
## A certification program is only as strong as the software behind it.

For nine years, Empathy Works has been that software — building, and rebuilding, the platform GBI's entire certification program runs on.

<GbiStatBand />

### The work that has to keep up

The Green Building Initiative (GBI) is a nonprofit that certifies how sustainable a building is. Its flagship program, Green Globes, is a science-based, whole-building certification that rates commercial real estate on environmental performance, health and wellness, and resilience — at every stage of a building's life, from design through to the spaces people work in today.

That sounds tidy on paper. In practice, a certification program is a mountain of moving parts: assessments that run to hundreds of questions, worked on by several people over days; expert reviewers trading revisions with building owners; scoring rules that have to be exactly right; reports that have to be trusted; and a growing catalogue of assessment products, each one a little different. For years that work lived in bespoke one-off tools and Excel spreadsheets — slow to change, easy to get wrong, impossible to scale.

And a program with ambitions to grow — more products, more buildings, more countries — can only move as fast as the software underneath it. That is the problem GBI brought us in 2017, and it is the through-line of everything since.

### 2017 — An engine for certification

GBI didn't want another one-off tool. They wanted a platform: somewhere they could design, test, and launch a new assessment product themselves, without a development cycle for every change — and one powerful enough to handle the real complexity underneath, from GBI's scoring and data processing to the back-and-forth between clients and the Green Globes Assessors who guide them.

So we built it. A single custom platform that runs every Green Globes product, instead of a tool per product. Assessments hundreds of questions long that multiple people work on together over days. Revision tracking and in-platform communication, so owners and expert assessors stay in sync. A scoring engine that grades everything from a single headline number down to an individual element. Automated, high-quality report generation. And portfolio-wide analysis, so an owner can compare every building they manage and see where to act first.

It launched in 2017 and became the engine behind the program — eventually powering 42 assessment products. Fully custom, and still ours to evolve.

<GbiModuleCadence />

### 2019 — One home for everything

Success brought its own problem. GBI had been operating for over twenty years, and its products had accumulated across different tools and platforms over time — each with its own login, its own account, its own way of doing things. A customer with buildings in three programs had three disconnected places to manage them. Re-certifying a building, or moving it to a different assessment, meant starting over.

In 2019, GBI asked us to build the home that tied it all together: one account, one place, every project and every building — no matter which tool actually ran the assessment underneath.

That meant meeting twenty years of history head-on. We consolidated data from every legacy tool into a single portal with single sign-on, a consistent interface, and a permission system that lets a customer share one project or an entire portfolio — with access flowing through to each assessment tool automatically. We gave clients a visual project directory they can view as cards, on a map, or as a table, with filtering, grouping, and bulk actions — so an owner can import dozens of buildings at once, or start a new project from an existing one as a template.

The detail we're proudest of is the least visible. Rather than force twenty years of users to reset their passwords, we reverse-engineered the authentication encryption from GBI's legacy tools so everyone could simply log in to their new account. The payoff came during a two-day launch window, when years of data — cleaned and imported by tooling we'd built for the purpose — landed in the new portal and just worked.

We also pulled GBI's stack apart, in a good way: decoupling their public website (WordPress) and their training and education platform (a LearnDash LMS) from the core portal, so the team could change content and run courses without touching engineering — while everything stayed tightly integrated.

That single home now holds every certified building in one place — a directory that has filled in across North America over two decades:

<GbiCertMap />

### Then, scale — in two directions

With the engine and the portal in place, GBI's ambitions pointed outward.

**Measuring what actually happens in a building.** A certification is a snapshot; owners increasingly wanted to track performance over time. We built a tracker around ENERGY STAR — the U.S. government's building-energy program — that pulls in energy, water, and waste data automatically, reports each building against its own baseline and trend, and rolls the whole portfolio up so an owner can see which buildings to prioritize.

**Going global.** Building standards aren't universal — criteria, currencies, and tax rules all change by region — so taking Green Globes international was far more than translation. We combined AI-powered automatic translation with expert manual overrides, so GBI's team can fine-tune technical content for each region rather than accept a literal translation, and we built regional billing and taxation in underneath. Today the same assessment ships in English and several more languages.

