SquadStack AI Adds Visual Intelligence to Sales Calls
SquadStack.ai has launched its Humanoid Vision Agent, a voice AI product that is already live with IndiaMART.
The system is designed to bring visual intelligence into customer engagement workflows by analysing product images, listings and seller data before a sales call begins. This gives the agent context before the conversation starts, rather than requiring it to begin from scratch.
The approach targets a common weakness in voice AI systems, which often rely on repetitive questioning because they do not know what a customer has already viewed or shown interest in. In sales and customer service settings, that can lead to longer calls and a disjointed experience.
Humanoid Vision Agent processes product images to identify attributes such as colour, material and design. It also extracts information from titles and descriptions and matches buyer requirements with seller data before the interaction starts.
That allows the agent to use visual and listing context during live conversations, reducing redundant queries and making calls more closely aligned with a customer's stated or implied requirements.
IndiaMART is the first named deployment. There, the system supports buyer-seller interactions by extracting signals from product images and incorporating them into calls in real time.
The launch reflects a wider shift in voice AI from a focus on speech quality to broader use of context and multiple data sources. Companies in the sector are increasingly combining conversational systems with information from product catalogues, browsing behaviour and transaction histories to improve how automated calls are handled.
Apurv Agrawal, Co-Founder and Chief Executive Officer at SquadStack.ai, said the company sees that shift as central to the next phase of AI calling.
"AI calling is at an inflection point where improving voice quality alone is no longer enough. The real opportunity lies in making conversations more intelligent and context-aware," said Apurv Agrawal, Co-Founder and CEO, SquadStack.ai. "With Humanoid Vision Agent, we are addressing this gap by enabling the system to interpret what the customer has already seen and understood before the interaction begins. This approach reduces unnecessary friction, makes conversations more relevant, and allows businesses to deliver faster, more efficient, and outcome-driven engagement at scale."
Founded in 2021, SquadStack.ai handles more than 750,000 calls through AI-led workflows for Indian brands including Tata, Bajaj Finserv, Kotak Securities, Axis Securities, PhonePe, Zepto, AngelOne, IndiaMART, ShipRocket and Eureka Forbes.
The product is part of the company's broader work in sales and customer experience automation. Its stated aim is to turn complex sales and service tasks into agent-led workflows supported by AI.
Context shift
Humanoid Vision Agent is built around what SquadStack.ai describes as a context-first intelligence framework. In practice, that means the system draws on several sources of information before a call rather than relying only on what is said once the conversation has started.
For marketplaces and product-led businesses, that matters because customers often arrive at a call after browsing listings, comparing products or reviewing images. A system that can recognise those earlier steps may be better placed to ask narrower questions and move the conversation forward more quickly.
The company says the system can describe products in relation to what a user is viewing while matching visual information with known buyer intent. That can make live conversations more precise and reduce the need for callers to repeat details already available elsewhere in the workflow.
SquadStack.ai also outlined plans to extend the product through broader multimodal integrations and deeper context layers. The immediate commercial test, however, will be how well the technology performs in live use cases such as IndiaMART, where visual information and buyer intent are closely linked in real time.