Updates and new features in Spark
Spark and Copilot may look similar, but they serve different worlds: Copilot targets general-interest use cases, while Spark is built for deep-tech R&D in advanced engineering, aerospace, electronics, energy, medical devices, and materials. It connects to specialized data (patents, publications, company sources), translates your prompts into queries that each database can handle, and provides agents like Research Matrix, Patent Analysis, and Value Chain Analysis to automate tasks engineers usually do manually. Instead of essays, it delivers structured outputs engineers can work with directly - and with new agents and topic spaces coming soon, Spark’s R&D focus will only deepen.
We've upgraded Spark's chat with automatic tool selection, AI-powered prompt improvement suggestions, and smart follow-up recommendations across all features.
AI structures complex research topics into manageable subproblems and creates custom evaluation criteria, forming a research matrix. You can then deploy AI agents to systematically investigate every cell in the matrix. Download results into Excel for integration with existing R&D workflows. Resume stored matrices anytime to continue building on your research.
The news analysis workflow now lets you define topics using narrative descriptions instead of just lists. This makes it easier to describe complex or nuanced topics. Additionally, all analysis results are now automatically saved to your personal Spark data vault.
Spark’s new patent analysis workflow automates the tedious process of generating patent search queries, assessing patent relevance against specific criteria, and extracting key technologies and applications from relevant patents. Results are saved to your personal vault and downloadable as Word documents.
Spark's news analysis enables customized topic tracking. It features ready-made templates, regional filtering, category organization, and role-specific reporting—delivering tailored intelligence in multiple languages to different stakeholders across your organization.
Spark's new 'wand' feature helps transform your questions into comprehensive discovery workplans, exploring finance, technology, regulatory, and application perspectives to reveal deeper insights and blind spots in your thinking.
Spark now lets you upload and discuss PDFs with smart follow-up suggestions. It handles both scanned and digital documents (up to 100 pages). We also added relevant tool suggestions in the chat, and built a new autonomous workflow for text classification.