Overview
As sole UX designer on GridBidder, an autonomous bid-optimization platform for battery energy storage trading, I designed the interface that lets energy traders and asset operators work alongside an algorithmic bidding engine — trusting its recommendations across volatile real-time markets, while keeping a fast, auditable way to step in when something needs a human call.
Introduction
GridBidder's optimization engine bids battery storage assets into the CAISO day-ahead and real-time markets automatically, reacting to price signals faster than any person could. But traders and operators were being asked to hand real financial and operational decisions to a system they couldn't see inside — no visibility into why a bid was placed, and no fast way to intervene if market conditions changed underneath it.
Problem
How might we let energy traders and asset operators trust an autonomous bidding engine with real financial and operational stakes — without asking them to babysit every bid, and without losing the ability to override a decision the moment it stops making sense?
Research
Interviews with two personas — the Energy Trader, focused on market strategy and P&L, and the Asset Operator, focused on physical battery health and availability — mapped out where each needed visibility versus where they needed control. Full research detail is in the presentation deck.
"
I'm not going to let a black box touch my book.
Ideation
Explored how much of the optimizer's logic to expose without overwhelming a trader mid-session — landing on a layered approach: a portfolio-level view for status at a glance, with drill-down into reasoning and controls only when something needs attention. Full exploration in the presentation deck.
Solution
1. A Portfolio Dashboard for at-a-glance status. A fleet-wide view of battery storage assets — power output, availability, active alerts — across every market the portfolio trades in, so traders and operators can tell in seconds whether anything needs attention.
2. Explainability over opacity in Bidding & Settlements. Every bid ties back to the price and market signal that drove it, so a trader sees the why, not just the what — turning the optimizer from a black box into something they can reason about and stand behind.
3. A Bid Override built for speed, with a full audit trail. Overriding an autonomous bid takes as few steps as approving one — intervention needed to be as fast as the market moves — and every override is logged, so stepping in never creates risk of its own.
Impacts
GridBidder is on track to fully replace the trading team's existing insights platform. Two concrete wins came directly out of the design: day-ahead CAISO price data that used to require a manual pull-and-send handoff to a third-party forecasting partner is now visible directly in the platform ahead of the daily trading stand-up, and settlement/P&L review no longer needs a separate reporting tool — traders can ask Gridbidder AI about any anomaly (like why state-of-charge was held back on a given day) and get an explained answer in place, without leaving the workflow. The team is currently validating the interface against live CAISO day-ahead and real-time trading workflows ahead of full production rollout.
