Yellow.ai

Yellow.ai currently positions itself as an enterprise Agentic AI platform for customer and employee service. It markets autonomous voice, chat and email interactions, 15+ LLMs, 150+ plug-and-play integrations, multimodal context, automated testing and 1,300+ global brands. The page packages Agent Builder 2.0, Conversational Knowledge Base, AI Copilot, Automated Testing and Analyze rather than leading with the earlier Dynamic Automation Platform name.

Current documentation describes an enterprise customer-service automation platform for autonomous chat, voice, email and messaging across support, sales, onboarding and workflow execution. Automate, Assist and Analyze cover agent building, Agentic RAG, copilot/testing, suggested human-agent responses and actions, User 360, contextual human escalation, resolution/CSAT/automation/goal analytics, custom dashboards, 35+ channels, multilingual deployment and named integrations including Zendesk, Salesforce, Genesys, Workday and SAP.

Agent Builder 2.0 turns stated goals into structured prompts, recommends prompt and workflow repairs, and builds multi-step agents that choose and execute actions such as support-ticket creation, order initiation and approval routing. Workflows can trigger from abandoned carts, payment confirmations and order updates. A Super Agent delegates to domain agents while retaining context; agent-to-agent and human handoffs preserve context. Versioning, rollback, experiments, reasoning/action traces, prompts, decision paths, logs and replay are documented. Page-level counters claim 10,000 agents, 400,000 conversations in 30 days, 200,000 workflows in the first 30 days, 3x faster deployments, 93% containment, 40% higher CSAT and 60% operating-cost reduction without a named cohort.

Nexus is documented as a next-generation workspace at nexus.yellow.ai, distinct from the cloud.yellow.ai platform. Its v3 engine supports a Super Agent, specialist agents and tools, two-stage routing, ordered numbered-step execution, parent-child triggers with return to the parent, history compaction, per-turn traces, routing logic, widget building, AI Trust Centre and versioned configuration. Existing customers are told to switch workspaces or contact their account team for migration/access.

Nexus Vox is marketed as a current enterprise voice system supporting 500+ languages and dialects, including mixed-language examples. Yellow.ai describes the voice agent as able to book a flight, refund a charge, reset a password or resolve a ticket, and claims sub-400-millisecond response latency. These are product claims rather than named-customer outcomes.

The Agentic RAG knowledge product asks clarifying questions, remembers conversations, reasons across steps and can collect user data, trigger workflows or escalate. It ingests documents, sitemaps, S3, Google Drive, SharePoint, APIs, ServiceNow and Salesforce and supports scheduled hourly, daily or weekly synchronization. Anonymous page aggregates claim 60% operating-cost reduction, 2x faster go-to-market, up to 90% search accuracy and 90% deflection without metric lineage.

Inbox is the platform's unified human-helpdesk layer across 35+ channels, covering live chats, email tickets, contacts, analytics, mobile access and escalation. It supports AI-agent orchestration and preserves history for human support.

Yellow.ai documents Freemium and Premium plans. Freemium includes 5,000 monthly bot conversations, FAQ automation, unlimited human-agent seats, 500 chat/email tickets per month, two channels and development/live environments. Voice, SMS and advanced analytics are restricted. Premium is customized through a product expert, adds sandbox/staging/production and enterprise controls, and can be presented monthly or yearly. Module consumption includes monthly reached users and channel usage; no public Premium amount or typical contract value is disclosed.

The terms identify Bitonic Technology Labs Inc. and affiliates as Yellow.ai and define the subscribed CX Automation Platform through an Order. Paid subscriptions renew for successive one-year terms absent timely non-renewal, and support is supplied to paid users under the agreement.

The privacy policy applies to services offered by Bitonic Technology Labs Inc. and its affiliates collectively as Yellow.ai. It distinguishes direct enterprise customers from their end users but does not publish the full affiliate ownership structure.

Yellow.ai says it began in 2016 as Yellow Messenger and now names Raghu Ravinutala, Rashid Khan and Jaya Kishore Gollareddy as its three co-founders. The current leadership page lists Raghu as CEO, Rashid as CMO, Jaya Kishore as CPO and Madhav Chinta as CTO. Locations include San Mateo, Bengaluru, Singapore, UAE, Indonesia and Malaysia. The timeline records a US$4M Series A, the rebrand from Yellow Messenger to Yellow.ai and a US$78.15M Series C; investor logos name Sapphire Ventures, Lightspeed, Salesforce Ventures and WestBridge. Historical timeline claims include 4x revenue growth around Series A, 90% year-over-year revenue growth in 2022 and 1,300+ enterprises across 85+ countries in the Dynamic Automation Platform era, but no current revenue amount is disclosed.

