + Korean + Tagalog + French + Arabic + Russian + Haitian Creole + Polish + Portuguese + Hindi + Urdu + Bengali) + per-draft confidence tier.
Per-canonical-multi-LLM-ensemble-spec — per-10-LLM-judge-pool (GPT-4o + Claude Sonnet + Claude Opus + Gemini Pro + Mistral Large + Cohere Command + Llama 3 70B + Meta LlamaGuard safety filter + ShieldGemma safety filter + NVIDIA NeMo Guardrails safety filter) + per-7-eval-pattern (chain-of-thought + tree-of-thought + self-consistency + Constitutional AI + aspect-based empathy/accountability/resolution/tone/cadence/CTA + G-Eval + LLM-as-jury with Borda-count + Condorcet aggregation) + per-4-inter-LLM-agreement (Cohen kappa + Krippendorff alpha + Fleiss kappa + Gwet AC1) + per-6-bias-mitigation (position + verbosity + self-enhancement + length + style + cultural) + per-prompt-injection-defense (Lakera Guard + Protect AI Rebuff + Lasso Security + Robust Intelligence + Calypso AI + Aporia) + per-jailbreak-detection + per-FBC-feedback-loop + per-ensemble-confidence-tier.
Per-canonical-auto-publish-threshold-spec + per-canonical-rep-approval-spec + per-canonical-multi-arm-bandit-spec + per-canonical-compliance-gate-spec — per-5-star-low-threshold (~90% auto-publish when all aspects pass + confidence high + no claims flag + no PHI + no legal threat + no fake-review signal) + per-4-star-medium-threshold (~50% auto-publish) + per-3-2-1-star-high-threshold-requires-human-review + per-vertical-threshold-modifier (healthcare lower per HIPAA + legal lower per state bar + financial lower per FINRA + cannabis lower per state board + alcohol lower per TABC) + per-state-threshold-modifier + per-6-tier-rep-approval (auto-publish + rep-review + manager-review + compliance-review + legal-review + executive-review-crisis) + per-five-destination-routing-handoff + per-per-platform-policy-check + per-multi-arm-bandit-UCB + per-Thompson-Sampling + per-Epsilon-Greedy + per-LinUCB + per-Contextual-bandit + per-causal-uplift-CATE-T-S-X-DR-CausalML-DoubleML-EconML + per-FTC-fake-review-rule-of-2024 + per-FTC-substantiation + per-FTC-MARS + per-FTC-AI-disclosure + per-FTC-endorsement-guides + per-FTC-Made-in-USA + per-FTC-Green-Guides + per-FTC-Negative-Option-ROSCA + per-FTC-MAP + per-FTC-Health-Products-Compliance-Guide + per-CFPB-UDAAP + per-FDD-Item-12-Item-17-Item-19 + per-state-FRR + per-state-UDTPA + per-Lanham + per-Robinson-Patman + per-HIPAA-no-acknowledge-as-patient (the most operationally distinctive constraint for healthcare reviews under 45 CFR 164.502) + per-HIPAA-marketing-authorization-164.508 + per-FCRA + per-ECOA-Reg-B + per-Fair-Housing-Act + per-GLBA + per-FINRA-2210 (financial-services-review = FINRA-regulated-communication requiring supervisory review pre-use) + per-SEC-Reg-S-K + per-state-bar-advertising (some states prohibit responding to negative reviews) + per-state-professional-licensing + per-FDA-DSHEA + per-FDA-OPDP-Rx-drug + per-DEA-Schedule-II-V + per-cannabis-state-board + per-alcohol-TABC-CalABC-SLA + per-Surgeon-General-warning + per-tobacco-FDA + per-TCPA + per-CASL + per-CCPA-CPRA + per-GDPR-Article-6-7-17-22 + per-LGPD + per-DPDP + per-PIPEDA + per-COPPA + per-EU-AI-Act-Article-50-transparency-the-AI-must-disclose-itself + per-EU-AI-Act-Article-13-14-15 + per-EU-AI-Act-Article-22-automated-decision + per-Digital-Services-Act-Article-30 + per-Digital-Markets-Act + per-WCAG-2.2-AA + per-ARIA + per-EAA-EN-301-549 + per-Section-508 + per-ADA-Title-III + per-NIST-AI-RMF + per-ISO-42001 + per-ISO-27001 + per-SOC-2-Type-II + per-OPA-Cedar-Casbin-Cerbos-Oso-policy-as-code + per-compliance-confidence-tier.
Per-canonical-cross-skill-handoff + per-canonical-audit-trail — per-handoff-to-30-sibling-skills + per-per-draft-canonical-audit-record + per-immutable-WORM-storage + per-7-year-IRS-tax-retention + per-7-year-FTC-substantiation-retention + per-7-year-HIPAA-medical-record-retention + per-7-year-state-bar-record-retention + per-6-year-SEC-record-retention + per-3-year-FINRA-record-retention.

Why per-vendor-Birdeye-AI-account-flat-review-reply-template breaks at multi-location AI-review-response-generator scale

At 1-account-1-flat-review-reply-template scale per-account per-flat-review-reply-template primitive is enough. At multi-location AI-review-response-generator scale per-canonical-review-source-pointer + per-canonical-class-differentiated-drafting-engine-spec + per-canonical-multi-LLM-ensemble-spec + per-canonical-rep-approval-spec + per-canonical-auto-publish-threshold-spec + per-canonical-multi-arm-bandit-spec + per-canonical-compliance-gate-spec + per-canonical-audit-trail.

