Ask Ascend is an AI-powered chatbot that lets business users explore data using natural language—no SQL or technical expertise required. Built on a fully serverless architecture on AWS and powered by a large language model (LLM), it pulls live data from Ascend data platform, interprets user intent, and returns intelligent, data-grounded answers in real time. Users can see not just the answer, but also the underlying data references, with responses delivered as clear narratives, tables, or charts. Additionally, the response is specific and contextual to what the user is looking at, not just generic answers based on large dataset.
By allowing users to query any Ascend report or screen directly, Ask Ascend democratizes access to insights across the organization. It shortens time-to-answer, reinforces consistent business definitions, and standardizes metrics—while intelligent context handling ensures accurate results even when questions are phrased in different ways.
What Makes Ask Ascend Different
Most native AI add-ons only “see” the single worksheet that happens to be active on the screen. Ask Ascend takes a broader view. It analyzes data across all open worksheets within an Ascend report, so answers are informed by the full scope of relevant data rather than a narrow slice. That cross-worksheet awareness leads to richer, more contextual responses.
Ask Ascend also preserves conversational context over time. Follow-up questions build on prior answers, allowing users to interact with their data in a natural, dialogue-driven way—more like talking to an expert than running one-off queries.
Under the hood, the chatbot runs on a completely serverless, AI-driven architecture on AWS, making it both highly scalable and cost-efficient. As long as data is available in an Ascend worksheet, Ask Ascend can work with it with minimal configuration. That means:
- Rapid scalability: Easily supports large data volumes and spikes in usage.
- Effortless extensibility: New worksheets and reports in Ascend are quickly enabled for Ask Ascend.
- Brand and client flexibility: Multiple brands, clients, and specialty pharmacies can be supported in parallel, with straightforward expansion to new data sources.
What Business Users Can Ask
Ask Ascend is designed for the kinds of questions commercial, patient services, and operations teams ask every day, for example:
- “Are more patients abandoning therapy in 2026 vs. 2025 (absolute and percentage, YTD), and why?”
- “What’s historically the worst time of year for discontinuations, and what are the drivers?”
- “Which source shows disproportionately high discontinuations vs. patient volume?”
- “Show discontinuation trends for <brand> at <specialty pharmacy>.”
- “Which enrollment source had the highest total enrollments and what percentage of total did it represent?”
Each of these prompts can be answered in plain language, with the option to dive into the underlying data when needed.
How It Works: Retrieval-Augmented Generation
Ask Ascend is powered by Retrieval-Augmented Generation (RAG), a Generative AI pattern that grounds model outputs in real, verified data rather than generic training information.
- Retrieval: Data from all open Ascend worksheets is passed to the LLM, assembling a comprehensive context window.
- Augmentation: That retrieved data is incorporated into the prompt, giving the model the precise, up-to-date information it needs.
- Generation: The LLM then produces accurate, human-like responses rooted in that data—rather than guessing or hallucinating.
By combining a large language model, a serverless AWS architecture, and the Ascend platform, Ask Ascend turns exploration into a simple conversation. Every business user can move from “What happened?” to “What should we do next?” without waiting on an analyst.
While “What should we do next?” can also be addressed by our prediction models, that’s a separate story—one we’ll cover in another blog post, where we’ll also highlight our existing AI features in more detail.
