Navisio AI slashes monthly cluster costs by over 60% to scale its AI platform with MongoDB Atlas

THEIR CHALLENGE

Using MongoDB Atlas to scale bottlenecks for Navisio AI

In an insurance sector where many intermediaries still process documents much as they did three decades ago, Italian InsurTech startup Navisio AI is spearheading a modern digital shift. 

Launched by two university graduates under Nano i-Tech, it set out to transform an insurance market burdened by soft rates, an aging workforce, and repetitive processes. Insurance intermediaries in Italy frequently spend 70% of their day on manual administrative work, highlighting paper contracts and sifting through complex policy wordings across dozens of carriers. Navisio AI’s solution is an AI-powered platform with a conversational interface that allows brokers to query private documents and compare them against a public library of 1,600 Italian insurance products.
To ensure the platform met real-world brokers’ needs, the startup originally partnered with 50 different intermediaries who helped co-develop the system's features and usability. As Navisio AI transitioned from research to commercial launch, the team strategically graduated from local MongoDB Community Edition to MongoDB Atlas to power their cloud operations. However, their initial architectural setup—a standalone database for each customer organization—created an unexpected roadblock to their rapid growth and as customer adoption accelerated, Navisio AI encountered severe infrastructure bottlenecks. The original, multi-database design triggered latency spikes, elevated CPU and RAM consumption, and forced costly vertical scaling. 

"We were reaching a point of no return,” said Simone Alberto, Co-founder and CTO of Navisio AI. “If we wanted to grow our customer base, the only possible solution was to scale the platform vertically, doubling the cost each time.” 

Seeking expert architectural guidance without the expense of traditional consultancies, Navisio AI turned to MongoDB Professional Services. 

OUR SOLUTION

Designing a secure, unified AI platform

By partnering with MongoDB Professional Services, Alberto and his team gained an immediate solution to Navisio AI’s scaling bottleneck, plus a clear architectural path forward. On first meeting, MongoDB architects recommended a rapid, tactical fix; introducing new collection structures and optimized database views which Navisio AI successfully put into production within two days. It instantly stabilized CPU, RAM, and latency surges.

Next, the team tackled the primary architectural change: migrating from a single-database-per-organization set up to a unified database structure utilizing tenant identifiers within the different collections.

To power its flagship "Ask Navisio" conversational assistant, the startup designed a hybrid search architecture leveraging MongoDB Atlas Vector Search and MongoDB Search for text search side-by-side to locate exact information within seconds. Because the Italian insurance market is so heavily regulated, accuracy is paramount. To build absolute trust with brokers, Navisio AI uses MongoDB to return grounded answers that are cited down to the exact document, page, and line numbers, allowing users to instantly counter check the AI's source. This foundation also supports their automated ‘workflows’—allowing brokers to rapidly extract policy data and compare coverages and pricing across a public library of 1,600 Italian insurance products to generate optimized client offers.

To satisfy the strict compliance frameworks of the Italian insurance regulator (IVASS), GDPR, and the EU AI Act, Navisio AI deployed its server on Google Cloud in Belgium, ensuring strict European data residency. “It was a really useful way to put the database and the machines within the same virtual private network,” said Alberto. The team also used MongoDB’s compliance documentation to enable their lawyer and data protection officer to create Navisio AI’s Data Processing Agreement (DPA) paperwork. 

“Before meeting MongoDB Professional Services, we were comparing apples with bananas,” said Alberto. “They suggested the right indexes to improve performance, reduce costs and improve the accuracy of results. It was a really great experience - now we’re comparing apples with apples.”

OUTCOME

Slashed costs, rapid ROI, and massive scale

Through its architectural redesign on MongoDB Atlas and collaboration with MongoDB Professional Services, Navisio AI saw immediate, business-defining results.

By migrating to a multi-tenant collection structure, the company saw its monthly MongoDB cluster costs plummet from €1,500 down to €500. Factoring in the cost of MongoDB Atlas, the Professional Services consulting fees, and internal developer hours, Alberto calculated that the startup would achieve a complete return on their investment (ROI) in just eight months.

Beyond the balance sheet, the unified developer data platform radically simplified life for Navisio AI’s engineering team. Managing transactional data, vector search, and text search through a single database connection eliminated the need to manage complex environments across multiple third-party database vendors. Developer velocity skyrocketed; because MongoDB's documentation is optimized for AI tools, the team was able to seamlessly use coding agents like Claude and MongoDB's Model Context Protocol (MCP) tools to debug and build new features much faster.

Most importantly, resolving database bottlenecks freed up Navisio AI’s exceptionally lean IT team of just four and a half people. "We must focus on the product," Alberto explained. "We need some sort of north star to follow that we can trust is a good solution – one we can follow without performing all the research around it.'" Now, instead of spending weeks researching database fixes or debugging latency peaks, the team can hand architectural questions over to MongoDB experts and shift their undivided attention back to building new product features.

With this robust foundation, Navisio AI’s customers have already reported faster, more accurate results. Today, the company stands at 500 users. As it prepares to scale to 1,000 by the end of the year, and eventually toward 5,000 users, the team is confident that MongoDB has given them a database built for massive, cost-effective growth. Looking ahead, Navisio AI is already collaborating with MongoDB Professional Services to explore advanced AI concepts like GraphRAG to keep the platform at the absolute cutting edge.

“We’re focused on building the harness around our solution—creating an architecture where, even if we change a model tomorrow, we can still deliver the best output for our customers,” said Alberto. “And we are certain that MongoDB is a key pillar of this harness.”