Wattif Methods
The papers behind every dollar of verified savings.
Wattif Methods is a small shelf of long-form technical references on how the platform earns its claims. Each paper is written for a specific audience, whether that is a CMVP reviewer, an operator or a partner, and traces every claim back to a file, table, or column shipped in the codebase.
Open work is listed honestly inside each paper. External CMVP review of the methodology is scheduled for Q3 2026; until then the alignment claims here are the engineering team's, not an independent reviewer's.
Papers
7 on the shelfHow Wattif performs IPMVP Option B measurement and verification
The Option B engine as built: the tables that hold the audit trail, the rule that gates each row, the two crons that produce the baseline and the verification windows, and the named failure modes that hold a row out of settlement. Every dollar of verified savings is anchored to one sub-circuit, one boundary meter, one twenty-eight day baseline, and one append-only row signed by a named verifier.
How Wattif delivers continuous control policies end to end
Every strategy is born from a Diagnostic finding, previewed against the customer's own historical telemetry, commissioned at Recommend authority, and settled through the same Option B engine as paper 01. This paper covers the lifecycle, the audit trail, the three delivery modes, and the safety contract.
How seven agents run a building's energy operation
The Operator, the Engineer, the Broker, the Accountant, the Consultant, the Dispatcher, and the Analyst. Each agent owns one slice of the building's energy operation through a distinct tool surface, runs at the cadence its work is meaningful over, and writes to shared tables. The team escalates to humans when an action crosses an authority ceiling. This paper covers what each agent does, how they coordinate, and what stops them.
Any algorithm that touches customer equipment should be replayable before it acts
The hardest moment in deploying a control algorithm is the leap of faith between running a model and trusting it on a customer's equipment. Replay closes the gap by paying it in evidence the customer already owns: the past, as the algorithm would have written it, played through the same code path live operation runs. This paper extracts the principle from the strategy lifecycle and states it as a standalone deployment posture.
How a finding earns the right to exist
A finding is the unit of platform output that turns telemetry into a dollar claim. Every finding row carries the evidence window it was drawn from, the hypothesis label that names the agent's confidence, the dollar amount it sized, and the recommended action. This paper covers the contract every finding satisfies, the H1/H2/H3 labels and what distinguishes them, the five sizing methods, the lifecycle from Recommended to Resolved, and the promotion path from a finding to a financed Intervention.
Authority is the contract between an algorithm and the equipment it touches
Algorithmic confidence and operational consequence are not the same axis. A strategy can be statistically excellent and still wrong about a single freezer at three in the morning. The authority ladder is a seven-level scale that gates what a strategy may do unsupervised, set per action rather than per agent, with class-based ceilings that reflect consequence rather than confidence. This paper extracts the principle from the strategy lifecycle and aligns it with IEC 61508, ANSI/ISA 84, and ISO 26262 functional-safety frameworks.
How the Wattif AHU supply air temperature reset works
AHU Supply Air Temperature reset is the single highest-impact, lowest-risk control optimisation in commercial buildings with chilled water HVAC. Every chiller kilowatt-hour passes through an AHU dispatch decision. This paper specifies the thermodynamic mechanism, the ASHRAE Guideline 36 sequence T-3-11 the strategy implements, the state architecture that makes replay faithful, the IPMVP Option B verification path, the named limits, and the responses to twelve common reviewer challenges. Validated on a two-AHU pilot at The GEAR, Singapore.
About these papers
The papers describe systems Wattif has shipped and is operating today. They are not white papers in the marketing sense; each one is a reference document that names files, tables, columns, crons, and failure modes by the names they have in the codebase. Where work is incomplete, it is named as such. Where alignment with an external standard has not yet been independently reviewed, the engineering team's claim is labelled as such.
Contact
Questions on a specific claim, a request for a clarifying paragraph, or notes from a review go to the team directly.
Dharani Malladi · dharani@wattif.io
Aish Prasad · aish@wattif.io