The master workbook
2.3 million cells attempting to consolidate 60 business units. Minutes to open. Crashes weekly. Version control is _v2_final_FINAL.xlsx, and the number that reaches the board came out of it.
Enterprise GHG accounting platform
The multi-tenant system of record for corporate carbon accounting, from field-level activity data to audit-ready disclosure across Scope 1, Scope 2 and Scope 3.
LAUNCHING 10 JUNE 2027
Multi-unit organizations manage regulated emissions data in disconnected workbooks, with no version control, no governed emission factors, and no audit trail. Regulators, investors and auditors have stopped accepting that.
2.3 million cells attempting to consolidate 60 business units. Minutes to open. Crashes weekly. Version control is _v2_final_FINAL.xlsx, and the number that reaches the board came out of it.
EPA, IEA, DEFRA and the API Compendium, at different vintages in every unit. Consolidated totals nobody can defend.
Who changed what, when and why lives in inbox archives. Verification surfaces it as a finding, every year.
SEC climate rules, EU CSRD and CDP land together, with no workflow and no visibility of what is at risk.
A 340% one-month diesel spike disappears inside a quarterly total. Nobody sees it until an auditor asks.
Activity data flows in from every site. Governed calculation turns it into defensible tCO₂e. Workflow, audit trail and reporting carry it all the way to the regulator.
Stationary and mobile combustion, flaring, venting, process and fugitive emissions, captured at equipment level through forms generated from the source's own parameter schema.
Purchased electricity, steam, heat and cooling. The location-based and market-based chains resolve independently and both are stored, because the GHG Protocol requires both and neither can be switched off.
Purchased goods and services, transport, waste, business travel and the rest of the value chain, by spend-based and activity-based method, on the same governed factors as everything else.
EPA's Emission Factors Hub and eGRID, UK DESNZ, IPCC and the API Compendium ship as cited seed data you adopt and then own. Every factor is effective-dated, and versions chain append-only so history never changes underneath a filed report.
Commercial catalogues stay licensed to you, never redistributed by us
Every calculation freezes what it used, the factor version, the GWP set, the unit conversions and the boundary approach, into a snapshot that is never rewritten. A recalculation supersedes; it does not overwrite.
Supersede, never delete
One staged pipeline serves every configurable entity, from structure and users to factors, formulas and activity data. Templates carry your own vocabulary and regenerate themselves when it changes. Rows validate against exactly the rules the screens use.
30-day rollback, and versioned entities land as drafts
Whether a source needs a supporting document is a property of that source, inherited from its category and overridable per source. Required means required: the submission will not move without it.
Optional, required or disabled, per source
A completion grid for the person entering data, a review queue with variance already computed for the person approving it, portfolio status for the administrator, and an evidence-first view for the auditor.
Four roles, four landing screens
GHG Protocol, ISO 14064-1, CDP, TCFD, GRI 305 and EPA GHGRP outputs, generated only from approved and locked data. Each mapping is versioned data, so a questionnaire changing next year is an update, not a release.
Every report stamped so it can be reproduced exactly
Nothing about your industry, your structure or your methodology is written into the software. That is only useful if setting it up does not take a consultant, so the first thing a new administrator sees is not an empty dashboard.
Those three level names are seed rows, not an enumeration in a database column. Rename Business Unit to Legal Entity and every screen, every generated import template and every report follows. Add a fifth level called Asset or Region and nothing migrates.
Every step writes real configuration, validated by the same rules the admin screens use. Leave at step six and come back next week to step six.
Creating a source proposes the parameters it needs. Where one already exists that means the same thing, the system says so and offers it, instead of quietly creating a fourth field called Fuel Volume. Parameters version, and each record pins the version it was captured under.
Build a calculation from parameters, factors, GWP values, constants and named variables in a visual editor. It parses against a closed grammar, the unit dimensions have to reduce to mass of CO₂e, and activating it runs its stored test cases and an impact preview first.
Every submission moves through one state machine, one submission per node per window, with each transition timestamped, attributed, and kept.
Rejected returns the submission to Draft with a documented reason, and the second review looks only at what changed. Nothing disappears.
The data month has to fall within a reporting year somebody actually created. Data cannot arrive for a year that does not exist.
The accounting freeze. A locked month stays closed until somebody reopens it for a documented reason, and finalizing a year cascades the locks.
MONTH_LOCKED
The reporting cycle. Are we collecting for this window right now, or not? A month can be unlocked and still closed, which is the normal state between cycles.
