Artificial intelligence has made it easier to write, research, analyse information and solve complex problems.

But anyone who uses AI regularly knows that getting an answer and achieving the intended outcome are not always the same thing.

A user working on a business proposal might begin with ChatGPT, move to Claude for deeper reasoning, use another model for research and return to the original platform to refine the final output.

Along the way, the user may have to repeat instructions, explain the objective again, provide background information and correct responses that gradually move away from the original requirement.

The result can be more iterations, additional model calls, increased costs and time spent managing AI rather than completing the task.

Advertisment Add Startup Pedia As A Trusted Source This is the problem Srikanth Ravinutala, founder of US-based tech company Ekva Labs, wants to address with Ekva One , a multi-model AI platform designed around a different approach to using artificial intelligence.

“The idea is to make AI more aligned with what people are trying to accomplish, rather than simply responding to what they type,” said the founder of Ekva in an exclusive interview with Startup Pedia.

What Is Ekva One and How Does It Work? US-based Ekva One brings multiple AI models into a single platform, combining intelligent model routing with a contextual intelligence layer.

The platform is designed to help users access different models without having to manually determine which one is best suited for every query.

Its routing approach considers the nature and complexity of the task to identify an appropriate model.

For instance, a complex research assignment may require a more capable model, while summarising an existing report, rewriting a paragraph or converting information into a checklist may be handled by a smaller, more cost-efficient model.

But Ekva's approach goes beyond model selection.

At its core is Contextual Intelligence (CI), built around the broader concept of Objective-Based Contextual Intelligence (OCI).

This approach is designed to maintain the objective, relevant information, previous decisions and constraints associated with a task.

“The distinction matters because a follow-up query is rarely independent of what came before it,” said the founder of Ekva.

He further explained that a user asking AI to refine a business strategy expects the system to retain the reasoning and requirements established during the initial analysis.

Ekva aims to preserve that continuity while allowing different models to handle different parts of the work.

Also Read : Hyderabad-based AI Startup Turns Everyday Work Into “Organizational Intelligence”, Reaches Rs.

5.4 Cr ARR How Ekva Routes Queries to Reduce AI Costs Consider a user preparing a market-entry strategy for a new product.

The initial request might involve analysing the market, evaluating competitors, identifying risks and developing recommendations.

Such a task may require a powerful model with advanced reasoning capabilities.

Once the initial analysis is complete, however, the user might ask the AI to summarise the findings, create a presentation outline, rewrite the executive summary or convert the recommendations into an action plan.

These follow-up tasks still depend on the original analysis, but they may not require the same level of model capability.

In a conventional workflow, users may continue using the same model throughout the conversation, even when simpler and less expensive models could handle subsequent tasks.

Ekva is designed to approach this differently.

The initial query can be routed to a more capable model to establish the analysis and direction.

A subsequent query can then be routed to a smaller model, carrying forward the relevant context, objective and previous decisions.

The smaller model can use this context to complete the follow-up task without requiring the user to recreate the entire background from scratch, subject to the context available to that model.

This allows Ekva to allocate intelligence according to the needs of each stage rather than automatically using the most expensive model throughout the interaction.

“This approach can reduce overall AI costs by approximately 40–60%, depending on usage patterns and the comparison baseline,” the founder of Ekva told Startup Pedia.

The savings are intended to come from a combination of appropriate model selection, context continuity and fewer unnecessary iterations.

Actual results will depend on the models used, task complexity, pricing and the amount of repeated work avoided.

The larger idea is to optimise the cost of completing an objective, rather than simply the cost of an individual query.

Shared Contextual Intelligence: One Context Across Multiple AI Tools Another part of Ekva's approach is Shared Contextual Intelligence.

Today, people increasingly use different AI tools for different tasks.

One application may be used for research, another for writing, another for coding and another for analysis.

Although each tool may be capable, the context surrounding the user's work can remain fragmented across applications.

Consider a team preparing a product launch.

Its objectives, target audience, product positioning, approved messaging, budget and previous decisions may already be established.

A research tool needs that context to analyse the market.

A writing tool needs it to create relevant campaign material.

An analytical tool needs it to evaluate performance against the original objectives.

Without a shared contextual layer, each application may require the team to repeat instructions or....