HIPAA-no-acknowledge-as-patient under 45 CFR 164.502 is the most operationally distinctive constraint here. Under the HIPAA Privacy Rule, a covered entity cannot publicly acknowledge anyone as a patient — even in response to a glowing review where the reviewer self-discloses (a five-star review reading "Dr. Smith treated my anxiety for the past year and changed my life" cannot be answered with "Thank you for trusting us with your anxiety treatment, Sarah!" — that response acknowledges Sarah as a patient and is a HIPAA violation). Safe-pattern responses generically reference "the team is committed to" without acknowledging the reviewer as a patient. Per-vendor account-flat-review-reply-template primitives have no HIPAA-no-acknowledge awareness — they generate context-matching warm responses that propagate HIPAA violations at scale.

State bar advertising rules are the second operationally distinctive constraint. Some states prohibit responding to negative reviews at all (or impose strict content rules specific to legal-services advertising). EU AI Act Article 50 transparency is the third: any auto-published response becomes an Article-50 transparency-required system whose disclosure trace must be retained. FTC AI-disclosure under the 2024 fake-review rule completes the regulatory floor.

The operator-side architecture above per-vendor-flat-review-reply-template primitive is canonical-review-source-pointer + per-class-differentiated-drafting-engine-spec + per-multi-LLM-ensemble-spec + per-rep-approval-spec + per-auto-publish-threshold-spec + per-multi-arm-bandit-spec + per-compliance-gate-spec + per-cross-skill-handoff + per-audit-trail + per-portfolio-audit-trail.

What is in market today

Per-platform per-AI-review-response-vendor

Birdeye AI, Glowing, Trustpilot AI, Reviews.io AI, Yotpo Reviews AI, Yext Reviews, SOCi Reviews, Reputation.com Reply, ReviewTrackers Reply, GatherUp Reply, Podium Reply, Swell Reply, Grade.us Reply, Chatmeter Reply, NiceJob Reply, Listen360 Reply, Synup Reply, BrightLocal Reply, Womply Reply, Demandforce Reply, Rio SEO Reply, Brandify Reply, Reputation Studio Reply. Per-account per-flat-review-reply-template primitive. Per-canonical-review-source-pointer-canonical-class-differentiated-drafting-engine-canonical-multi-LLM-ensemble-canonical-rep-approval-canonical-auto-publish-threshold-canonical-multi-arm-bandit-canonical-compliance-gate-canonical-audit-trail is not the primitive.

Per-platform per-LLM-foundation-vendor

OpenAI GPT-4o, Anthropic Claude Sonnet + Claude Opus, Google Gemini Pro + Gemini Flash, Mistral Large + Mistral Nemo, Cohere Command, Meta Llama 3 70B + Llama 3 405B, Meta LlamaGuard, Google ShieldGemma, NVIDIA NeMo Guardrails, Azure AI Content Safety, AWS Bedrock Guardrails, Google Vertex AI Guardrails Aegis. Per-account per-flat-completion primitive (typically blind to per-class differentiated template + per-LLM ensemble with Borda/Condorcet + 4 inter-LLM agreement metrics + 6 bias mitigation + prompt injection defense semantics). Per-canonical-per-draft-class-differentiated-template-canonical-per-draft-LLM-pool-canonical-per-draft-eval-pattern-canonical-per-draft-inter-LLM-agreement-canonical-per-draft-bias-mitigation-canonical-per-draft-prompt-injection-defense-canonical-per-draft-jailbreak-detection-canonical-per-draft-FBC-feedback-loop-canonical-per-draft-ensemble-confidence-tier is not the primitive.

Per-platform per-eval-platform-vendor

Braintrust, LangSmith, Weights & Biases Weave, Humanloop, Promptfoo, Helicone, Galileo, Arize Phoenix, Patronus AI, Confident AI DeepEval, MLflow LLM, Comet ML LLM, Logfire, OpenLLMetry, Langfuse, TruEra, WhyLabs, Fiddler AI, Robust Intelligence, Latitude, Vellum, PromptLayer. Per-account per-flat-eval-run primitive (typically blind to per-draft multi-arm bandit + causal uplift CATE + auto-publish threshold + 6-tier rep approval semantics). Per-canonical-per-draft-auto-publish-threshold-canonical-per-draft-per-vertical-threshold-modifier-canonical-per-draft-per-state-threshold-modifier-canonical-per-draft-rep-approval-tier-canonical-per-draft-five-destination-routing-canonical-per-draft-per-platform-policy-check-canonical-per-draft-multi-arm-bandit-canonical-per-draft-causal-uplift-CATE is not the primitive.

Per-platform per-HIPAA-compliance-vendor + per-state-bar-compliance-vendor + per-CMP-vendor

HIPAA Vault, Atlantic Health Solutions, Compliancy Group, HIPAA Secure Now, NetSec, Foley & Lardner Lawyer Compliance, ABA Center for Professional Responsibility, LegalServ, LawPay, Clio Compliance, OneTrust, TrustArc, Ketch, Securiti, Privacera, Skyflow, BigID, DataGrail, Transcend, Osano, Cookiebot, Didomi, Hyperproof, Drata, Vanta, Thoropass, Tugboat Logic. Per-account per-flat-HIPAA-compliance-report or per-flat-state-bar-rule or per-flat-consent primitive (typically blind to per-draft HIPAA-no-acknowledge-as-patient under 45 CFR 164.502 + state bar advertising rules vary per state including some that prohibit responding to negative reviews + state professional licensing + FINRA 2210 financial-services-review supervisory review pre-use + FDA OPDP + DEA + cannabis state + EU AI Act Article 50/22 + Digital Services Act Article 30 + ECOA disparate impact semantics). Per-canonical-per-draft-HIPAA-no-acknowledge-as-patient-canonical-per-draft-HIPAA-marketing-authorization-canonical-per-draft-state-bar-advertising-canonical-per-draft-state-professional-licensing-canonical-per-draft-FINRA-2210-canonical-per-draft-SEC-Reg-S-K-canonical-per-draft-FDA-OPDP-canonical-per-draft-DEA-canonical-per-draft-cannabis-state-board-canonical-per-draft-alcohol-TABC-CalABC-SLA-canonical-per-draft-FTC-fake-review-rule-of-2024-canonical-per-draft-FTC-AI-disclosure-canonical-per-draft-EU-AI-Act-Article-50-canonical-per-draft-EU-AI-Act-Article-22-canonical-per-draft-Digital-Services-Act-Article-30 is not the primitive.

