What it is
Yellow.ai is an enterprise platform for AI customer-service agents across voice, chat and email. The core product, Nexus, turns a company’s service procedures into agents that take an action and close the request, not just answer from a help center, and it runs on multiple underlying models rather than one provider. Nexus Vox, launched 5 May 2026, is the voice line: the vendor claims sub-conversational latency, 500+ languages and dialects, and a cloned brand voice from 10 seconds of audio, all on one stack instead of separate speech, synthesis and telephony vendors. Nexus EDGE is a desktop agent for employee-experience tasks with human approval on each action. The company calls Vox its fastest-growing product.
It was founded in Bengaluru in 2016 as Yellow Messenger by Raghu Ravinutala (CEO), Rashid Khan and Jaya Kishore Reddy, and raised $100M+ from Lightspeed, Salesforce Ventures, Sapphire Ventures and WestBridge Capital. Its latest disclosed numbers, from the August 2026 deal announcement: $34M+ revenue in the last fiscal year (unaudited), 650+ enterprise clients, 16B+ conversations a year, deployments in 85+ countries, and 135+ languages on the platform. Forrester rated it a Strong Performer in its Q2 2026 Wave for customer-service conversational AI. It holds SOC 2 Type II, ISO 27001, ISO 27701, HIPAA and PCI-DSS v4.0.1.
The fork it resolves
The buying question in 2026 is no longer “can Yellow.ai automate our contact center?” It is “is Yellow.ai a safe multi-year bet while it turns into a call-center owner?”
On 3 August 2026, Yellow.ai signed a definitive agreement to merge with Bluerock Acquisition Corp (Nasdaq: BLRK), a SPAC. The combined company would list on Nasdaq as YAI at a pro forma equity value of about $550M, on a pre-money of about $300M, roughly nine times reported revenue. The stated plan for the proceeds is to buy business-process-outsourcing (BPO) operators and rebuild them as AI-run operations on Yellow.ai’s platform. Both boards approved; Bluerock’s shareholder vote was still pending at our check, with close targeted for the second half of 2026.
That changes what you are buying. Today you license software. After close, your vendor may also own call centers that compete with your outsourcer, or buy your outsourcer outright. If the product fits and you can contract around that, Yellow.ai is a credible, lower-entry alternative to the US-priced agent platforms. If you need a vendor whose ownership and strategy will look the same in 24 months, it is not the default pick right now.
Pricing reality
Sales-led above the free tier. The public pricing page lists a Free plan (1 bot, 2 channels, 2 human agents, 100 monthly tracked users) and an Enterprise plan billed on usage by monthly tracked users, with 35+ channels and unlimited bots. Neither carries a dollar figure.
The only public price points are on the AWS Marketplace listing, checked 29 October 2026: $10,000 for 12 months on Basic (one chatbot use case, tiered conversation limits) and $25,000 for 12 months on Standard (four use cases). Those are entry contracts, not a voice deployment at contact-center volume, and AWS notes infrastructure costs may apply on top. We found no Vendr or other procurement dataset for Yellow.ai deal sizes, so budget a production voice rollout as unknown until you have a quote. For scale, comparable platforms on ooligo run from about $85K median at Netomi to $150K-$350K+ at Sierra. Yellow.ai’s realistic advantage is a lower floor than those, not a published rate.
Best for
Customer-experience and contact-center operations leaders at mid-size to large enterprises in India, Southeast Asia and the Middle East who need voice and chat agents in many local languages, and who want a single vendor for speech, reasoning and telephony at a lower entry contract than Sierra or Kore.ai. The scoped case where it wins: after-hours and overflow voice for a utility, logistics or travel operator, where Waste Connections reports after-hours voice savings and VIPdesk, itself a BPO, reports 70%+ of interactions automated with a six-week launch (both vendor-published case studies).
Not for
A US or EU buyer with a deal that must survive three years unchanged, until the SPAC closes and redemption levels are public. A team whose need stops at deflecting FAQs from a help center: Intercom Fin or Zendesk AI agents on your existing desk cost less to stand up. And a company that outsources its contact center to a BPO Yellow.ai might buy or compete with, unless legal has reviewed the conflict.
Versus the alternatives
- Kore.ai — the closest like-for-like and the longer-standing enterprise platform, a Gartner Leader with a $150M round in 2024 and a 2026 growth investment. Pick it when you need the broader platform (employee service plus customer service) and a vendor whose strategy is pure software.
- Sierra — the fastest-growing entrant, past ~$200M ARR with a $950M raise in May 2026. Pick it for a consumer brand that wants a vendor-built, outcome-priced agent and can clear a $150K+ floor.
- PolyAI and Parloa — voice-only specialists. Pick one when voice is the whole job and you value a narrower product over Yellow.ai’s breadth in languages.
- Ada, Netomi, Decagon — digital-first agent platforms. Pick these when chat and email carry most of your volume.
If none fit, start with the AI layer of the contact-center suite you already run (NICE CXone, Genesys Cloud, Five9) and measure containment on your top five intents before you add a vendor.
Watch-outs
- Your vendor may become your outsourcer’s owner or rival. A platform that owns BPOs running its own software has a conflict over who keeps the automation savings. Guard: add change-of-control and termination-for-convenience clauses, plus a full data and conversation-log export right, before signing; if your BPO appears on Yellow.ai’s acquisition list, reopen the contract.
- The headline cash is not committed cash. The $200M+ proceeds figure assumes zero redemptions from a ~$175M trust; only ~$30M of PIPE is committed. Guard: keep the first contract to 12 months with no multi-year prepayment until the post-close filing shows how much cash actually arrived.
- Headcount cuts landed on product and engineering. Inc42 reported 100+ layoffs in late 2025, mainly in engineering and product, after 40-50 in August 2025. Guard: name the implementation team and escalation path in the SOW, put support SLAs with service credits in the order form, and ask for a reference customer that went live in the last six months.
- Customer counts and Vox claims are vendor numbers. A December 2024 release said 1,100+ enterprises; the August 2026 deal release says 650+ enterprise clients. Guard: ask which definition changed, take two references in your industry and region, and pilot Vox in your top three languages, measuring latency and containment yourself rather than relying on the 500-language claim.