Consider the peculiar bargain of a customer-service phone call. You have a question. The company has an answer somewhere. Between you sits a recorded voice, a menu and a stretch of music that becomes less charming with every repetition. The frustrating part is the mismatch: a request that might fit inside a short message has become an appointment with a queue.
- The job: move callers toward virtual assistants, forms, messaging or human help.
- The approach: connect to the existing IVR rather than demand a replacement.
- The lesson: judge the route by whether the customer finishes.
Deflect.ai’s proposition began inside that mismatch. Introduced by Kore.ai in 2020, it was a cloud-based call automation and digital deflection offering. A telephone call could enter through an existing interactive voice response system, or IVR, then continue through a different channel. The customer’s original choice of a telephone would no longer dictate every subsequent step.
The queue was the clue
The timing helps explain the pitch. In May 2020, Deflect.ai appeared in coverage of Kore.ai and its partner Raqmiyat’s response to the pandemic. Banks in the Gulf faced customers whose ordinary routines had been interrupted. Branch-based service was constrained. People needed help through channels they could reach from elsewhere. A crowded telephone queue was an obvious place for that pressure to become visible.
This was a service problem with a technological opening. The relevant change was in the circumstances of customers and banks. More requests had to travel through remote channels. Deflect.ai offered a way to redirect some of that traffic into automated conversations and digital forms, with human support still part of the route.
By June 30, Kore.ai had announced a dedicated financial-services division. Deflect.ai and its Virtual Banking Assistant were named among the solutions being launched. Raj Koneru, Kore.ai’s founder and chief executive, identified financial services as a focus for 2020. Deflect.ai belonged to that broader enterprise effort, giving its small, descriptive name a much larger organizational context.
Keep the switchboard. Add a door.
The interesting commercial choice was compatibility. Launch-era descriptions presented Deflect.ai as able to work with existing IVR systems and chat-agent software. A bank could consider a different customer journey while retaining infrastructure it already used. That matters to the shape of the proposition: the product addressed what happened around the queue, without making a wholesale replacement the opening requirement.
Raqmiyat’s August 2020 banking explanation made the sequence concrete. Deflect.ai received calls from an IVR, including through SIP domain transfers, and checked agent availability. It could direct the request into chat or SMS, or collect the requirement through a digital form. Ambiguous cases could go to a live agent. Routing depended on the situation the system encountered.
A form can be the clever part
There is something pleasingly unglamorous about the form in this story. Conversational AI invites attention to the voice or the bot. Yet a form can give a request structure. In an illustrative workflow, a customer might supply the details of an inquiry on screen rather than repeat them over a telephone connection. The useful question becomes which medium best suits the next piece of work.
Deflect.ai’s product category sits at this intersection of contact-center automation and digital self-service. Its intended buyers were businesses running support operations, with banks a documented target. The people experiencing it would be their customers. That distinction gives the product two tests: an operation needs a manageable workflow, and a caller needs a reason to accept the alternate route.
Kore.ai’s sales page calls the proposition Intelligent digital deflection for call centers
. The word “deflection” is revealing. To an operator, it suggests traffic diverted from an expensive channel. To a customer, the same move has to feel like progress. A shorter queue is valuable only if the destination can do something useful.
A shorter queue is valuable only if the destination can do something useful.
The economics of a useful exit
The business case was straightforward in direction: move suitable requests into more cost-effective channels and reduce waiting. Deflect.ai was marketed through a tailored demo, a sales-led enterprise approach. For a buyer evaluating such a proposition, the meaningful cost calculation extends beyond software. Integration, maintaining workflows and staffing escalations all belong in the calculation. Those are evaluation questions, rather than a reported Deflect.ai customer budget.
A useful pilot would also follow the work after submission. Suppose a digital form accepts a request but leaves it in another unattended queue. The phone line has become quieter, yet the service has barely changed. For an evaluator, the telling observation is what happens next: acknowledgment, action, an answer and a clear route back when the answer is insufficient. This is an analytical test of the proposition, not a claimed customer result.
The comparison also needs care. A conventional queue, a callback and an automated digital interaction each distribute work differently. Deflect.ai’s distinguishing pitch was the connection between the existing voice entrance and several possible destinations. Its value would depend on the requests being routed, the systems behind them and the customer’s willingness to use the channel offered.
The human at the other end
The human exit deserves equal attention. A caller who reaches an agent after trying automation should have made some progress. In the surrounding Kore.ai contact-center material, contextual handoff included passing interaction history to a person. That is an important design ambition: the next conversation should be able to use the work already done.
There is a practical experiment a reader can borrow from this design. Choose one frequent, clearly bounded request. Offer a digital route that can complete it. Keep escalation easy to reach. Then examine completion, repeated contacts and the effort required from customers. Counting the people who left the telephone queue answers only the first question.
That exercise makes the product easier to understand. The visible assistant is one part of a service chain. The business must decide which requests belong on that chain, what information travels with them and who owns the exceptions. Those decisions determine whether changing channels saves the customer any effort.
The conditions matter. A person without access to the offered digital channel still needs another route. A request requiring judgment needs someone able to exercise it. A form disconnected from the business process can merely relocate the wait. Read this way, Deflect.ai’s 2020 proposition offers a useful standard for automation: follow the customer all the way to the answer.