How the architecture is built

Per-portfolio per-banner per-location per-canonical-review-source-pointer-substrate. Per-40-canonical-review-source canonical-source.
Per-portfolio per-canonical-class-differentiated-drafting-engine-spec. Per-7-class-template (5-star + 4-star + 3-star + 2-star + 1-star + multi-star-nuanced + fake-review-detection) + per-brand-voice-spec + per-claims-allowlist + per-forbidden-phrase-library + per-trademark-aware-substitution + per-14-language-drafting + per-draft-confidence-tier canonical-drafting.
Per-portfolio per-canonical-multi-LLM-ensemble-spec. Per-10-LLM-judge-pool + per-7-eval-pattern + per-4-inter-LLM-agreement + per-6-bias-mitigation + per-prompt-injection-defense + per-jailbreak-detection + per-FBC-feedback-loop + per-ensemble-confidence-tier canonical-multi-LLM.
Per-portfolio per-canonical-auto-publish-threshold-spec + per-canonical-rep-approval-spec + per-canonical-multi-arm-bandit-spec. Per-5-star-low-threshold + per-4-star-medium-threshold + per-3-2-1-star-high-threshold + per-vertical-modifier + per-state-modifier + per-6-tier-rep-approval + per-five-destination-routing + per-per-platform-policy-check + per-UCB + per-Thompson-Sampling + per-Epsilon-Greedy + per-LinUCB + per-Contextual-bandit + per-causal-uplift-CATE canonical-threshold-approval-bandit.
Per-portfolio per-canonical-compliance-gate-spec. Per-FTC-fake-review-rule-of-2024 + per-FTC-AI-disclosure + per-FTC-substantiation + per-FTC-MARS + per-FTC-endorsement-guides + per-FTC-Made-in-USA + per-FTC-Green-Guides + per-FTC-Negative-Option-ROSCA + per-FTC-MAP + per-FTC-Health-Products-Compliance-Guide + per-CFPB-UDAAP + per-FDD-Item-12-Item-17-Item-19 + per-state-FRR + per-state-UDTPA + per-Lanham + per-Robinson-Patman + per-HIPAA-no-acknowledge-as-patient + per-HIPAA-marketing-authorization-164.508 + per-FCRA + per-ECOA-Reg-B + per-Fair-Housing-Act + per-GLBA + per-FINRA-2210 + per-SEC-Reg-S-K + per-state-bar-advertising + per-state-professional-licensing + per-FDA-DSHEA + per-FDA-OPDP-Rx-drug + per-DEA-Schedule-II-V + per-cannabis-state-board + per-alcohol-TABC-CalABC-SLA + per-Surgeon-General + per-tobacco-FDA + per-TCPA + per-CASL + per-CCPA-CPRA + per-GDPR-Article-6-7-17-22 + per-LGPD + per-DPDP + per-PIPEDA + per-COPPA + per-EU-AI-Act-Article-50 + per-EU-AI-Act-Article-13-14-15 + per-EU-AI-Act-Article-22 + per-Digital-Services-Act-Article-30 + per-Digital-Markets-Act + per-WCAG-2.2-AA + per-ARIA + per-EAA-EN-301-549 + per-Section-508 + per-ADA-Title-III + per-NIST-AI-RMF + per-ISO-42001 + per-ISO-27001 + per-SOC-2-Type-II + per-OPA-Cedar-Casbin-Cerbos-Oso-policy-as-code canonical-compliance.
Per-portfolio per-canonical-cross-skill-handoff. Per-handoff-to-30-sibling-skills canonical-handoff.
Per-portfolio per-canonical-audit-trail + per-portfolio-audit-trail. Per-per-draft-canonical-audit-record + per-immutable-WORM-storage + per-7-year-IRS-tax-retention + per-7-year-FTC-substantiation-retention + per-7-year-HIPAA-medical-record-retention + per-7-year-state-bar-record-retention + per-6-year-SEC-record-retention + per-3-year-FINRA-record-retention canonical-audit.
Per-portfolio per-review-response-agent-canonical-bundle. Per-review-response-drafting + per-cs-agent-assist + per-auto-publish-threshold-gating + per-post-crisis-seo-repair + per-review-classification + per-crisis-escalation + per-review-source-subscription canonical-bundle.
Per-portfolio per-canonical-end-to-end-SLA. Per-review-source-resolve-to-class-differentiated-template-to-multi-LLM-ensemble-to-inter-LLM-agreement-to-bias-correction-to-prompt-injection-defense-to-auto-publish-threshold-to-rep-approval-tier-to-compliance-gate-to-EU-AI-Act-Article-50-disclosure-to-multi-arm-bandit-selection-to-publish-to-FBC-feedback-loop-SLA canonical-end-to-end-SLA.
Per-portfolio per-canonical-end-to-end-replay. Per-class-differentiated-template-rewind + per-multi-LLM-ensemble-rewind + per-auto-publish-threshold-rewind + per-rep-approval-rewind + per-multi-arm-bandit-rewind + per-compliance-gate-rewind + per-replay-confidence-tier + per-replay-explainability canonical-replay.