The official company page lists Yellow.ai as privately held, founded in 2016, headquartered in San Mateo and in the 1,001-5,000 employee band. It says 1,300+ brands across 85+ countries use the platform, that it is trained on 16B+ conversations annually, operates across 35+ voice/text channels and 135+ languages, and has US$102M+ in funding. The associated-member count is not payroll verification.

Sapphire Ventures says it co-led Yellow.ai's Series C and describes the company's India and APAC roots, automation across support, sales and marketing, text/voice transactions, low-code/no-code tooling and contextual human handoff.

WestBridge identifies Yellow.ai as founded in 2016 by Raghu Ravinutala, Jaya Kishore Reddy and Rashid Khan and records its Series C investment in 2021. The portfolio page says the company served 1,000+ enterprises across 70+ countries, an older scale statement than current company surfaces.

Salesforce Ventures lists Yellow.ai's status as Private. Its portfolio metadata also names Anik Das as a co-founder, which conflicts with the current Yellow.ai About page's three-founder list and is retained as stale or unresolved investor metadata rather than current leadership evidence.

Yellow.ai documents role-based permissions for developer, admin, tester, engagement manager, insights and analytics roles, including rights over workflows, integrations, databases, training and intent mapping. It describes SSO, OAuth2, encrypted credentials, data hosting in the US, EU, Singapore, India, Indonesia and UAE, PII masking, vulnerability scanning, VPC/ACL controls and current compliance claims including ISO 27001/27701, SOC 2 Type II, HIPAA and PCI-DSS 4.0.1 service-provider coverage.

Automated Testing generates cases from SOPs, documents, website content and copilot conversations; executes hundreds of persona-, channel- or risk-specific scenarios in parallel; validates end-to-end journeys; and reports pass/fail, accuracy, empathy, expected-versus-actual responses, reasoning and transcripts. Yellow.ai says teams can enforce SLAs and tone controls and integrate quality signals with CI/CD.

Analyze provides conversation logs, unified performance dashboards, topic clustering, sentiment, resolution and containment analysis, AI-versus-human comparisons and voice-call diagnostics. It uses LLM evaluation to identify success patterns and can auto-generate knowledge-base articles from resolved interactions. Anonymous counters on the page are not customer-attributed.

The integration catalog documents CRM, ITSM, HR, utility, retail/ecommerce, live-chat and payment connectors. Examples include Salesforce, HubSpot, Zendesk, Freshdesk, ServiceNow, Workday/SAP tools, Shopify, Magento, Google Sheets, JIRA and payment systems such as Cashfree, Razorpay, Stripe, PayU and Paytm. Documented actions include creating/updating tickets, calendar events, reading/writing sheets, displaying products, placing and tracking orders, creating payment links, checking payment/refund state and connecting to live agents.

The current MCP Data Connector connects accounts such as HubSpot and Postgres, exposes their tools, lets administrators select permitted actions and allows AI agents to invoke those tools dynamically during conversations or workflows. The Call MCP Tool action can fetch or update records and trigger workflows through connected third-party applications.

Automation combines AI-driven and rule-based agents, workflows, database/API integration and testing. Both agent types can call workflows to update databases or external systems for tasks such as order tracking, appointments and ticket management. Copilot provides manual development testing and test suites provide automated regression, performance and accuracy reports.

A Super Agent determines intent and delegates to task-specific agents for examples including flight booking, order tracking and returns. Agents collect input, call workflows, invoke MCP tools, handle fallbacks and use variables for continuity. Workflows can fetch, process or update backend data, and MCP actions can fetch contacts, update records or trigger workflows.

Publishing from development/sandbox to staging and production sends an approval request to configured approvers; only authorized roles may approve or decline. Specific flows can be published separately, and production supports reverting to a previous version. This is a deployment/change-control gate, not evidence that every sensitive runtime transaction supports human approval.

The Shopify connector synchronizes user and store events into User 360, supports targeted campaigns and automated workflows, and documents abandoned-cart reminders, new-customer welcome, order confirmation and cancelled-order re-engagement with offers and recommendations. Events include checkout creation, order updates, refunds, products and paid orders. The documentation lists monitoring order changes, adjusting inventory and notifying customers as uses of order-update events, but does not fully document a standalone stock-allocation or replenishment engine.

Engage lets marketers build and execute multichannel outbound, flow and inbound campaigns across the lifecycle from acquisition through transaction, retention and growth. Channels include email, SMS, WhatsApp, Facebook Messenger, Viber and Slack; workflow campaigns support fallback channels and event-driven execution. Engagement admins can send campaigns and create audiences while engagement users can schedule them.