### Where it stands today

Nine years on, Empathy Works is GBI's software backbone. The custom platform we built — and keep building — has certified thousands of buildings, runs 42 assessment products, and operates across multiple regions and several languages: the assessment engine and the portal their customers live in, both fully custom, both ours to grow as the program does. Behind it all, the platform has now processed close to three million assessment answers.

A certification body lives or dies on whether its software can keep up with its ambition. For nine years, GBI's has.

<GbiResponsesArea />


Since 2017 we've been GBI's software team — building the custom platform behind Green Globes certification and the unified portal their customers run on. Together it has certified thousands of buildings across North America, in English and several more languages.

Nine years building the software that runs Green Globes certification


Expertise is only as useful as the number of people who can act on it.

JLL, one of the largest real-estate services firms in the world, had spent years learning what actually makes a workplace work — from the obvious (people want control over their own temperature) to the quiet (how much sound privacy shapes whether anyone can think). They'd distilled all of it into a book, _The Green and Productive Workplace_. But a book reads the same for everyone. It can describe a good workplace in general; it can't walk your floor and tell you what's wrong with _yours_. JLL wanted to close that gap — to hand every client the book's judgment, applied to their own offices.

## Turning a book into a product

We started where the expertise lived: with the people who wrote it. Rather than work from the text alone, we sat with the book's authors and drew the reasoning out of their heads — which factors mattered, how they traded off against each other, what a good answer actually looked like. You can't productize judgment you don't understand.

The translation was the hard part. A few hundred pages of expert prose had to become a fixed set of questions and recommendations without losing the nuance that made the advice worth having — and no two clients wanted the same thing. Some chased environmental performance, others raw productivity, most a mix. So we didn't build one assessment; we built one JLL could tailor per client, standing up a focused version aimed at each customer's objectives, where they only ever saw the questions that mattered to them.

<JllAssessmentBuilder />

Collecting the answers was its own problem. A real assessment pulls on several people — a facilities lead, an HR head, a sustainability specialist — each holding a different piece, each free at a different time. We built a permission system and a collaborative workflow so the right expert answered the right question, over days or weeks, without anything getting entered twice or stalling on one person. And for clients running whole portfolios, we built powerful cloning and templating: a new building could start from one already assessed, turning fifty blank forms into a handful of real edits.

## From a score to a decision

A recommendation only matters if someone acts on it, and different people act on different things — so the same assessment came back out as several reports, each built for a different reader. A two-page executive summary made the case at a glance for the people who sign off. A detailed breakdown gave building managers the full to-do list — what to change, and in what order. Portfolio views let an owner compare their own buildings against each other, and against an anonymized industry average, to see where they stood and where to start.

<JllSectionScores />

<JllIndustryComparison />

And underneath all of it, the numbers. Because payroll is the largest line item most organizations carry, we put real weight into modelling the financial case — estimating what each improvement would cost and the productivity it would return, and explaining _why_. A decision-making body could see the highest-bang-for-buck moves, weigh them against a budget, and understand the reasoning behind every figure rather than taking it on faith.

<JllFinancialImpact />

The tool ran quietly for the better part of a decade, assessing more than a thousand office environments on infrastructure cheap enough to leave running in the background.

The book told people what a better workplace looked like. The software told them how to get to one — one office at a time.


Expertise is only as useful as the number of people who can act on it. We turned JLL's Green and Productive Workplace into an assessment that could look at a real office and hand its manager a plan.

Making a book about better workplaces into a tool that could fix one


## A standard is only as good as the data people can put into it.

For the EMerge Alliance, we turned a complex data standard into a guided form any manufacturer can fill in — no code, no spec to decipher, clean data in minutes. It's a small tool doing a big job: unblocking the shift to a new kind of power grid.

The story starts, improbably, in the 1880s.

### The world is catching up to DC

In the 1880s, Thomas Edison's direct current (DC) lost a famous fight to alternating current (AC). AC could travel long distances and change voltage along the way; DC couldn't. So the world wired itself for AC, and stayed that way for over a century.

But the conditions that made AC the obvious choice are falling away. Almost everything we plug in today — solar panels, batteries, LED lights, EV chargers, the device you're reading this on — actually runs on DC, and on an AC grid each one wastes energy converting back and forth. So a new kind of system is emerging: the DC microgrid, a local, resilient setup that keeps power as DC from the solar panel straight to the socket. It's a cleaner, more efficient way to power a building — and EMerge Alliance, a non-profit industry group, writes the standards meant to take it mainstream.