Frequently asked questions

What is per-location AI review-response drafting at multi-location scale — and how is it different from a CS-agent-assist co-pilot?

Per-location AI review-response drafting runs per-portfolio per-banner per-location per-canonical-review-source-pointer (per-Google-Business-Profile + per-Yelp + per-Facebook-Page-reviews + per-Apple-Maps-Ratings-and-Reviews + per-Bing-Places-reviews + per-TripAdvisor + per-OpenTable + per-Resy + per-Tock + per-Healthgrades + per-Zocdoc + per-RateMDs + per-Vitals + per-WebMD + per-Avvo + per-Lawyers.com + per-Martindale + per-Cars.com + per-DealerRater + per-Edmunds + per-KBB + per-AutoTrader + per-BBB + per-Trustpilot + per-SiteJabber + per-Glassdoor + per-Indeed-employer-reviews + per-Angi + per-HomeAdvisor + per-Thumbtack + per-Houzz + per-Porch + per-Nextdoor + per-Google-Play-Store + per-Apple-App-Store + per-Amazon-product-reviews + per-Shopify-reviews-via-Yotpo + per-Loox + per-Stamped + per-Okendo + per-Judge.me + per-Reviews.io + per-canonical-review-source) + per-canonical-class-differentiated-drafting-engine-spec + per-canonical-multi-LLM-ensemble-spec + per-canonical-rep-approval-spec + per-canonical-auto-publish-threshold-spec + per-canonical-multi-arm-bandit-spec + per-canonical-compliance-gate-spec + per-canonical-audit-trail + per-portfolio-audit-trail. The distinction from a CS-agent-assist co-pilot (sibling build-pillar at /how-to-build-multi-location-review-response-agent-assist on the same review-response agent) is artifact-vs-interaction: review-response-drafting produces a publishable draft artifact gated through auto-publish-threshold + rep-approval workflow; CS-agent-assist is the real-time co-pilot surface inside Zendesk/Salesforce Service Cloud/Intercom/Front/Gladly/Help Scout/Kustomer/Freshdesk/HubSpot Service Hub/ServiceNow/Zoho Desk/Sprinklr/Khoros that suggests responses to a live CS rep during active chats/calls/tickets. The drafting skill ships artifacts; the co-pilot skill ships in-context suggestions.

Why does per-vendor-Birdeye-AI-canonical-account-flat-review-reply-template break at multi-location AI-review-response-generator scale?

Per-vendor-Birdeye-AI-canonical-account-flat-review-reply-template ships per-account per-flat-review-reply-template primitive — typically a content manager configures a single rubric (one tone + one set of stock phrases + one auto-publish threshold for 5-star), Birdeye AI generates a 50-word response per inbound review using GPT-4 default, and the response posts to GBP. No per-canonical-review-source taxonomy across the 40+ surfaces, no per-class-differentiated drafting engine resolving per-5-star-praise (low-friction acknowledgment + thank-you + invitation-to-return) vs per-4-star-praise-with-suggestion (acknowledgment + thank-you + brief response to the suggestion + invitation-to-return) vs per-3-star-neutral (acknowledgment + brief detail-clarification + offline-resolution-CTA) vs per-2-star-concern (deeper-empathy + accountability + offline-resolution-CTA-with-specific-channel-and-rep-name + no-defensive-language) vs per-1-star-crisis (immediate-empathy + accountability-without-admission-of-legal-liability + executive-escalation-path + offline-resolution-CTA-with-direct-line + no-PHI-no-PII-no-account-detail + crisis-escalation-via-PagerDuty/Opsgenie) vs per-multi-star-nuanced-mixed-sentiment (per-aspect response addressing both positive and negative elements separately), no per-canonical-multi-LLM-ensemble resolving GPT-4o + Claude Sonnet + Claude Opus + Gemini Pro + Mistral Large + Cohere Command + Llama 3 70B with per-aspect agreement (tone + empathy + accountability + resolution + CTA), no per-canonical-rep-approval resolving auto-publish-tier (when all aspects pass + confidence high + no claims flag + no PHI + no legal threat) + rep-review-tier + manager-review-tier + compliance-review-tier + legal-review-tier + executive-review-tier-crisis, no per-canonical-auto-publish-threshold resolving per-5-star-low-threshold (auto-publish in 90% of cases) vs per-4-star-medium-threshold vs per-3-2-1-star-high-threshold-requires-human-review, no per-canonical-multi-arm-bandit resolving UCB / Thompson / Epsilon-Greedy / LinUCB / Contextual + causal-uplift-CATE T/S/X/DR-learner / CausalML / DoubleML / EconML for response-variant selection, no per-draft compliance gate with FTC fake-review-rule-of-2024 (AggregateRating must tie to real reviews) / HIPAA-no-acknowledge-as-patient (the most operationally distinctive constraint for healthcare reviews) / state bar advertising rules / state professional licensing / FINRA 2210 financial-services / FDA OPDP Rx drug / DEA Schedule II-V / cannabis state board / alcohol TABC + Surgeon General / EU AI Act Article 50 transparency (the AI must disclose itself) / Digital Services Act Article 30 / WCAG / ADA enforcement, no per-draft audit trail with regulatory-defense retention. Per-vendor-Glowing + Trustpilot-AI + Reviews.io-AI + Yotpo-Reviews-AI + Yext + SOCi + Reputation.com + ReviewTrackers + GatherUp + Podium + Swell + Grade.us + Chatmeter + NiceJob + Listen360 + Synup + BrightLocal + Womply + Demandforce + Rio-SEO + Brandify + Reputation-Studio-canonical-account-flat-review-reply-template ship per-vendor per-native account-flat-review-reply-template primitives. At 1-account-1-flat-review-reply-template scale per-account per-flat-review-reply-template primitive is enough. At multi-location AI-review-response-generator scale per-canonical-review-source-pointer + per-canonical-class-differentiated-drafting-engine-spec + per-canonical-multi-LLM-ensemble-spec + per-canonical-rep-approval-spec + per-canonical-auto-publish-threshold-spec + per-canonical-multi-arm-bandit-spec + per-canonical-compliance-gate-spec + per-canonical-audit-trail.

