For a global technology research and advisory firm providing market intelligence, provider evaluations and competitive benchmarking, proprietary intelligence was a core asset. Its research spans cybersecurity, digital engineering, cloud and enterprise technology, supported by thousands of reports, provider assessments, rankings and structured datasets developed through surveys, interviews and analyst research. Technology providers and enterprise clients use these insights to evaluate market performance, compare vendors and support strategic decisions.

But that intelligence was fragmented across PDF reports, structured databases and multiple studies. Comparing two technology providers across several research areas—including cybersecurity, digital engineering, cloud and enterprise technology—could require analysts to identify relevant studies, search reports and databases, correlate rankings and qualitative findings, validate them against original research and turn the result into a coherent response.

Individual queries could take two to three hours. Comprehensive reports could require several weeks to analyse, assemble and review. The model depended heavily on analyst expertise and constrained personalised research at scale. Clients largely relied on standard reports or manually prepared responses, with limited ability to explore the underlying research independently.

The client wanted a self-service intelligence product that could answer complex questions across structured and unstructured data, preserve proprietary rankings and scores, ground responses only in approved research, trace every material claim to source, support customised analysis and create a scalable subscription-led digital offering.

Research had scale. Access did not.

Building a Multi-Agent Research Intelligence Platform

Persistent developed a customised agentic AI research platform that lets users ask questions in natural language and receive synthesised, evidence-backed answers from the firm’s proprietary research. The solution combines an orchestration agent, a specialised SQL agent and multiple analytical tools to interpret queries, identify relevant information and generate contextual responses.

An orchestration agent interprets each query’s intent and scope, including providers, research categories, geographies and time periods, then determines which tools and data sources to invoke. A specialised SQL agent discovers relevant database structures and fields, resolves variations in company, study and category names, builds and refines complex queries, retrieves exhaustive datasets across studies, quadrants and time periods and validates information before it reaches the response.

The platform also searches unstructured reports and supporting documents, combining structured rankings and scores with qualitative analyst commentary. This supports straightforward requests as well as historical comparisons, provider-performance trends and analysis across multiple research domains.

McKinsey’s 2025 State of AI survey found that 62% of respondents said their organisations were at least experimenting with AI agents, while nearly two-thirds had not yet begun scaling AI across the enterprise. The gap between experimentation and scale makes workflow design, orchestration and grounding increasingly important.

Where Orchestration Meets Grounded Research

Persistent designed the solution to prevent the model from answering with general training knowledge. Responses are generated only from approved repositories, while rankings, scores and other structured values are preserved from source database to final response.

Every material statement links back to the relevant report, document or database record. Users can review and rate answers, adding feedback when more depth, context or domain specificity is needed. That feedback refines prompts, contextual logic and response quality across releases while retaining human oversight.

Persistent also built ingestion pipelines that connect the platform with source systems. Changes in the source environment are synchronised so answers reflect the latest available research.

Deloitte’s 2026 State of AI in the Enterprise reports that only one in five companies has a mature governance model for autonomous AI agents. For a research product where scores, rankings and assessments must remain intact, that finding reinforces the importance of source lineage and human oversight. Persistent used its internal AI playground capabilities as the starting foundation and customised them around the client’s research, integration and governance requirements. The accelerator shortened development time by approximately 30–40%.

Grounding mattered because faster answers still had to remain faithful to the research.

Business Impact

From Manual Research to Decision-Ready Intelligence

• Approximately 80% reduction in report-generation time, reducing automated research outputs from days or weeks to less than a days.
• 60–70% reduction in human intervention across retrieval, correlation and narrative generation
• Up to 90–95% reduction in human intervention for more mature and fully automated workflows
• 80–90% reduction in operational effort and associated costs for customised intelligence
• Competitive intelligence delivered two to three times faster than through the previous manual research model.
• Complex questions reduced from two to three hours of analyst effort to seconds or approximately a minute, depending on query complexity and data volume.
• More than 95% accuracy for structured-data responses while preserving rankings, scores and provider assessments
• More than 90% accuracy across unstructured research
• Research synthesis across more than 5,000 documents

Analysts could spend more time on interpretation and advisory work while the platform handled retrieval, correlation, validation and initial narrative generation.

Self-service changed more than turnaround time. It changed the product model.

Turning Proprietary Research into a Scalable Digital Product

The platform gives clients direct access to proprietary intelligence, allowing them to move beyond static reports and explore the underlying research through follow-up questions. For the firm, that creates a more scalable operating model and a foundation for subscription-led revenue by monetising existing intellectual property through a differentiated self-service offering.

Persistent’s contribution combined agentic database exploration, grounded retrieval, source-level lineage, human feedback and synchronised ingestion. Its reusable AI foundation accelerated development while the final solution remained tailored to the client’s data structures, query patterns and traceability standards.

Transform Your Research Repository. Build a Grounded Agentic Intelligence Product. Talk with Persistent.