The platform analyzes target audience, goals and historical data to generate customized interactive campaign flows. Users can specify product promotion, feedback or payment-collection goals, channels and tones such as formal, urgency or humor, and generate an end-to-end flow. Product descriptions, reminder counts and AI-generated campaign themes can be configured.

The retail solution says buyers can view and compare products, check availability, place orders, track status, initiate exchange/refund, give feedback and make payments in chat without a human agent. It documents product and complementary-item recommendations and discounts based on purchase history/preferences, ERP-backed availability, back-in-stock opt-ins, product-delivery updates, CDP-based profiles and contextual human escalation. Generic 90%, 60% and 50% impact counters are not tied to a named cohort.

Agent Assist centralizes workflows inside the human-agent Inbox for actions such as ticket creation, order cancellation and refund processing. It receives ticket/chat context, and an agent can click Cancel order to have the order captured and marked cancelled without searching or reconfirming. Yellow.ai Support must configure/enable the assist bot, so this is a supported but gated human-in-the-loop execution surface.

Yellow.ai offers technology, channel and referral partner tracks with reselling, referrals, technical support and joint marketing. The page names Tata Consultancy Services, Eguardian and Tech Mahindra as partners for enterprise CX/EX automation.

The current official WordPress case-study inventory exposes 31 published canonical case-study URLs. The hub surfaces current featured metrics for Cape Coral, InteleTravel, Lulu Hypermarket, VIPdesk and four anonymous enterprise deployments. Each canonical page is separately retained below; the hub is not used to promote vendor-hosted proof to grade A.

Cape Coral deployed English/Spanish chat and after-hours Voice AI connected to city systems. It answers only from approved city information, conversationally creates 311 or emergency tickets and returns ticket numbers. The vendor page reports nearly/over 250,000 residents supported, 24/7 availability and two unified channels; these are deployment-scale measures, not a quantified service-quality lift.

InteleTravel supports 140,000+ advisors and launched English/Spanish chat and voice across four regions in 70 days. Yellow.ai reports a 92% resolution rate, 75% containment, average interactions over five minutes and 0.62% negative feedback while extending service to 24/7 without additional headcount. The source does not disclose metric formulas, cohorts or common measurement windows.

Lulu Hypermarket deployed English/Arabic messaging for support, store discovery, order tracking, receipts, loyalty offers, product browsing and in-store pickup with contextual human escalation. The vendor reports 3M+ users engaged within weeks, 80%+ higher open rates, 90% lower store-discovery time and 85% lower wait times. Baselines and formulas are not disclosed.

An unnamed Fortune 500 energy infrastructure company handling 20,000+ monthly support contacts deployed chat self-service, RAG knowledge, incident logging/routing, approval workflows and GenAI quick-heal. Yellow.ai reports about 30% of requests handled by automated approvals, 185% growth in knowledge-base use, 95%+ KB utilization, 50%+ chat-adoption growth, 84%+ first-response-time improvement from 62.9 to 8.16 seconds, 98%+ SLA compliance, 70%+ fewer business escalations, 56%+ inherited-backlog reduction and 95%+ CSAT. The stated response-time improvement conflicts with the approximately 87.0% reduction implied by the endpoints.

An unnamed consumer-health company deployed Voice AI for ITSM and HR across nine languages and multiple regions, integrated with ServiceNow, Azure Graph, email and Genesys. The page reports go-live in 60 days, about 40% of global call volume diverted from human service desks within 60 days and AI summaries on 100% of SIP transfers. Customer scale statements—165+ countries, 20,000+ employees, US$15B+ annual revenue and 1.2B people reached daily—describe the customer, not Yellow.ai outcomes.

An unnamed grocery group deployed English/Spanish Voice AI across 7,000+ US stores for password, account, ID and MFA support, integrated with IAM, Nexthink, Systrack, ServiceNow and Amazon Connect. Yellow.ai reports a 92% automated resolution rate within 60 days and 1.28M+ calls handled since go-live. Store count is deployment context, not a business outcome.

An unnamed automaker deployed an employee Voice AI flow in about 60 days and expanded within 90 days. The page reports 85% automation/success, 5,000+ automated interactions, 93% CSAT and a 7x call-volume surge absorbed without added headcount. It also says daily calls rose from 20-30 to 150-200, a range implying roughly 5x-10x rather than one exact 7x ratio, and notes no permanent ASR fix was yet available.