How does per-review class-differentiated drafting-engine + per-review multi-LLM-ensemble work?

Per-portfolio per-banner per-location per-review per-canonical-class-differentiated-drafting-engine-spec runs per-portfolio per-canonical-per-5-star-praise-template (per-low-friction-acknowledgment + per-thank-you + per-detail-specific-callback + per-invitation-to-return + per-cross-sell-opportunity + per-loyalty-program-mention-when-applicable + per-no-promotional-overreach) + per-canonical-per-4-star-praise-with-suggestion-template (per-acknowledgment + per-thank-you + per-brief-response-to-suggestion + per-invitation-to-return + per-no-defensive-explanation) + per-canonical-per-3-star-neutral-template (per-acknowledgment + per-brief-detail-clarification + per-offline-resolution-CTA + per-no-over-explanation) + per-canonical-per-2-star-concern-template (per-deeper-empathy + per-accountability + per-offline-resolution-CTA-with-specific-channel-and-rep-name + per-no-defensive-language + per-no-blame-of-customer + per-no-public-detail-disclosure) + per-canonical-per-1-star-crisis-template (per-immediate-empathy + per-accountability-without-admission-of-legal-liability + per-executive-escalation-path + per-offline-resolution-CTA-with-direct-line + per-no-PHI + per-no-PII + per-no-account-detail + per-crisis-escalation-via-PagerDuty-Opsgenie-VictorOps-xMatters-Better-Stack + per-PR-firm-handoff + per-executive-team-paging) + per-canonical-per-multi-star-nuanced-mixed-sentiment-template (per-aspect-by-aspect-response + per-positive-acknowledgment + per-concern-acknowledgment + per-offline-resolution-for-concern + per-thank-you-for-balanced-feedback) + per-canonical-per-fake-review-detection-template (per-skip-draft + per-FTC-fake-review-rule-of-2024-flag + per-platform-removal-request + per-fraud-investigation-handoff) + per-canonical-per-brand-voice-spec-adherence (per-tone-attribute + per-cadence-attribute + per-vocabulary-attribute + per-banned-construction-pattern + per-required-construction-pattern + per-per-vertical-tone-variation) + per-canonical-per-claims-allowlist + per-canonical-per-forbidden-phrase-library + per-canonical-per-trademark-aware-substitution (Super Bowl → the big game + Olympics → summer games + NFL/MLB/NBA/NHL/NCAA team trademark respect + March Madness/Final Four/World Cup/Stanley Cup substitution + trademark fair use + nominative fair use + tarnishment + dilution avoidance) + per-canonical-per-14-language-drafting (Spanish + Mandarin + Vietnamese + Korean + Tagalog + French + Arabic + Russian + Haitian Creole + Polish + Portuguese + Hindi + Urdu + Bengali) + per-canonical-per-draft-confidence-tier + per-canonical-per-draft-explainability. Per-canonical-multi-LLM-ensemble-spec runs per-portfolio per-canonical-per-draft-LLM-pool (per-GPT-4o + per-Claude-Sonnet + per-Claude-Opus + per-Gemini-Pro + per-Mistral-Large + per-Cohere-Command + per-Llama-3-70B + per-Meta-LlamaGuard-safety-filter + per-ShieldGemma-safety-filter + per-NVIDIA-NeMo-Guardrails-safety-filter) + per-canonical-per-draft-eval-pattern (per-chain-of-thought + per-tree-of-thought + per-self-consistency + per-Constitutional-AI + per-aspect-based-empathy-accountability-resolution-tone-cadence-CTA + per-G-Eval + per-LLM-as-jury-with-Borda-count-Condorcet-aggregation) + per-canonical-per-draft-inter-LLM-agreement (per-Cohen-kappa + per-Krippendorff-alpha + per-Fleiss-kappa + per-Gwet-AC1) + per-canonical-per-draft-bias-mitigation (per-position-bias + per-verbosity-bias + per-self-enhancement-bias + per-length-bias + per-style-bias + per-cultural-bias) + per-canonical-per-draft-prompt-injection-defense (per-Lakera-Guard + per-Protect-AI-Rebuff + per-Lasso-Security + per-Robust-Intelligence + per-Calypso-AI + per-Aporia) + per-canonical-per-draft-jailbreak-detection + per-canonical-per-draft-FBC-feedback-loop (per-realized-vs-predicted-acceptance + per-realized-vs-predicted-helpfulness + per-realized-vs-predicted-de-escalation + per-realized-vs-predicted-resolution-rate + per-pattern-learning + per-multi-arm-bandit-regret + per-recalibration) + per-canonical-per-draft-ensemble-confidence-tier + per-canonical-per-draft-ensemble-explainability.

What does per-draft auto-publish-threshold + per-draft rep-approval + per-draft multi-arm-bandit + per-draft compliance-gate do?

