Behind the scenes

Our beginnings.

We have invested over 6 years and orchestrated the expertise of many building industry specialists to build Condense.  We started this adventure trying to make it easier to build green.  We were seeing how hard it was to accurately budget for green building projects, and how this would snowball into other problems during the course of the project, requiring value engineering and delaying the schedule. This project started as a simple spreadsheet combining Aide Fitch’s green building and architecture expertise with Barry Howard’s real estate finance expertise.  We quickly saw how this could be even more useful if we could improve the accuracy of the energy and cost modeling beyond a simple spreadsheet.

With help from many building industry experts, we developed innovative, fast, web based approaches for both of these. Once we had that working well and had tested our modeling system on real building projects, we began searching for ways to speed up our access to new data to feed our system.  We found the largest gap in construction cost.  As anyone in the building industry knows, it’s notoriously hard to pin down building costs.  Since buildings are built out of building products, we thought that working directly with real building products could help us increase our access and accuracy.  We started talking with building product manufacturers and salespeople, and discovered a few happy things. Building product salespeople were excited to include their products in our modeling system, because it’s not just advertising, it’s a way for them to get directly connected to customers who are specifically searching for products like theirs. But what we’re even happier about: building product salespeople were asking us if they could use our tool themselves, as a presentation tool to their customers.  This is a win-win-win because they get a great tool, we get them interacting with our tool and actively updating their data, and building owners and their teams get much greater clarity in their design process. We’re building similar relationships with product installers and subcontractors, so we can get a total cost picture including labor costs.  This is a lot of work, but we’re taking it one step at a time, focusing on specific geographic localities for our outreach (San Antonio and Denver are first).

None of this would be possible, of course, without an excellent, solid core modeling platform.  Below is a description of the major parts of our core platform.

What is Condense?

A web-based tool for quickly determining design feasibility on building performance and cost estimating.

How does it do this? 

We’ve built an energy modeling platform using a machine-learned algorithm derived from hundreds of thousands of Energy Plus models, that can also use real life building data and any other types of energy calculations.  To set up the seed Energy Plus models, we’ve performed a comprehensive sensitivity analysis to find ways to simplify energy models, excluding variables that aren’t impactful, and most importantly, eliminating the need for the user to draw building geometry (no CAD or BIM skills needed!), while achieving results within 2% of a detailed 3D model.  Baseline model creation is even easier because inputs are preloaded from the user’s choice of energy code (ASHRAE 90.1- 2013, IECC 2015, ASHRAE 90.1-1989, etc.).

What is the Condense energy algorithm?

This is a new type of energy model using techniques similar to older “reduced order normative models” (see ). The difference is the Condense model is based on thousands of full, detailed Energy Plus models rather than historic building data, which is always limited in its ability to represent all conditions and systems. The historic data could also skew an algorithm with specific or atypical samples, whereas Condense’s Energy Plus models represent a consistent range of conditions with all inputs completely controlled.


Condense can quickly generate Energy Plus & Open Studio models. Since the algorithm is based on Energy Plus models, users can also extract the Energy Plus model(s) that match a desired set of inputs from the algorithm. This Energy Plus model can then be used for engineering or more detailed documentation.