PERIOD_NOT_OPEN
All three are checked in the service layer, never only in the interface. A node with nobody assigned to report on it cannot be opened at all, because a period that is open and unfillable helps no one.
Multi-gas, unit-safe, asynchronous, and frozen the moment it is computed.
EtCO₂e = Σ ( AD × EFgas × GWPgas )
Three tiers, one fixed order, resolved against the data month. The first that matches wins, and the ones below it are never consulted.
An append-only ledger where every entry is hash-chained to the one before it. Break a link and the integrity check tells you exactly where.
Submission approved by the regional manager, with the variance comment attached
Calculation snapshot frozen against the factor version and GWP set in force for that data month
Period locked. Reopening it needs a stated reason, and the reason is part of the record
Access is granted as a role at a point in your structure, and it covers everything beneath that point. The figures below are the design targets the product is built against, not results from customers we do not yet have.
Head of GHG reporting
Owns the structure, the people, the taxonomy, the factor library, the GWP set and the final sign-off. Runs the wizard, finalizes the year, and can export everything at any time without asking us.
Design target: organization set up in under a day
Regional sustainability
Opens periods, curates sources and factors within their part of the tree, and reviews what comes back with anomaly flags and prior-period variance already computed. Cannot approve their own work.
Design target: a flagged review in 45 minutes
Field operations
The only role that can write activity data. Guided forms, automatic unit conversion, a live calculation preview, and drafts that survive a lost connection.
Design target: 3 minutes per source
External assurance
Read-only across the whole organization, with the frozen snapshots, the workflow history and the hash chain. Self-service evidence packages, so the engagement stops being a document hunt.
Design target: any figure traced in 3 clicks
Report mappings and methodology follow the frameworks your regulators, investors and verifiers name. Each one is versioned data, so a questionnaire that changes for next year is an update rather than a release.
Every organization starts from the GHG Protocol's own category set. Anything sector-specific arrives on top of it as an installable pack of seed data: categories, parameters and factors that you can edit, extend or archive. None of it is vocabulary compiled into the software, so no sector is a second-class citizen and none of it needs us to ship a release.
Pre-configured emission categories with their activity parameters, the emission factors that go with them, and the methodology references behind each one.
As data, at any point in an organization's life, not only at setup. Installing twice changes nothing, and archiving it puts the taxonomy back where it was.
Everything a pack brings is editable the moment it lands. Rename it, change a factor, add a category the pack never thought of. It is a starting point, not a schema.
Build the taxonomy yourself in the setup wizard, or import it. The baseline works unassisted, and a pack is a shortcut rather than a requirement.
No organization can reach another organization's data. That is not a configuration setting, and it is not enforced in one place where a single mistake would undo it.
Row-level security in PostgreSQL, a request-context guard, tenant filtering at the data-access layer, and role-and-node authorization above that. The database layer fails closed, and the application connects as a restricted role that cannot bypass it.
AES-256 at rest, TLS 1.3 in transit, multi-factor authentication for privileged roles, and enterprise single sign-on over OIDC or SAML.
Platform operators provision and support organizations. They cannot read an organization's emissions data, and where cross-organization read access exists at all it is read-only and written to the same audit trail as everything else.
The calculation builder parses a closed grammar with a fixed function whitelist. There is no evaluation of arbitrary code anywhere near it, so a formula cannot become a way into the system.
Published so you can hold us to it. If it is not on this list, nobody is quietly promising it to you in a sales call either.
Scope 1 and 2 complete, Scope 3, the setup wizard, the calculation builder, universal import and export, and the framework report sets.
AI working where the data is entered and reviewed, not in a separate chat window: flagging a reading that does not fit its own history, suggesting the parameter or factor a new source needs, drafting the variance explanation a reviewer would otherwise write by hand, and answering questions about an inventory in plain language. Every suggestion is attributable, refusable and recorded, because a figure a person did not choose is not a figure an auditor can accept.
The same machinery extended past carbon to health, safety and environment reporting as a whole: incidents, water, waste, air quality and workforce safety, on the one structure, the one approval workflow and the one audit trail. One book for everything an operation has to report on.
We did not design GHG Book by shipping screens and bending the data model around them afterwards. Every table, every endpoint, every audit invariant and every access boundary was specified, argued, and ratified, with the alternatives we rejected written down, before an interface existed. It is a slower way to start. It is also why an auditor will be able to trust what we build.
We are taking on a limited number of design partners ahead of the June 2027 launch. Early-access organizations shape the roadmap and lock in founding terms.
NO NEWSLETTER, NO LIST SHARING, NO SPAM