Per-portfolio per-banner per-location per-draft per-canonical-auto-publish-threshold-spec runs per-portfolio per-canonical-per-draft-per-5-star-low-threshold (auto-publish in ~90% of cases when all aspects pass + confidence high + no claims flag + no PHI + no legal threat + no fake-review signal) + per-canonical-per-draft-per-4-star-medium-threshold (auto-publish in ~50% of cases) + per-canonical-per-draft-per-3-2-1-star-high-threshold-requires-human-review + per-canonical-per-draft-per-vertical-threshold-modifier (healthcare requires lower auto-publish-rate per HIPAA + legal requires lower auto-publish-rate per state bar + financial requires lower per FINRA + cannabis requires lower per state board + alcohol requires lower per TABC) + per-canonical-per-draft-per-state-threshold-modifier (state bar advertising rules + state professional licensing + state UDAP). Per-canonical-rep-approval-spec runs per-portfolio per-canonical-per-draft-auto-publish-tier + per-canonical-per-draft-rep-review-tier + per-canonical-per-draft-manager-review-tier + per-canonical-per-draft-compliance-review-tier + per-canonical-per-draft-legal-review-tier + per-canonical-per-draft-executive-review-tier-crisis + per-canonical-per-draft-five-destination-routing-handoff (sibling skill on governance-decision-router agent) + per-canonical-per-draft-per-platform-policy-check (per-Google-reply-policy + per-Yelp-content-guidelines + per-Meta-Community-Standards + per-Apple-Business-Connect + per-TripAdvisor + per-Healthgrades + per-Zocdoc + per-Avvo + per-BBB + per-Trustpilot + per-Glassdoor + per-Nextdoor + per-App-Store-developer-response-guidelines + per-Google-Play-developer-policy + per-Amazon-Community-Guidelines). Per-canonical-multi-arm-bandit-spec runs per-portfolio per-canonical-per-draft-UCB + per-canonical-per-draft-Thompson-Sampling + per-canonical-per-draft-Epsilon-Greedy + per-canonical-per-draft-LinUCB + per-canonical-per-draft-Contextual-bandit + per-canonical-per-draft-causal-uplift-CATE-T-S-X-DR-learner + per-canonical-per-draft-CausalML + per-canonical-per-draft-DoubleML + per-canonical-per-draft-EconML for response-variant selection. Per-canonical-compliance-gate-spec runs per-portfolio per-canonical-per-draft-FTC-fake-review-rule-of-2024 (AggregateRating must tie to real reviews + Review must tie to real reviewer + no incentivized without disclosure + no suppression of negative) + per-canonical-per-draft-FTC-substantiation-doctrine + per-canonical-per-draft-FTC-MARS + per-canonical-per-draft-FTC-AI-disclosure (the AI must be disclosed if response is auto-generated) + per-canonical-per-draft-FTC-endorsement-guides + per-canonical-per-draft-FTC-Made-in-USA + per-canonical-per-draft-FTC-Green-Guides + per-canonical-per-draft-FTC-Negative-Option-ROSCA + per-canonical-per-draft-FTC-MAP + per-canonical-per-draft-FTC-Health-Products-Compliance-Guide + per-canonical-per-draft-CFPB-UDAAP + per-canonical-per-draft-FDD-Item-12-Item-17-Item-19 + per-canonical-per-draft-state-FRR + per-canonical-per-draft-state-UDTPA + per-canonical-per-draft-Lanham-Act-false-advertising + per-canonical-per-draft-Robinson-Patman + per-canonical-per-draft-HIPAA-no-acknowledge-as-patient (the most operationally distinctive constraint for healthcare reviews — under HIPAA Privacy Rule 45 CFR 164.502 a covered entity cannot publicly acknowledge anyone as a patient even in response to a glowing review where the reviewer self-discloses; safe-pattern responses generically reference "the team is committed to" without acknowledging the reviewer as a patient + no diagnosis + no treatment + no condition + no physician name + Safe-Harbor 18-identifier de-identification check) + per-canonical-per-draft-HIPAA-marketing-authorization-164.508 + per-canonical-per-draft-FCRA + per-canonical-per-draft-ECOA-Reg-B + per-canonical-per-draft-Fair-Housing-Act + per-canonical-per-draft-GLBA + per-canonical-per-draft-FINRA-2210 (financial-services-review-response = FINRA-regulated communication requiring supervisory review pre-use + content standard + fair-and-balanced + substantiation + 1-or-3-year retention) + per-canonical-per-draft-SEC-Reg-S-K + per-canonical-per-draft-state-bar-advertising-LegalService (state bar rules on responding to negative reviews vary significantly — some prohibit any response to negative reviews; the system must respect per-state bar rules) + per-canonical-per-draft-state-professional-licensing + per-canonical-per-draft-FDA-DSHEA + per-canonical-per-draft-FDA-OPDP-Rx-drug + per-canonical-per-draft-DEA-Schedule-II-V + per-canonical-per-draft-cannabis-state-board-CA-CO-MA-NY-MI-IL-NJ-WA-OR-NV-AZ-AK-ME + per-canonical-per-draft-alcohol-TABC-CalABC-SLA + per-canonical-per-draft-Surgeon-General-warning + per-canonical-per-draft-tobacco-FDA + per-canonical-per-draft-TCPA + per-canonical-per-draft-CASL + per-canonical-per-draft-CCPA-CPRA + per-canonical-per-draft-GDPR-Article-6-7-17-22 + per-canonical-per-draft-LGPD + per-canonical-per-draft-DPDP + per-canonical-per-draft-PIPEDA + per-canonical-per-draft-COPPA + per-canonical-per-draft-EU-AI-Act-Article-50-transparency (the AI must disclose itself as the author of the response when the draft is auto-published) + per-canonical-per-draft-EU-AI-Act-Article-13-14-15 + per-canonical-per-draft-EU-AI-Act-Article-22-automated-decision + per-canonical-per-draft-Digital-Services-Act-Article-30 + per-canonical-per-draft-Digital-Markets-Act + per-canonical-per-draft-WCAG-2.2-AA + per-canonical-per-draft-ARIA + per-canonical-per-draft-EAA-EN-301-549 + per-canonical-per-draft-Section-508 + per-canonical-per-draft-ADA-Title-III + per-canonical-per-draft-NIST-AI-RMF + per-canonical-per-draft-ISO-42001 + per-canonical-per-draft-ISO-27001 + per-canonical-per-draft-SOC-2-Type-II + per-canonical-per-draft-OPA-Rego-AWS-Cedar-Casbin-Cerbos-Oso-policy-as-code + per-canonical-per-draft-compliance-confidence-tier + per-canonical-per-draft-compliance-explainability.

