Why I Left Autodesk and Started a Data Company

It was in September 2019 when I started thinking about my next move. The entrepreneur "bug" didn't go away after five years at Autodesk, which I had joined through the acquisition of Lagoa – a cloud-based 3D mechanical CAD and visualization company.

I had started Lagoa to make 3D collaborative and bring it to the cloud. We would first build a rendering product and then grow into mechanical modeling. When we began, nobody thought it was possible – the tech was experimental, most investors did not want to get into the space, and customers were skeptical of the cloud. Still, I had the unshakeable belief that it was the right thing to build and that we would succeed. In 2012 we launched rendering, and by 2014 we had a full mechanical CAD modeler. We got backed by great investors, paying customers, and 1000 true fans.

My Time at Autodesk

Soon after Lagoa's Series A, we received offers to buy the company. The decision to sell to Autodesk was not easy. Once there, we had the opportunity to be part of the company during a massive transition to the cloud. We joined forces with the Fusion 360 team to make cloud-based design tools, and in 2016 I helped launch the Autodesk Forge platform and became the Director of Data for Autodesk. I met incredible customers and led a large team building critical data features into the Autodesk product line.

The decision to leave was even more challenging. After 5 years, my co-founders and I had learned the drill of working at a large company and loved what Autodesk customers do. Autodesk was a great company to work for and had a fantastic future ahead. But then something changed for me.

From Bits to Bricks

I became more and more curious about how the world gets built, how the digital & physical world intersects, and the opportunities for better software. I decided to leave and make an unorthodox investment: build a multifamily residential building from the ground up. I had some experience in the manufacturing space. Still, I had never done a construction project before (except for a couple of medium-sized renovations.) I invested in startups, crypto, and public markets and needed help, so I took my uncle out of retirement – a civil engineer – and started a construction and real estate management company. We began scoping the project for a 20 unit apartment building, which we would repeat if successful, bought a piece of land and kicked off the project.

The Data Mess

I was now a real estate investor. I thought I would experience some of what Autodesk customers experienced. The beautiful design outputs and back and forth with architects and general contractors to get the building off the ground. All that we talked so authoritatively at Autodesk about during my time there. But I was wrong.

It turns out that while design is fun, and the design tools by Autodesk make that process very smooth, when it got down to estimating, it was a mix of shock and awe for me.

It didn't matter that we had a 3D design model with the latest BIM (building information management) technology available. The most advanced technology available for making estimates and looking at the project data was a spreadsheet with a pivot table. I was floored that after all the investment in the design process - all the amazing tools to make the process easier, I was left with a 30+ tabbed spreadsheet.

The good news is: I am a master of spreadsheets.

But then it just kept going onto more and more spreadsheets – even for basic things like gross sqft, quantities and lengths, materials, you name it. For each question, there was yet another spreadsheet. And to compound the problem - spreadsheets are almost out of date the minute that they are statically created. For example, the design data (which drives much of the cost) changes faster than the spreadsheet updates you get from your design team. In the meantime, we were making decisions about the project in a spreadsheet – faster than you can update the BIM model itself. So the team is often catching up on both ends, and people cannot agree with what is in these files, constantly working from outdated information. It is not uncommon to spend an entire design review session trying to find out why a change was made, who made it, and if it was agreed upon or not.

It gets worse. Comparing the cost of designs was impossible. Figuring out what is in a GC bid and comparing it to the design is also an art and another spreadsheet. As it goes into schedules and project timelines, it just gets even more challenging.

A large number of problems at their core in pre-construction are essentially data problems, and it is easy to miss something and throw your budget and timelines off.

There is much information to cope with manually in today's data-rich world, and people naturally forget stuff. The consequences are tough to swallow and can have material effects in construction.

Spreadsheets are specialized one-off tools designed to solve a problem, making them easy to adapt to virtually any situation. You can make software that replaces a spreadsheet; however, there are so many different workflows that the number of tools you would have to build grows exponentially. It also changes from project to project – building a hospital is not like building linear infrastructure or a warehouse. You also don't want to fragment all of your data into multiple tools as the real value is on connecting it.

We completed our project in 18 months (from breaking ground to final inspection), six months late. We ran over by about 10%. Remember: this was a small project with 20 units. I could give a masterclass on spreadsheets and had a very experienced team running the project. We also only had one active project to run.

There are 5000 companies doing 100x the volume of projects we did, and 500 companies doing 10000x the volume. For companies working on megaprojects it gets even more daring – almost all projects suffer cost overruns and are delivered 40% late. So I thought: there is something here to be solved. It was clear that the ability to use and leverage the data created in the construction process to better manage projects and timelines was non-existent, and the industry needed something more modern.

The Opportunity

Construction is a $10.5 trillion industry – 16% of the global GDP. The cost of bad data in the US is estimated at $3 trillion per year. I started talking to construction companies all over the world to understand their pain points with data. I learned that spreadsheets are indeed what runs most of all professional construction projects. Even companies running $100M+ projects are still relying heavily on them for all kinds of problems. It is not uncommon to see spreadsheets with 100+ tabs. Companies report that 60% of the time is spent chasing data they don’t trust, or cleaning up data before it’s ready to be used.

It was also clear that better data drives more accurate bids, project plans, and better outcomes for everyone involved – from architects to GCs and owners.

Some companies are trying to solve this internally by making custom data pipelines. Some have already spent millions of dollars and still don't have all their needed data connectors and tools. Internal tools often break and are hard to use and maintain. Invariably the people building custom internal tools end up leaving the company and taking the knowledge with them. Building custom software is hard - especially for a construction business where software engineering is not a core competency. It is just not practical for every construction company to build an internal data product from the ground up.

Companies are also adopting more tools for project management, payments, workforce management, finance, and BIM management. Companies that only had Autodesk data now also have data in Procore and others – doubling or tripling the amount of data a construction company manages. An already fragmented data landscape is becoming even more fragmented as you can no longer rely on a single vendor. Data is going dormant and unused (80% of data is reported to be unused). It's not enough to have data; you also need software to leverage it.

My New Mission

We spent 18 months in stealth working with engineers, architects, GCs, and operators to solve the industry's data problem. With better data, budgets will be more precise, projects will run more efficiently, and it will be easier to reduce waste and increase profits. With better data, businesses and communities will have better outcomes.

It stands to reason that a data-workspace for construction should exist. The industry needs more than just spreadsheets and dashboards.

We must automate data workflows to gain insights, spot errors, and build a data-driven culture that doesn't rely on manual data entry and manual processing.

At Toric, we have set out to build exactly that. A data workspace that lets you explore, clean, transform, blend and visualize data without coding. We made it easy for data-creators to make a data pipeline in minutes and for consumers to get the insights they need in real-time – without maintaining a software stack. With it you do not need to write complex queries to databases or do any ETL. We connected Toric to all the industry tools, including Autodesk, Procore, Microsoft, and many others that the industry loves and we want to support.

We kept the familiarity of a spreadsheet, grew it with smart data inferencing and data apps. With Toric, you can build an intelligent data pipeline, which is reusable from project to project and leverage all of your data and hard work.

We think construction data will never be the same after Toric. We can't wait to show it to you.

Let us know what you think,

‍Thiago da Costa

‍CEO and Co-founder of Toric

‍