### The challenge of making it real

A microgrid is assembled from parts — batteries, converters, breakers — made by dozens of different manufacturers. To combine them, you need to know how each one behaves, in a form you can compare. But every manufacturer describes their products differently, buried in PDFs that don't line up. Comparing two products, or checking whether they'll work together, has meant reconciling mismatched data by hand — friction that slows a microgrid down before it's even designed.

EMerge's Interoperability Data Model is the answer: one shared, machine-readable format for describing this equipment, so any design tool can read it. But a standard only helps if the data actually exists in that format — and that data has to come from manufacturers. Asking them to hand-write technical files against a specification was never going to happen at scale. The standard was ready; the data wasn't.

### What we built

Empathy Works built the missing piece: a system that makes the standard itself the thing manufacturers fill in. It works in four movements.

<div className="not-prose space-y-8 my-10">
  <NumberedStep number="1" title="Define">
    The Interoperability Data Model became the foundation of the tool — every field, requirement, and rule in the
    standard became structure in the form. EMerge owns the standard; the tool follows it exactly.
  </NumberedStep>
  <NumberedStep number="2" title="Submit">
    A manufacturer fills in a clear, guided form — the right input for every field, helpful prompts where people
    get stuck, and no spec-reading required. Where the information already lives in an existing document or
    product page, it can be pre-filled for review rather than re-typed.
  </NumberedStep>
  <NumberedStep number="3" title="Review">
    Submissions arrive already validated against the standard — checked field by field before a person ever sees
    them, so no one is reconciling mismatched files by hand.
  </NumberedStep>
  <NumberedStep number="4" title="Publish">
    Every product entered comes back out two ways at once: machine-readable data any design tool can consume, and
    a clean spec sheet that's readable at a glance. Entered once, usable everywhere.
  </NumberedStep>
</div>

No code. No deciphering a spec. What once needed an engineer now takes anyone a few minutes — and what comes out is clean, standardized data that's correct by construction.

### What it unlocks

Now the data exists — and that changes everything built on top of it. Engineers can search and compare equipment in seconds instead of days. They can lay out an entire microgrid on screen and simulate exactly how it will perform — every battery, converter, and load working together — before ordering a single part. That kind of simulation is genuinely powerful, and it only works because every component speaks the same language; feed it mismatched PDFs and it has nothing to run on. That same structured data is also ready for the AI tools that increasingly work with data directly rather than reading documents. And every product a manufacturer adds makes the whole ecosystem more useful — and the shift to DC power a little faster.

The hard part was never defining the standard. It was getting the world to fill it in — and now that's the easy part.


We turned EMerge Alliance's Interoperability Data Model into a simple, guided form, so any manufacturer can produce clean, standardized product data with no code and no spec to decipher — speeding the shift to DC power.

Turning a data standard into something any manufacturer can use


The fastest way to ship something complex is to have already built the hard parts.

The Building Intelligence Quotient (BiQ) is an assessment designed to help building owners see how well they're using building-automation systems, and where they could get more from them. On paper it's the kind of project that becomes a long, expensive build: hundreds of questions across multiple sections; answers gathered from several contributors over weeks; multi-axis scoring; generated reports; and the whole thing offered in both English and Mandarin. That spec normally means months of engineering. BiQ went live in about a week.

## Configured, not coded

It could, because every hard part was already solved. BiQ was the first assessment built on Measure/Change — our platform for exactly this kind of program — and the platform already did the things that make these builds expensive. A long assessment that several contributors can fill in together over many sittings, sharing specific pages around as they go. Multi-axis scoring — a headline number, percentages, and scores on subsets of questions, all at once. Report generation. And first-class localization, so offering the assessment in Mandarin alongside English was a setting, not a translation project bolted on afterward.

So this wasn't really a development project; it was a configuration one. We stood the platform up, and the BiQ team shaped their own assessment, scoring, and reports — launching in days what would otherwise have been a year of work and a budget to match.

BiQ was the first thing we launched on Measure/Change. It proved the premise the platform was built on: the second time you build something this complex, it shouldn't cost what the first one did.

## The Solution

Every piece of BiQ was configured on top of capabilities the platform already had:

<BiqSolution />


The fastest way to ship something complex is to have already built the hard parts. BiQ was the first assessment launched on Measure/Change — live in about a week.

Our Work