Hardware-agnostic dispatch · FERC Order 2222

The neutral dispatch layer
powered by agentic intelligence.

Co-optimization is locked inside hardware ecosystems or built for 20-person trading desks. Vela couples market revenue, live battery degradation, and your contractual constraints in one explainable solve, then hands the schedule to whatever EMS you already run.

Shadow mode onboards in 48 hours · No live dispatch, no regulatory exposure
Operating across the markets that matter
ERCOT CAISO PJM NYISO 2026 MISO 2027
app.vela.energy / Command
What shadow mode has surfaced across live portfolios
0MW
Assets under management
$0M
Revenue recovered in shadow mode
0%
Average uplift vs. threshold dispatch
0hr
Time to first shadow-mode bid
Shadow mode · Verified

+44% over threshold dispatch, proven before a single live dispatch.

The ancillary services row is the one to watch: $3,190 found from zero, a program the existing strategy was never enrolled in. Degradation-aware stack optimization finds the slack that static thresholds can't see. Thirty days, no dispatch risk, no regulatory exposure. The table is what Vela would have earned.

View the shadow-mode panel →
Revenue stream
Agent
Baseline
Delta
Energy arbitrage
$12,840
$9,210
+39%
Demand response
$5,620
$5,100
+10%
Ancillary servicesNEW
$3,190
$0
Capacity revenue
$2,410
$2,410
Total (30 days)
$24,060
$16,720
+44%
Shadow mode · no live dispatch · no regulatory exposure
Platform

Three modules.
One continuous loop.

Each panel runs the real engine. Open any one to step through the full interactive experience.

01 · Bid intelligence
Dispatch Queue
Generated bid recommendations with full rationale: LMP spike probability, degradation cost, constraint audit. Approve or decline in one keystroke.
MILPSOC FLOORAUTO-APPROVE
Open →
02 · Shadow mode
30-Day P&L Proof
Run alongside your existing system. Vela shows what it would have done, side by side. Flip the switch only when you trust the numbers.
BACKTESTATTRIBUTIONNO RISK
Open →
03 · Settlement
Reconciliation
Verify delivery against commitments to the 15-minute interval. Attribute revenue per asset and reconcile against ISO settlement statements.
ISO RECONPER-ASSETAUDIT LOG
Open →
Why Vela

Each piece exists somewhere. The combination doesn't.

Fleet co-optimization isn't new (PCI, AutoGrid, Tesla) but it's locked inside a hardware vendor or built for desks with traders. Vela sits in the gap: hardware-agnostic, mid-market, live degradation state, full transparency in the solve.

Vela never touches your hardware. It reads any asset (battery, EV depot, HVAC, load) and hands a schedule to your existing EMS. The lack of hardware is the moat.

SunSpec Modbus DNP3 IEEE 2030.5 OpenADR 2.0 OCPP 2.0.1
01
Fleet-relative degradation arbitrage
Degradation is a live per-asset state (cycles, capacity fade, age) that couples assets in the solve. If Battery A is 87% through warranty and B is fresh, Vela cycles B harder today, even when A fits the price better. That edge compounds across a fleet; incumbents optimize revenue, not wear.
02
Neutral execution: no hardware skin in the game
Fluence's software sells Fluence boxes; Tesla's optimizer serves Tesla. A truly neutral decision layer can only come from a company with no hardware to defend. Vela reads any asset via protocol adapters and outputs schedules; your EMS executes. Operators don't rip out controls, they add a brain on top.
03
Explainability via shadow prices
Vela reads the dual variables of the binding constraints and surfaces them in plain language: which limit drove the call, the unconstrained play, and the dollar cost of each. Operators who see the trade-off trust the tool; those who can't override it, and that destroys the value.
04
Risk-aware dispatch (CVaR), not point forecasts
Forecasts are distributions, not numbers. Vela runs a CVaR constraint at a tunable tail percentile: max expected revenue with a floor on the bad-but-plausible case. Point forecasts go occasionally catastrophic; tail-risk control is standard in finance, rare in DER dispatch.
Who it's for

Built for operators,
not trading desks.

VPP Operators
Your DERMS dispatches. Vela decides what to bid.
Threshold-based dispatch leaves ancillary revenue uncaptured most hours. Vela solves the full revenue stack simultaneously (energy arbitrage, spinning reserve, DR, and capacity) every 15 minutes, using real-time LMP signals and your constraint set.
Per-interval stack optimization across all enrolled programs
Shadow-mode validation before any live dispatch
Human approval queue with full bid rationale
Aggregators
Each customer has different constraints. Vela handles all of them.
Managing per-customer opt-out windows, SOC floors, and ISO-specific enrollment rules across a mixed fleet is an ops problem that doesn't scale with headcount. Vela encodes those constraints and runs the bidding layer so your team stays on the relationship side.
Per-asset constraint configuration
Auditable P&L reporting by customer
Works with your existing SCADA or DERMS
C&I Asset Owners
You bought the battery for demand-charge savings. Vela adds a second stream.
Wholesale market participation has been inaccessible to most C&I sites: too complex, too much risk. Vela runs in shadow mode first, proving the incremental P&L before going live. Works with any behind-the-meter battery, regardless of vendor.
Shadow mode first, no dispatch risk day one
Stacks on top of your existing demand-charge program
Hardware-agnostic, any BESS vendor
ISO coverage

Five markets.
One agent.

FERC Order 2222 unlocks wholesale market access for distributed energy aggregations across every U.S. ISO. We're already integrated where it counts.

CAISO GA · Nov 2024 · NYISO Q4 2026 · PJM Feb 2028 · MISO 2027–29
ERCOT
Real-time & day-ahead · ADER pilot
LIVE
CAISO
DA + RT + ancillary · Order 2222 GA
LIVE
PJM
Capacity + regulation · Integration ready
LIVE
NYISO
Targeting Q4 2026 rollout
2026
MISO
Phased rollout 2027–2029
2027
FAQ

Questions operators ask.

No. Vela is a decision layer, not a control system. It reads asset state through a protocol-adapter layer (SunSpec, DNP3, IEEE 2030.5, OpenADR, OCPP) and outputs a recommended schedule. Your existing EMS or DERMS executes it. Nothing about your control stack changes.

Vela runs the full solve every interval against live market and asset data, but never dispatches. It logs what it would have bid and reconciles that against your actual results, so you see the dollar delta with zero dispatch risk and zero regulatory exposure before going live. Most operators see their first P&L comparison within a week.

A DERMS dispatches against thresholds; it doesn't co-optimize the full revenue stack against live degradation and tail risk. A trading desk does, but costs 20 headcount and is built for utility-scale. Vela is the mid-market intersection: hardware-agnostic co-optimization with the bidding and explainability layer built in.

Every solve surfaces its binding constraints and their shadow prices in plain language: which limit drove the decision, what the unconstrained play would have been, and the exact dollar cost of each active constraint. You approve or decline with the full rationale attached. It's a co-pilot you can audit, not a black box.

Shadow mode onboards in about 48 hours once we have read access to your asset telemetry and market enrollment. No hardware changes, no dispatch. You're watching the delta within the week.

Shadow mode · no dispatch risk on day one

The grid clears
in 15 minutes.

Shadow mode onboards in 48 hours. You'll see your first P&L comparison within the week. No dispatch, no exposure, just the delta.