Briefing Paper

THE NEXT WAVE OF

INTELLIGENT DESIGN
AUTOMATION

Sponsored by
SPONSOR PERSPECTIVE

Autodesk generative design technology: Shaping the future of intelligent design

automation and empowering teams to deliver innovation and productivity.

The way things are designed, made, and used is changing at lightning speed, and
the pressure to get to market with groundbreaking products in less time is very real.
At Autodesk, a company that makes software for people who make things, we see
generative design technology delivering the ideal combination of innovation and
productivity required to help companies address the challenges of this disruption.

Powered by artificial intelligence (AI) and the computing power of the cloud, Autodesk
generative design technology gives us the ability to simultaneously generate multiple
CAD-ready solutions based on real-world manufacturing constraints and product
performance requirements. From there, the designers or engineers can filter and select
the outcomes to best meet their needs.

With its ability to explore thousands of valid design solutions, built-in simulation,
awareness of manufacturability, and part consolidation, generative design impacts
far more than just the traditional notion of design. It’s really about the entirety of
the manufacturing process providing companies with new insight to bridge the gap
from design to manufacturing, enabling major advantages from reductions in cost,
development time, and material consumption to improvements in product performance.

All of a sudden, we are no longer limited by our own imaginations or past experience;
we have the ability to design with engineering and manufacturing expertise in a way
that is almost superhuman.

What originated at Autodesk as design software inspired by nature’s evolutionary

approach, has evolved with further development and additional customer insight.
And after several months in a closed beta, Autodesk generative design is being released
within Autodesk’s Fusion 360 platform. Now, companies everywhere have access to
intelligent design automation software, giving them the power to shape the future of
making things.

As you will read here, a few of the most forward-looking companies in the world,
including Airbus, Under Armour, and Stanley Black & Decker, are already using generative
design to solve engineering challenges, demonstrating immense opportunity for design
and manufacturing across industries. As AI becomes part of all work processes and
generative design becomes the norm for product design, it will be exciting to see what
we can achieve.
THE NEXT WAVE OF
INTELLIGENT DESIGN
AUTOMATION

For decades, the process of product development has been painfully HIGHLIGHTS
iterative and rigid. Designers and engineers would take customer
With the democratization of a
requirements, create a few design concepts, experiment with possible
n

new wave of intelligent design

forms and materials, test designs virtually and physically to determine automation technology, we are
on the cusp of a paradigm shift in
how they held up in various conditions, and tweak them until they got how products are designed
one that met specifications—before they ran out of time and money. n A new breed of generative design
There simply were not enough hours in the day to come up with the software has emerged that will offer
ideal design. And little thought was given to the manufacturability of a step change in computer-aided
engineering
the resulting design.
With the democratization of a new wave intelligent design automation technology,
we are on the cusp of a paradigm shift in how products are designed—a world in
which engineers might dedicate their creative focus to what they are building rather
than how.
Product designers long ago swapped their drawing boards for computer-aided-
design (CAD) programs, which made modifications and revisions easier and, over
time, increased in functionality and decreased in price. Despite the name, however,
CAD software did little to aid in the design process; even with later bells and
whistles, it was fundamentally a tool for documenting what an engineer had come
up with in his or her head.
Thanks to low-cost, high-performance computing power available via the cloud,
and intelligent design algorithms powered by machine learning capabilities, a
new breed of generative design software has emerged that will offer a step change
in computer-aided engineering (CAE). “The nexus of software capabilities, fast
computers, thoughts about manufacturing capabilities and ability to engineer
new materials will totally revolutionize engineering and product innovation in the
future,” says Keith Meintjes, practice manager for simulation and analysis with
product life cycle management consulting and research firm CIMData.
Manufacturers of all sizes can use advanced generative design tools, which
employ algorithms to transform a designer’s requirements into product geometry,
optimizing the product design based on the conditions and constraints provided.
The tools can provide a plethora of design options in the amount of time it would
take engineers to set up a meeting to discuss just one.

Briefing Paper | The Next Wave of Intelligent Design Automation Harvard Business Review Analytic Services 1
With generative design, the options are pre- What Is Generative Design?
The phrase “generative design” has
validated to work within the manufacturing been used in architecture and civil
engineering to describe the use of
and business constraints provided. advanced computer programs to
synthesize designs for structures or
circuit diagrams, for example, based
upon input requirements.
In recent years, some manufacturers
have worked with advanced topology
The primary benefits are twofold: optimization tools to make existing
generative design tools decrease the parts or components perform better or
amount of time it takes to design a make the finished products lighter or
product, and open the door to design stronger. General Motors, for example,
options that might never have been used topology optimization to remove
dreamed up by human engineers 400 pounds of weight from the design
on their own. Engineers, with only of its latest Chevrolet Equinox, while
their own experience and skills to retaining interior space and vehicle
guide them, are naturally limited by performance characteristics.
“what they’ve seen in the past and
what they believe to be true,” says Generative design, however,
Frank DeSantis, vice president of goes much further than topology
breakthrough innovation at Stanley optimization. Generative technology
Black & Decker, an early adopter of can take the many specific goals
generative design technology. input by a designer or engineer—size,
weight, strength, style, materials,
In addition to enabling the exploration cost, schedule, manufacturability—
of multiple design alternatives for each and employ algorithms running in the
product, such software considers the cloud to produce a plethora of possible
manufacturing, cost, and schedule design solutions. Incorporating
impacts of the design as part of the machine learning and advanced
process. In the past, designers and simulation, the intelligent design
engineers would pass a design back software can rapidly cycle through
and forth in seemingly “endless thousands or millions of design
manual iteration loops,” says Bastian choices, and test configurations to
Schäfer, innovation manager with produce options that would be difficult
Airbus. With generative design, the for designers and engineers to discover
options are pre-validated to work and model efficiently. FIGURES 1 AND 2
within the manufacturing and business
constraints provided, reducing long- Generative design is just beginning
standing friction between engineering to be applied to the development of
and manufacturing, providing parts and products in a holistic way,
increased business value to enterprises with early adopters using such tools
and their customers. to design everything from aircraft
components to athletic shoes. It takes
With the potential to continuously
the way companies have designed
feed learning data into the software,
for the past 30 to 40 years with CAD
generative design systems will be able
software—taking something in our
to produce more effective designs over
heads and pushing the geometry
time. Some early adopters are already
around on the screen with a mouse—
extending the use case for generative
and flips it. The software works at the
design technology beyond product
service of humans instead of the other
and parts development to designs
way around.
for the components and tools used
to manufacture those parts. Down An engineer seeking to design a frame
the line, manufacturers could utilize for a quadcopter drone, for example,
the same technology to design better might specify needed space for a
factories or supply chains. battery, electronics, and cameras;
the weight of those electronics; the