What does per-draft cross-skill-handoff + per-review-response-agent-canonical-bundle do?

Per-portfolio per-draft per-canonical-per-draft-cross-skill-handoff runs per-portfolio per-canonical-per-draft-handoff-to-per-location-review-response-drafting (parent commercial pillar at /per-location-review-response-drafting) + per-canonical-per-draft-handoff-to-review-response (parent agent) + per-canonical-per-draft-handoff-to-cs-agent-assist-build-pillar (sibling build-pillar on the same review-response agent at /how-to-build-multi-location-review-response-agent-assist — the co-pilot equivalent for live CS interactions) + per-canonical-per-draft-handoff-to-auto-publish-threshold-gating (sibling skill on review-response agent) + per-canonical-per-draft-handoff-to-post-crisis-seo-repair (sibling skill on review-response agent) + per-canonical-per-draft-handoff-to-online-review-management (sibling commercial at /online-review-management) + per-canonical-per-draft-handoff-to-customer-review-software (sibling commercial at /customer-review-software) + per-canonical-per-draft-handoff-to-ai-review-response (sibling commercial at /ai-review-response) + per-canonical-per-draft-handoff-to-brand-voice-management (sibling skill on brand-spec-authoring agent) + per-canonical-per-draft-handoff-to-forbidden-phrase-library + per-canonical-per-draft-handoff-to-claims-allowlist-substantiation + per-canonical-per-draft-handoff-to-voice-attribute-extraction + per-canonical-per-draft-handoff-to-structured-spec-authoring + per-canonical-per-draft-handoff-to-borderline-routing (sibling skill on governance-decision-router agent) + per-canonical-per-draft-handoff-to-five-destination-routing + per-canonical-per-draft-handoff-to-fbc-override-learning + per-canonical-per-draft-handoff-to-multi-dimensional-threshold-routing + per-canonical-per-draft-handoff-to-tiered-pre-filter-deterministic-gates-build-pillar (sibling build-pillar at /how-to-build-tiered-pre-filter-deterministic-gates-for-ai-content-compliance) + per-canonical-per-draft-handoff-to-marketing-content-llm-as-judge-build-pillar (sibling build-pillar at /how-to-build-marketing-content-llm-as-judge-semantic-compliance-scorer — every draft passes through this LLM-as-judge before auto-publish) + per-canonical-per-draft-handoff-to-marketing-ai-autonomy-profile-configuration-build-pillar + per-canonical-per-draft-handoff-to-per-jurisdiction-compliance-multi-state-franchise-build-pillar + per-canonical-per-draft-handoff-to-master-record-build-pillar + per-canonical-per-draft-handoff-to-customer-change-event-emission-build-pillar + per-canonical-per-draft-handoff-to-cross-touchpoint-identity-resolution-build-pillar + per-canonical-per-draft-handoff-to-deterministic-probabilistic-hybrid-identity-resolution-build-pillar + per-canonical-per-draft-handoff-to-runtime-readable-behavioral-cohorts-build-pillar + per-canonical-per-draft-handoff-to-versioned-customer-history-DSAR-build-pillar + per-canonical-per-draft-handoff-to-versioned-history-regulatory-defense-build-pillar + per-canonical-per-draft-handoff-to-crm-record-creation-build-pillar + per-canonical-per-draft-handoff-to-per-location-missed-call-crm-creation-and-callback-workflow-build-pillar + per-canonical-per-draft-handoff-to-callback-schedule-link-build-pillar (sibling build-pillar at /how-to-build-per-location-callback-scheduling-for-missed-call-recovery — when a review-response includes an offline-resolution-CTA the system schedules via this pillar) + per-canonical-per-draft-handoff-to-multi-source-attribution-preserving-lead-ingestion-build-pillar + per-canonical-per-draft-handoff-to-event-tie-in-drafting-build-pillar + per-canonical-per-draft-handoff-to-per-location-dynamic-content-build-pillar + per-canonical-per-draft-handoff-to-per-location-per-cohort-two-sigma-anomaly-detection-build-pillar + per-canonical-per-draft-handoff-to-hyper-local-search-trends-build-pillar + per-canonical-per-draft-handoff-to-routing-audit-trail-build-pillar + per-canonical-per-draft-handoff-to-jsonld-generation-build-pillar + per-canonical-per-draft-handoff-to-rich-result-eligibility-scoring-build-pillar. Per-review-response-agent-canonical-bundle integrates the review-response-drafting skill with sibling skills on the same review-response agent: per-canonical-review-response-drafting (this skill) + per-canonical-cs-agent-assist + per-canonical-auto-publish-threshold-gating + per-canonical-post-crisis-seo-repair + per-canonical-review-classification + per-canonical-crisis-escalation + per-canonical-review-source-subscription. Per-canonical-end-to-end-SLA runs per-canonical-per-draft-review-source-resolve-to-class-differentiated-template-to-multi-LLM-ensemble-to-inter-LLM-agreement-to-bias-correction-to-prompt-injection-defense-to-auto-publish-threshold-to-rep-approval-tier-to-compliance-gate-to-EU-AI-Act-Article-50-disclosure-to-multi-arm-bandit-selection-to-publish-to-FBC-feedback-loop-SLA canonical-SLA.

