Quant has launched the AI agent Ava with IBM for customer service work at Fortitude Re.
The system has reduced call handling times and improved first-call resolution, according to the companies.
The launch focuses on an insurance contact centre where voice and chat interactions are managed through a mix of AI systems and human staff. Quant said the system is already in use at Fortitude Re, which it described as a provider of reinsurance solutions.
Figures released with the launch show Ava resolves 84% of calls. Average handling time has fallen from 11 minutes 30 seconds to 8 minutes 30 seconds, while first-call resolution has risen from 71% to 86%.
Quant also outlined the scale it expects over the next few years, projecting contact volumes of about 910,000 by 2027. Around 160,000 of those interactions are expected to come through chat, with the rest through voice.
Quant and IBM presented the project as a response to the weak returns many AI programmes have delivered. Chetan Dube, Chief Executive Officer of Quant AI, said many organisations have spent heavily on AI systems without seeing clear financial results.
"The challenge is that despite massive investment in AI, most organisations are still not seeing meaningful returns. While there is a lot of marketing around 'agentic transformation' promising 80-90% automated resolution and dramatic improvements in first-call resolution, the reality reported by many CFOs is very different. Industry analyses, including MIT-linked findings, suggest that tens of billions have been invested in AI, yet around 95% of initiatives have failed to deliver measurable ROI. For us at Quant, the relationship with IBM has been different because it has been driven by a quantitative focus on changing that outcome. In our work with Fortitude Re, the emphasis has been on ensuring that outcomes are measurable and that the numbers tell a different story. We are operating at a projected call volume of around 910,000 contact points by 2027, with approximately 160,000 chat interactions and the remainder voice. Within this system, we are already seeing a resolution rate of 84%. Average handle time has also reduced significantly, moving from 11.5 minutes previously to 8.5 minutes with IBM Consulting Advantage and Quant working together. First-call resolution has increased from about 71% to 86%. Quant, as the name suggests, is fundamentally driven by delivering outcomes that are provable and measurable, and we are privileged to be partnering with IBM in this effort," said Chetan Dube, Chief Executive Officer of Quant AI.
Workflow Model
The product combines AI voice and digital chat in a single customer service flow. It handles requests related to policies, claims, payments and documentation. The system can authenticate customers, process payments and send forms. It can also pass more complex cases to live agents while preserving the context of the interaction.
Quant said the service works in English and Spanish, and positioned it as part of a broader shift from isolated automation tools to connected workflows involving both software and people.
Dube said many businesses have built separate automation systems without linking them into a single process, limiting the gains they expected from AI investment.
"If you look deeper at why AI investments are not delivering expected returns, there are two major points of failure that organisations need to understand as they move toward becoming agentic enterprises over the next few years. The first issue is siloed automation. Many enterprises have implemented isolated automation systems, but the connective tissue between them is missing. Without that glue that binds these systems together, the weakest link determines the strength of the entire chain. When automation is manually handed off between systems, it creates inefficiency. As one executive famously put it, 'manual automation is an oxymoron.' The second issue is that organisations have forgotten anthropomorphism in system design. When building what is essentially a conveyor belt of automation, it is not enough to connect only systems; humans must also be part of that same flow. Internal and external automation components need to be bound together with human actors integrated into the same lifecycle. Without this, end-to-end value cannot be fully realised, and the promise of an agentic enterprise remains incomplete," said Dube.
He described the Quant and IBM approach as an "active reasoning conveyor belt". In practice, that means AI tools, existing enterprise systems and human decision-makers all sit within one operating flow rather than passing work between disconnected steps.
Insurance Use
In the insurance example outlined by Quant, a customer calls to check a claim. An AI voice agent verifies identity, confirms policy details and gathers information. The case then moves through intake systems and validation stages. Claim valuation systems assess the case, fraud checks run, and beneficiary details are confirmed.
Human staff then enter the same workflow. A Senior Life Adjudicator reviews and approves the claim decision, and an AP Disbursement Specialist authorises the payment release. System agents then close the case.
Dube said this differs from a standard chatbot model because it covers the full process rather than only the first customer interaction.
"What is different in our partnership with IBM is the creation of an active reasoning conveyor belt. This system natively binds together siloed automation systems as well as human actors into a single continuous workflow, enabling end-to-end gains that would otherwise not be achievable. To illustrate this, consider a typical insurance claim scenario at scale, involving millions of policyholders and a large volume of claims, disputes, and renewals. A customer initiates a call to check the status of a claim. The interaction begins with an AI voice agent that verifies identity, confirms policy details, and collects necessary information such as policy numbers and identity verification data. What is important, however, is what happens beyond this initial interaction. The process does not remain a simple chatbot exchange. Instead, the case transitions through a full lifecycle: it is routed to intake systems, AI agents validate documentation, and claim valuation engines assess the case while checking for fraud and confirming beneficiary details. From there, human actors enter the same workflow. A Senior Life Adjudicator reviews and approves the claim decision. An AP Disbursement Specialist is then engaged to authorise payment release. Finally, system agents handle case closure and complete the processing cycle. This entire sequence demonstrates the 'conveyor belt' model in practice, where AI agents, automated systems, and human decision-makers all operate on a unified workflow. The outcome is not just point-in-time resolution but end-to-end lifecycle completion. This is fundamentally different from traditional chatbot or point-solution automation approaches. Instead of isolated interactions that fail to resolve underlying processes, this model enables full lifecycle execution of complex enterprise workflows. The difference it makes is significant. What previously might have taken 15 to 16 minutes with unresolved outcomes can now be handled more efficiently with higher resolution rates and a seamless transition from initiation to completion across both automated and human-driven stages," said Dube.
IBM also described the work as a break from the incremental improvements that have shaped insurance operations for years. Yogendra Goyal, Global Managing Partner and Head of AI First Business Operations at IBM, linked the Fortitude Re deployment to a broader rethink of how customer service is structured.
"For 25 years I've been working with insurance carriers to resolve this issue, but it was always about incremental solutions. With Quant, it moves to exponential, resulting in a truly AI-first customer experience," said Yogendra Goyal, Global Managing Partner and Head of AI First Business Operations at IBM.
The launch places customer service at the centre of Quant's pitch for what it calls an AI operating model. Rather than treating voice bots, chat tools and back-office systems as separate projects, the company argues for a single workflow that carries work from first contact through to completion.
That position reflects a wider debate in the market. Many companies have introduced AI assistants into front-end service channels. Fewer have connected those tools with core processing systems and human decision points in a single chain. Quant is presenting its figures from Fortitude Re as evidence that this model can improve both service speed and resolution rates when those links are in place.
Quant said the system retains context when cases move from automation to staff, a common weak point in many contact centre operations. In insurance, where claims, disputes and renewals often move across multiple functions, that continuity is central to the case the company is making for connected automation.
For now, the clearest measure of the launch is the operational data attached to the Fortitude Re rollout: 84% call resolution, a three-minute reduction in average handle time, and first-call resolution rising from 71% to 86%.