BESS Grid Optimization Agent using AI SDK & Vercel AI Gateway
Saudi Arabia signed $1.16bn in Battery Energy Storage System (BESS) agreements to stabilize peak grid load. Build an autonomous telemetry agent that ingests hourly dispatch signals and battery health metrics to calculate optimal charge/discharge cycles. Achieve grid balancing cost savings above 15% with zero safety constraint violations.
What you are building
The core problem, expected build, and operating context for this challenge.
Implement an AI SDK stream agent integrated with Vercel AI Gateway to parse battery telemetry and optimize real-time energy dispatch.
How work is evaluated
Evaluates the BESS dispatch agent on charge strategy optimization accuracy and tariff savings.
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.
soc_safety_check
Ensures state-of-charge never falls below 10% or exceeds 95%
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
cost_efficiency
Percentage of theoretical maximum arbitrage profit achieved • target: 85 • range: 0-100
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
What you should walk away with
Stream battery telemetry data using Vercel AI SDK runtime handlers.
Integrate Vercel AI Gateway for resilient tool calling and fallback management.
Calculate degradation-aware battery dispatch schedules based on dynamic power pricing.
Ensure adherence to strict thermal and state-of-charge safety limits.
Reference links and supporting material
Hourly grid power price data and real battery degradation characteristics derived from GCC energy grid telemetry.
Challenge setup
Evaluates the BESS dispatch agent on charge strategy optimization accuracy and tariff savings.
Challenge input
JSON containing hourly grid price, SOC baseline, and battery limits
AI SDK
Provides core agentic runtime for TypeScript.
Vercel AI Gateway
Offers reliable streaming and fallback routing.
Evaluated output
JSON array of actions (charge, discharge, hold) with target MW power
- Ensures state-of-charge never falls below 10% or exceeds 95%
- Percentage of theoretical maximum arbitrage profit achieved • target: 85 • range: 0-100
- Benchmark: GridAgent-Bench
- Cost Efficiency target: 85
- 1 public reference case
- JavaScript execution harness
- javascript 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
- RAI · Optional
- All Hands AI · Optional
- Vercel AI Gateway · Required
- Vercel AI Gateway · Required
Evaluation contract
- soc_safety_check · Weight 1
- cost_efficiency · Weight 1
Metrics
- cost_efficiency · Target 85
Benchmarks
- GridAgent-Bench
- energy
Reference cases
- 1 public reference case
Workflow tasks
- Task 1
Input: JSON containing hourly grid price, SOC baseline, and battery limits
Output: JSON array of actions (charge, discharge, hold) with target MW power
Execution harness: JavaScript
Evaluation tools
- AI SDK · Agent Framework
- Vercel AI Gateway · Model Gateway
Recipe freeze marker
No recipe freeze time is recorded.