Spring 2026 - Now in private beta

Your savings are real. Now prove it.

Ed is an agentic workspace for energy data analysis. Upload your meters, describe what you need to prove, and get weather-normalized, IPMVP-aligned results your CFO can actually audit.

Start 7-day trial See how it works

7-day trial, full access. Cancel before day 7 - you pay nothing.

Trusted by energy teams at these organizations since 2020

Savings reports your finance team can sign off on

Degree Days for your building, not a regional average

Every number traceable to your source data

Your data never trains our models

The problem

Energy data is messy. Most tools make it worse.

Every meter, vendor, and bill speaks its own dialect. Weeks go into cleaning data that should take hours. Then the savings report lands on the CFO's desk and the questions start. Where did that number come from? Can you prove the weather was different? Most teams can't answer. That's the problem Ed solves.

Mismatched units, intervals, and timezones across meters
Utility bills that don't fit a database row
M&V reports that fall apart under CFO scrutiny
AI tools that produce confident numbers no one can audit
Weather normalization done with a lookup table, not a regression
No paper trail between raw data and reported savings

Legacy approach

Hours of data cleaning before a single chart
Rigid dashboards that can't model your portfolio
Weather normalization in a spreadsheet, no regression
Savings figures your finance team won't trust
AI that guesses instead of computes

With Ed

Upload anything - Ed maps, cleans, and structures it
Describe your portfolio in plain English, get a working workspace
Weather normalization via actual regression, not lookup tables
Savings numbers traced back to verified data
Methodology documented, auditable, defensible

What is actually different

Real engines. Not a language model guessing.

Most AI tools send your question to a model and hope the answer is close enough. Ed is built differently. Every important calculation is a real service - compiled code with a defined input, a defined output, and a traceable result. The AI decides what to call and explains what happened. The service does the math.

When Ed tells you your savings are 12.4%, it ran the regression. It did not estimate.

Real computation, documented

Regression, Degree Days, baseline scoring - all compiled services. Results include statistical quality metrics and the methodology used. Nothing is inferred from a language model's training data.

Degree Days calculated → regression run → results documented

Your data stays yours

Each workspace is isolated to your project. Your raw data is never used to train any model, never shared with third parties, and never leaves your boundary.

Your workspace only — never shared, never used for training

How a typical session works

From raw data to verified savings in one conversation

You tell Ed what you need. Ed runs the right calculations, checks the results, and hands you something you can use.

Upload your data

Drop any CSV, XLSX, or PDF. Ed reads the structure, maps your columns, identifies the meter type, and confirms what it found before doing anything else.

Reads your files and maps columns automatically

Describe what you want to prove

"Weather-normalize Building 7 against last year. Option B, Degree Days base 15.5°C." Ed selects the baseline period, explains the choice, and flags any data quality issues before proceeding.

Picks and validates your baseline period

Ed fetches weather and runs the regression

Degree Days are computed for your exact location using the Degree Day Database. The regression runs against your actual consumption data. Statistical quality is checked against IPMVP thresholds automatically.

Local weather → Degree Days → regression → quality check

Results, documented and stored

Adjusted baseline, verified savings, and statistical quality metrics — all saved to your workspace. You get a structured result you can hand to a client or a CFO without explaining how the number was produced.

Quality-checked report, ready to share

What you get

Everything your energy portfolio needs

One workspace, connected to your data, the Degree Day Database, and the M&V methodology your clients expect.

Weather-normalized regression

Degree Days from your location. Proper OLS regression. Statistical quality reported. Savings adjusted for climate, not estimated from averages.

Portfolio management

Multiple sites, multiple meters, multiple tenants. World-map overview, site-level drill-down, aggregate KPIs. Describe your structure in English - Ed sets it up.

Universal data ingestion

CSV, XLSX, PDF bills, irregular intervals. Ed maps your columns, normalizes units, and builds clean daily and monthly layers. One upload, done.

Dashboards by description

Say "monthly kWh by site, last 24 months." A draggable, resizable widget appears - right data, right scale, right resolution. No configuration screens.

Degree Day Database

Global weather station network. Heating and cooling Degree Days for any location, any base temperature, any historical period. Fetched once, cached, available instantly.

Auditable M&V reports

IPMVP Option B output: adjusted baseline, reporting period, verified savings, statistical quality indicators, methodology documented. Numbers your finance team can sign off on.

Under the hood

Real math behind every answer not a guess

Ed separates AI orchestration from computation deliberately. Methodology guides the decisions. Ed calls the right steps. Compiled services do the math. You see a traceable result.

Methodology 1

Domain rules and best practice

Encodes M&V best practice. Tells Ed what constitutes a valid baseline, when to warn, and how to frame uncertainty. Judgment, not guesswork.

Baseline selection
Prefer full calendar years
Exclude structural breaks
Warn if less than 12 months of data

Orchestration 2

Ed calls the right steps in order

Ed decides which calculations to run, passes structured inputs, and receives structured results. It orchestrates — it does not improvise the answer.

Select baseline period
Fetch weather for your location
Compute Degree Days
Run regression analysis

Computation 3

Real math, traceable output

Scores candidate periods, runs the regression, validates quality thresholds, and stores the result. Compiled code. Every number traces to your source data.

Score baseline candidates
Check statistical quality
Validate against IPMVP thresholds
Return structured, auditable results

Built-in capabilities in Ed today

Weather and Degree Day lookup Baseline quality check Baseline period selection Weather normalization Regression analysis Portfolio dashboards M&V report generation Secure data storage

Jan

Built by someone who has done this work.

Founder - Energy data consultant - Mauritius

I built Ed because I kept doing the same manual work for real portfolios: weather normalization in spreadsheets, savings figures that fell apart under finance scrutiny, no good tool for the job.

Ed is the tool. But I'm still here. If you're on MAX, you get a monthly call with me directly. Bring your messiest dataset - that's what the call is for.

Connect on LinkedIn

Pricing

Try it on a real problem.

Seven days, full access. Upload actual meter data and run a weather-normalized regression before you decide if Ed belongs in your workflow.

7-day trial

Trial

Free

7 days, full Pro access

Up to 5 locations
Full weather and Degree Day Database
Regression, normalization, M&V reporting
~50 analyses included

Start 7-day trial

Your savings are real. Prove it.

Seven days, full access, real data. If the numbers don't hold up under scrutiny, you haven't paid anything.

Start 7-day trial Talk to Jan directly

No overage charges. Cancel any time.