What does per-draft audit-trail + per-canonical-end-to-end-replay do?

Per-portfolio per-draft per-canonical-audit-trail runs per-portfolio per-canonical-per-draft-canonical-audit-record (per-draft-ID + per-banner-pointer + per-location-pointer + per-review-pointer + per-canonical-review-source-snapshot + per-class-differentiated-template-snapshot + per-5-star-praise-4-star-praise-with-suggestion-3-star-neutral-2-star-concern-1-star-crisis-multi-star-nuanced-fake-review-detection-template-snapshot + per-brand-voice-spec-snapshot + per-claims-allowlist-snapshot + per-forbidden-phrase-library-snapshot + per-trademark-aware-substitution-snapshot + per-14-language-drafting-snapshot + per-draft-confidence-tier-snapshot + per-LLM-pool-snapshot + per-eval-pattern-snapshot + per-inter-LLM-agreement-snapshot + per-bias-mitigation-snapshot + per-prompt-injection-defense-snapshot + per-jailbreak-detection-snapshot + per-FBC-feedback-loop-snapshot + per-ensemble-confidence-tier-snapshot + per-auto-publish-threshold-snapshot + per-per-vertical-threshold-modifier-snapshot + per-per-state-threshold-modifier-snapshot + per-rep-approval-tier-snapshot + per-five-destination-routing-snapshot + per-per-platform-policy-check-snapshot + per-multi-arm-bandit-snapshot + per-causal-uplift-CATE-snapshot + per-FTC-fake-review-rule-of-2024-snapshot + per-FTC-substantiation-snapshot + per-FTC-MARS-snapshot + per-FTC-AI-disclosure-snapshot + per-FTC-endorsement-guides-snapshot + per-FTC-Made-in-USA-snapshot + per-FTC-Green-Guides-snapshot + per-FTC-Negative-Option-ROSCA-snapshot + per-FTC-MAP-snapshot + per-FTC-Health-Products-Compliance-Guide-snapshot + per-CFPB-UDAAP-snapshot + per-FDD-Item-12-Item-17-Item-19-snapshot + per-state-FRR-snapshot + per-state-UDTPA-snapshot + per-Lanham-snapshot + per-Robinson-Patman-snapshot + per-HIPAA-no-acknowledge-as-patient-snapshot + per-HIPAA-marketing-authorization-snapshot + per-FCRA-snapshot + per-ECOA-Reg-B-snapshot + per-Fair-Housing-Act-snapshot + per-GLBA-snapshot + per-FINRA-2210-snapshot + per-SEC-Reg-S-K-snapshot + per-state-bar-advertising-snapshot + per-state-professional-licensing-snapshot + per-FDA-DSHEA-snapshot + per-FDA-OPDP-Rx-drug-snapshot + per-DEA-Schedule-II-V-snapshot + per-cannabis-state-board-snapshot + per-alcohol-TABC-CalABC-SLA-snapshot + per-Surgeon-General-warning-snapshot + per-tobacco-FDA-snapshot + per-TCPA-snapshot + per-CASL-snapshot + per-CCPA-CPRA-snapshot + per-GDPR-Article-6-7-17-22-snapshot + per-LGPD-snapshot + per-DPDP-snapshot + per-PIPEDA-snapshot + per-COPPA-snapshot + per-EU-AI-Act-Article-50-snapshot + per-EU-AI-Act-Article-13-14-15-snapshot + per-EU-AI-Act-Article-22-snapshot + per-Digital-Services-Act-Article-30-snapshot + per-Digital-Markets-Act-snapshot + per-WCAG-2.2-AA-snapshot + per-ARIA-snapshot + per-EAA-EN-301-549-snapshot + per-Section-508-snapshot + per-ADA-Title-III-snapshot + per-NIST-AI-RMF-snapshot + per-ISO-42001-snapshot + per-ISO-27001-snapshot + per-SOC-2-Type-II-snapshot + per-OPA-Cedar-Casbin-Cerbos-Oso-policy-snapshot + per-compliance-confidence-tier-snapshot + per-canonical-audit-record) + per-canonical-immutable-WORM-storage + per-canonical-7-year-IRS-tax-retention + per-canonical-7-year-FTC-substantiation-retention + per-canonical-7-year-HIPAA-medical-record-retention + per-canonical-7-year-state-bar-record-retention + per-canonical-6-year-SEC-record-retention + per-canonical-3-year-FINRA-record-retention. Per-canonical-end-to-end-replay runs per-portfolio per-canonical-per-draft-class-differentiated-template-rewind + per-canonical-per-draft-multi-LLM-ensemble-rewind + per-canonical-per-draft-auto-publish-threshold-rewind + per-canonical-per-draft-rep-approval-rewind + per-canonical-per-draft-multi-arm-bandit-rewind + per-canonical-per-draft-compliance-gate-rewind + per-canonical-per-draft-replay-confidence-tier + per-canonical-per-draft-replay-explainability.

Engage the review-response agent

Per-portfolio per-banner per-location per-canonical-review-source-pointer + per-canonical-class-differentiated-drafting-engine-spec + per-canonical-multi-LLM-ensemble-spec + per-canonical-rep-approval-spec + per-canonical-auto-publish-threshold-spec + per-canonical-multi-arm-bandit-spec + per-canonical-compliance-gate-spec + per-canonical-audit-trail + per-portfolio-audit-trail shipped as the orchestration layer above your existing per-AI-review-response-vendor + per-LLM-foundation-vendor + per-eval-platform-vendor + per-HIPAA-compliance-vendor + per-state-bar-compliance-vendor + per-CMP-vendor primitive.

Book a scope call Browse the 32-agent catalog