VIPdesk used AI agents, live escalation and Zendesk/Shopify/custom API integrations for multiple client programs. One client launched in six weeks after two years of stalled work with another vendor and exceeded 70% automation; another reached 60%. One client reached 92% CSAT while others were 80%+. The page reports 250,000 sessions. Metrics describe different VIPdesk client deployments and must not be combined into one cohort.

Hero MotoCorp deployed WhatsApp agents integrated with its lead-management system for dealer/test-drive routing, nurture, generative campaigns, outbound engagement and upsell. Yellow.ai reports 33% higher overall lead conversion, recovery of 30% of lost leads in a pilot, 5M+ customers engaged, a 21% festive-season WhatsApp lead-conversion level across 16+ models and 40% journey completion among intended/high-intent customers. The page does not disclose common baselines or formulas.

Lion Air Group deployed English, Bahasa and Thai self-service on web and WhatsApp across six use cases and 52 journeys, including booking, rescheduling, luggage, policy and complaints, with human escalation and outbound campaigns. Yellow.ai reports 90% query handling, 1.8M+ engaged users and 28M+ messages in Q1 2024. The impact counter/body call the 1.8M user figure monthly, while the customer quote calls it quarterly; both reporting periods are retained as a contradiction.

Coda deployed localized website and Facebook self-service in six markets with a gamified interface and analytics. Yellow.ai reports 85% of queries automatically resolved, 2.6M+ messages per quarter and 178,000+ active users per quarter across the six markets. No baseline or formula is disclosed.

AirAsia deployed a generative knowledge agent on web and mobile for ground staff to search operational manuals and bulletins, including low-bandwidth environments. Yellow.ai reports 80% and growing query accuracy, 42,000+ critical queries handled in the initial phase, 30,000+ unique users onboarded and 400,000+ messages handled since go-live. No exact go-live date or formulas are disclosed.

Lion Parcel deployed English/Bahasa generative chat on WhatsApp, Facebook and Instagram for delivery status, tracking, payments and other shipping questions, plus WhatsApp campaigns and automated Salesforce Service Cloud ticket creation through Salesforce and OMS integration. Yellow.ai reports 85% query automation, 3.2M+ messages per quarter and 84,000+ active users per quarter.

Pelago deployed multi-LLM travel planning, bookings, cancellations, voucher retrieval, live support and Zendesk ticket synchronization. Yellow.ai reports 10,000+ quarterly users, including 5,000+ onboarded within six weeks, 50% of product inquiries automated, 60% bot accuracy and 250,000+ quarterly messages. The source does not explain why accuracy is lower than its automation rate or provide formulas.

Hyundai Motor India deployed web, Facebook, WhatsApp and Google Business Messages for model exploration, dealer discovery and test-drive booking, plus WhatsApp campaigns and Inbox. Yellow.ai reports about 1,000 vehicle sales via the agent in calendar 2022, 10% lead-to-retail conversion, 1.4M+ impressions across 14 outbound campaigns and 22-second average agent response. The narrative's 1M+ impressions is compatible as a lower-bound with the 1.4M impact counter.

An unnamed furniture retailer in Indonesia deployed English/Bahasa WhatsApp support, Zendesk ticket synchronization, promotional campaigns and commerce. Yellow.ai reports 73% of incoming queries self-served, a 49% lead-to-sales conversion rate and 27,000+ WhatsApp orders per quarter. The source discloses no baseline or calculation.

Domino's deployed an omnichannel assistant for order placement, status, special requests, post-delivery feedback and returning-customer recognition. The page title reports 70% lower query-resolution time; impact counters report 1.5M+ conversations, US$500,000 lower live-agent cost and 90% bot accuracy. The page provides no baseline amounts, measurement periods or formulas for these claims.

Bharat Petroleum deployed Urja across web, Facebook and WhatsApp in 30 days for 600 use cases and 13 languages; more than 45% of conversations were non-English. The case headline/body report US$10M in sales and 80M+ platform conversations. Impact counters report 500,000 LPG bookings via WhatsApp in four weeks, 4.4/5 CSAT and 99% bot accuracy. Attribution, baselines and metric formulas are not disclosed.

Leo Coffee deployed conversational retail on Facebook, WhatsApp and its store for service, subscriptions and orders after ecommerce had represented just over 1% of business. Yellow.ai reports INR150,000+ in revenue within a few days, 24/7 service and 100% query-resolution accuracy. No sample, denominator or independent validation supports the perfect-accuracy claim.

Bajaj Finserv deployed multilingual BLU across five channels plus voice for service, renewal and financial-product sales. The case reports more than US$100M in sales during 2020, 20M conversations handled, US$16M saved over the prior year and 95% bot accuracy. The source does not disclose revenue attribution, cost baseline or accuracy denominator.

