Pype Dream - Autodesk to Change Construction Workflow with AI
Roopinder Tara posted on July 23, 2020 |
Autodesk to acquire Pype and its AI-based applications for construction document processing


A little help? The specifications for a major construction project can be over a thousand pages. Pype’s AI-based software aims to automate the review of specs and change the workflow of construction. (Picture courtesy of LinkedIn )
A little help? The specifications for a major construction project can be over a thousand pages. Pype’s AI-based software aims to automate the review of specs and change the workflow of construction. (Picture courtesy of LinkedIn )
Autodesk has announced its plans to acquire Pype, a DC-area based startup that has created AI-based software to manage spec books and submittals, documents common to the construction industry. Submittals are composed of shop drawings, cut sheets and other data, in paper or digital form, and are shared between engineers, architects, contractors and architects.

Pype's main products, from what we can gather on their website, are:

  • AutoSpecs, which creates a draft submittal by reading specs, including action submittals, product data, closeout submittals, tests & inspections, and more.
  • CloseOut provides a portal for collecting closeout documents and compiles a “digital turnover package.”
  • eBinder will take the turnover documentation and make it “dynamic” PDFs
  • SmartPlans, introduced just this January, seems to semi-automate getting the information from PDF drawings into submittals and schedules. For example, you can select an area of a document with a product and equipment schedules, and SmartPlans will directly send it to Excel or several BIM applications, including Autodesk BIM360.

“Tackling unstructured data in construction drawings was a major challenge for our engineering and data science teams,” said Karuna Ammireddy, Pype’s CTO and co-founder, in the company’s announcement of SmartPlans. “We used five years of project data to build and train our AI engine to surface actionable project insights.”

Autodesk’s construction software platform can help companies create such forms manually, but Pype’s software does so automatically by recognizing which data to extract from existing documents.

CEO Andrew Anagnost said Pype will further Autodesk’s objective by helping users digitize the entire construction workflow process, from planning and design to project management and field operations.

What a Waste, the Construction Industry

The biggest industry, over $10 trillion in spending worldwide, may have the most waste and inefficiency. It is estimated as much as 35% of a construction worker’s time is wasted. AEC design software companies, like Autodesk, Bentley Systems and Nemetschek, have forever maintained that if only the industry would avail itself of the tools and technologies being offered them, their waste would go down and their efficiency would go up. But the construction industry is too big and too slow -- like a dinosaur.

AI to the Rescue


Shop drawings like for these aluminum doors are a big part of a submittal. (Picture courtesy of Forest)
Shop drawings like for these aluminum doors are a big part of a submittal. (Picture courtesy of Forest)

To speed up adoption of its technology, Autodesk answers with more technology. No more will the general contractor have to pore over a spec book that could be as much as a thousand pages of drawings, material specs, warranties, etc., marking it up with a red pen, a process that can take weeks with a big project.

Till now, the technology that was supposed to help may have been adding to the information overload. Digitizing the drawings was the first step, converting analog information into digital data. The raster data was useless and had to be “vectorized” into CAD geometry. Easier said than done. CAD data was supplemented by digital documents. Torrents of data were to come from new sources, each adding to the digital load. There came point clouds from LiDAR scans and photogrammetry models. Video cameras to monitor construction made more data files. IoT devices threaten to break the dam and make “data lakes.”

Managing data on such a massive scale is not what construction workers, their managers or their firms had signed up for. And data specialists, currently at the top of the tech food chain, are being scooped up by tech firms. Asking them to consider working in a giant, slow moving industry has got to be a tough sell.

Therefore, if the construction industry and its software tools are only capable of creating a tsunami of data, and there is no hope of a cavalry of data specialists to ride in to rescue it, AI would seem to offer the logical solution. Neural networks that observe and learn from piles of data sets over years, learning this input causes that output, then be able to generate the code to do it, with accuracy, as Pype claims to have done, would fulfill the promise of machine learning.

Wood Chipper in Reverse

You can imagine a scanner that is fed paper documents, drawings, PDFs and a cloud-based engine that makes one beautiful data model from it as a wood chipper in reverse, taking in a mish mash of chips and producing a tree.

If there is one of Pype’s applications that delivers on the promise of AI and machine learning, it would be Pype’s SmartPlans.

“I believe it’s SmartPlans, where AI most comes into play right now,” says software industry analyst Monica Schnitger on her blog. “When human beings do that, they count windows, doors, wall plates, and so on and tally them up into bills of materials. These are then checked by a checker to make sure nothing was missed. If anything was unclear, a submittal was made to the architect and everyone waited for a reply. Automating this makes so much sense.”

According to Autodesk, Pype will “empower general contractors, subcontractors and owners to gain even more value from Autodesk Construction Cloud by automating critical construction workflows such as submittals and closeouts to increase productivity and mitigate project risk” and AI will “reduce tedious manual entry and human error that can lead to rework, cost overruns and schedule delays on construction projects” and “by automating these traditional manual workflows and converting real-time data into actionable insights, construction teams have the power to increase collaboration and project efficiency.”

Pype's AI-based application scans the spec books, then applies natural language processing and machine learning to put all of its elements into categories, enabling the automatic creation of request forms and other documents that need to be shared with stakeholders, according to the Wall Street Journal.

Autodesk Continues Torrid Pace of AEC Investment

Autodesk seems to have picked up again after setting a torrid pace of AEC technology acquisition in the last half of 2018

Construction Cloud was announced at Autodesk University 2019 and the company had spent over a billion dollars to gather under it.

A selective list of Autodesk’s recent construction technology investments:

  • July 2020 - Bridgit Solutions, from Toronto, for construction workforce and resource planning. Autodesk is the leading investor in a $7M round of funding.
  • July 2020 – Investment in Factory_OS, in the Bay Area, for modular construction
  • April 2020 – Aurigo, for owners. Autodesk acquires a minority equity stake.
  • December, 2018 - Acquisition of BuildingConnected, $275M
  • November, 2018 – Acquisition of PlanGrid, $875M
  • July, 2018 - Acquisition of Assemble, not far from Autodesk’s San Francisco office.

Terms of the Pype acquisition were not disclosed. Autodesk did say recent acquisitions in AEC software were “over $1.1 billion.” However, since the lowest acquisition cost that Autodesk felt the need to report for almost 20 years is $25M (PlanPlatform in 2009), we can assume it paid less that for Pype.

About Pype

caption

Pype, founded in 2015, US headquarters in Chantilly, Virginia, just 10 miles west of Washington on I66, is led by Sunil Dorairajan, who graduated from India’s vaunted IIT with a bachelors in civil engineering and went on to get a masters in construction management from Virginia Tech. Pype had received a total of $6M in funding before the acquisition, according to Crunchbase. Pype has 33 employees in the US headquarters and 41 in India, including most, if not all, of the developers, according to the Wall Street Journal.

The Competition

With Pype, Autodesk has an AI-based answer to Bluebeam, arguably the market leader leading PDF tool for AEC. Bluebeam was acquired by Nemetschek for $100 million in 2014 and instantly made Nemetschek an AEC software leader in terms of users, much the same way Trimble did it with Sketchup. Bluebeam claims it is used by over 90% of the ENR top 50. Bluebeam’s Revu may be the penultimate PDF processor and similar to Pype, it also does quantity takeoffs and submittal reviews.


Recommended For You