Build a LlamaIndex Oil Pricing Advisor with Voiceflow Interface for ADNOC Benchmark Shift
ADNOC recently announced a strategic shift in its oil pricing benchmark from Murban futures to Platts Dubai. This challenge requires developers to build a RAG-based (Retrieval-Augmented Generation) advisor using LlamaIndex to help traders and energy analysts understand the implications of this change. The application will ingest historical pricing data, ADNOC's official announcements, and market analyst reports to provide nuanced answers to complex queries regarding the impact on refining margins and regional trade flows. You will utilize LlamaIndex's advanced data connectors and query engines to index diverse energy datasets. To make the tool accessible to non-technical stakeholders, you will build a conversational interface using Voiceflow, integrating it via API to trigger LlamaIndex queries and present structured advice. The system must prioritize explainability, citing specific documents for every claim about the Platts Dubai vs. Murban benchmark shift.
What you are building
The core problem, expected build, and operating context for this challenge.
Create an energy market analysis assistant using LlamaIndex for advanced RAG and Voiceflow for a production-ready conversational UI.
How work is evaluated
Evaluation is based on the system's ability to accurately describe the pricing shift and provide correct source citations.
Challenge datasets
Public datasets, plus the private datasets you have access to.
How submissions are scored
These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.
Citation Check
Ensures every answer includes at least one source citation.
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
Factual Accuracy
Accuracy against energy analyst ground-truth data • target: 0.9 • range: 0-1
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
What you should walk away with
Build a multi-document RAG system with LlamaIndex using heterogeneous energy data
Implement document citation and transparency in RAG outputs
Integrate Voiceflow's conversational canvas with external LLM services via API
Analyze the impact of regional oil benchmark shifts (Platts Dubai vs Murban)
Reference links and supporting material
A collection of oil market news articles from Zawya and MEED, official ADNOC press releases, and historical pricing spreadsheets for Platts Dubai and Murban.
Challenge setup
Evaluation is based on the system's ability to accurately describe the pricing shift and provide correct source citations.
Challenge input
Question string
LlamaIndex
Used for structuring and querying energy market data.
Voiceflow
Used for building the front-end user experience.
Evaluated output
Answer string with source citations
- Ensures every answer includes at least one source citation.
- Accuracy against energy analyst ground-truth data • target: 0.9 • range: 0-1
- Benchmark: Energy RAG QA Benchmark
- Factual Accuracy target: 0.9
- 1 public reference case
- Python execution harness
- Python sandbox (unavailable on Versalist)
Summary shows the first task, up to three tools, and up to three evidence items. Open the technical recipe to view all configured items.
View technical recipe
Configured tools
- Voiceflow · Required
- Llama Index · Optional
- Aim · Optional
Evaluation contract
- Citation Check · Weight 1
- Factual Accuracy · Weight 1
Metrics
- Factual Accuracy · Target 0.9
Benchmarks
- Energy RAG QA Benchmark
- general
Reference cases
- 1 public reference case
Workflow tasks
- Task 1
Input: Question string
Output: Answer string with source citations
Execution harness: Python
Evaluation tools
- LlamaIndex · Data Framework
- Voiceflow · Conversational UI
Recipe freeze marker
No recipe freeze time is recorded.