Heads Up For Tails deployed Poppins on WhatsApp, integrated with Shopify for product browsing, purchase and tracking across a catalog of up to 3,000 categories. The impact section reports 255,000+ of revenue within two weeks, 24/7 service and 38,000+ messages within a few weeks. The page rendering does not identify the currency for the 255,000 figure, so it remains unspecified.

An unnamed insurer deployed multilingual Voice AI integrated with policy administration, policy management and billing for verification, payment, updates and reminders. Yellow.ai reports 80% lower operating cost, 12% higher bot efficiency than live agents and 5x greater scaling capacity. Baselines, denominators and measurement periods are not disclosed.

Sayurbox deployed WhatsApp/web support for order placement, tracking, confirmation and COD payment through its API. Yellow.ai reports a 45% increase in CSAT and 24/7 service; cart-abandonment reduction is stated only qualitatively.

Asian Paints deployed multilingual distributor service on WhatsApp, PWA and web for orders, tracking and invoice state. Yellow.ai reports 25% lower critical call volume, 55+ enabled languages and 24/7 support. No baseline call count or period is disclosed.

Zenyum deployed service across website, Facebook, WhatsApp and Instagram for assessments, costs, installment plans and clinic locations. Yellow.ai reports more than 12,000 new users over one year, 900,000+ messages across four channels and 25,000+ sessions. The case notes over 50,000 monthly website users as customer traffic context.

ABFRL deployed Zoe on Microsoft Teams and PWA for policy, payslip, reimbursement and meeting workflows. Yellow.ai reports that more than 67% of employees engaged with Zoe over a year and more than 450 person-hours were saved over a year. The employee denominator and time-saved calculation are not disclosed.

Tiket.com deployed self-service for flight, hotel, train, activities, refunds and promotions with human escalation and deep APIs. The source says 70-80% of queries receive instant resolution, CSAT improved 2x, and impact counters report 80% CSAT since deployment, 99.6% resolution accuracy in the last month, 80% of monthly queries resolved without agent assist and 2.8M messages with 62,000 users per quarter. The original challenge was 40,000+ queries per day.

UnionBank of the Philippines deployed web and Facebook Messenger banking self-service. Yellow.ai reports 70-80% instant query resolution, a stated 3x increase in monthly active users from 28,000 to 120,000, 51% year-over-year operating-cost reduction and 98% accuracy. The endpoints imply about 4.29x, not 3x, so the stated ratio and endpoints are preserved as contradictory claims.

Bengaluru Metro deployed multilingual WhatsApp QR ticketing, payments, smart-card recharge, refunds, routes and schedules for a stated base of 380,000+ daily commuters. Yellow.ai reports 100,000+ users since go-live, 15M+ messages and 6,000+ WhatsApp travel-card recharge requests. No go-live date or metric formulas are disclosed.

Sony India deployed English, Hindi and Bengali Voice AI integrated with Sony CRM for registration, troubleshooting, inquiries, installation, stores and human escalation. Yellow.ai reports 46,000+ calls per quarter, 13,000+ monthly service requests and 5% month-over-month growth in product/demo requests. A Sony quote separately reports 21,000+ customer/dealer calls in the first two months; the windows differ and are preserved independently.

Waste Connections deployed English/French email, web and after-hours Voice AI for pickup scheduling, billing/payment, missed pickup and vacation hold with live escalation. The title claims 250,000+ unique users per year, while the impact section reports 100,000+ unique customers served by voice per year, 40,000+ served without human intervention and 15M+ messages per quarter. The original workload is 850,000+ calls/contact requests per month. The customer quote says millions of dollars were saved over the past year without a precise amount.

A vendor-hosted interview with InteleTravel COO Dayna Jordan discusses using chat and voice AI to support a global network of more than 140,000 advisors, provide faster answers and reduce dependence on support teams. It duplicates the deployment represented in the dedicated case but supplies no separate quantified outcome on its canonical interview page.

A vendor-hosted interview with Cape Coral ITS Director Michelle Hoffmann discusses generative AI for city-site navigation, backend automation, accessibility and a start-small implementation approach. At the interview's publication, Voice AI for after-hours service is described as upcoming; the later dedicated case documents it as live. No separate quantified outcome is supplied.

The canonical vendor-hosted video interview is with VIPdesk Chief AI Officer Jeff Kramp and Yellow.ai co-founder Rashid Khan about agentic AI in customer experience. It is a qualitative duplicate of the later VIPdesk case and provides no separate metric on the canonical page.