2 Harvard Business Review Analytic Services Briefing Paper | The Next Wave of Intelligent Design Automation
thrust required for propellers; and the FIGURE 1
forces that will act upon the aircraft.
The software would then resolve
TRADITIONAL COMPONENT DESIGN PROCESS
Characterized by many evaluation iterations
those requirements into a number
of optimized physical forms that the
engineer can review and further refine. HISTORY
“The ability to generate feasible designs Previous Designs
that a CAD engineer would never be Create ISSUES Evaluate
able to generate on his or her own and Design Design Design BUILD
then give a human the ability to choose (CAD) Iterations (CAE)
from them can revolutionize product CONTEXT

engineering,” says Meintjes.

Previous Designs

“We can actually feed our require-

ments into the algorithms and let them Design Performance
go—and over time it will be able to cre- Constraints Requirements
ate more complex and better results,”
says Schäfer at Airbus, which has been SOURCE: CIMDATA
experimenting with generative design
tools for five years. “This is the direc- FIGURE 2
tion in which we are headed.”
GENERATIVE DESIGN PROCESS
Optimization should directly create feasible and buildable designs
An Entirely New Way to Develop
Products: The Early Adopters
“We are only at the beginning of the
HISTORY
frontier of innovation that these tools
Previous Designs
will generate,” Meintjes says. “But
some companies are seeing the fruits Generative Document
Design BUILD
of generative design right now.” Algorithm
(CAD)
Airbus began using generative design CONTEXT
Previous Designs
tools to come up with a new design
for the partition that separates the
passenger compartment from the
Design
galley in the Airbus A320 cabin. The Constraints
partition had to be significantly lighter
than the current versions yet strong Performance
enough to anchor flight attendant jump Requirements
seats, have a cutout to move wide
SOURCE: CIMDATA
items in and out, attach to the airframe
in just four places, and be no more
than an inch thick. It also had to meet
strict parameters for weight, stress, and
displacement in the event of a crash.
generative design could be used to
The ultimate design—dubbed the
conceive an entire aircraft.
“bionic partition”—is an unusual-
looking, latticed structure unlike Hardware and toolmaker Stanley Black
any cabin partition in existence, but & Decker’s Infrastructure Innovation
optimized to be strong and light, team embraced generative design tools
requiring the least material to be built. to come up with a new design for the
Airbus 3-D printed the partition’s 15.4-pound hydraulic crimpers used
100-plus pieces, made of a new high- by workers to fix hanging electrical
strength metal alloy, and assembled the and telephones lines. Linemen had to
component. It fits exactly the same way lift these tools hundreds of times each
as the old one, but saves airlines 3,180 day. “The weight is unwieldy,” says
kilograms of fuel per plane per year. DeSantis. “It creates user fatigue.”
While Airbus started with a single The breakthrough innovation team
part, Schäfer envisions a time when used generative design tools to

Briefing Paper | The Next Wave of Intelligent Design Automation Harvard Business Review Analytic Services 3
“The designer’s role changes from CAD jockey to curator and
empathizer,” says Bastian Schäfer, innovation manager with Airbus.

prototype and test a hydraulic crimper they can ask “Which design best meets
head that was significantly lighter, but the requirements?” They can home
still as strong and durable. Specifically, in on the best design for a particular
they used the tools to redesign the circumstance rather than simply on a
metal attachment that helps make the design that works.
crimp on a metal connector. They set
Increasing agility and efficiency.
parameters around weight, size, and
The generative design approach can
manufacturing cost, and the software
enable manufacturers to look at their
generated about 100 designs, from
design and production processes
which the team chose the option
differently, saving time and money.
providing the best combination of
“We can create products and parts of
weight and manufacturing cost savings.
products that weren’t even imaginable
The final design weighs just over two in the past,” says DeSantis. That may
pounds and is held together by what mean using additive design to produce
DeSantis describes as a lattice that a product that had to be machined or
looks like it’s made of toothpicks. molded in the past, or simply reducing
“We started on this path with the the complexity and increasing the
goal of solving the problem of weight manufacturability of a design. “The
saving,” says DeSantis. “But what we ultimate benefit is speeding up the lead
uncovered is that there were a number time,” DeSantis says. “If it takes weeks
of other benefits with the generative to design an overall product, you can
design approach.” get it to market that much faster.”
Improving the relationship between
The Benefits of Intelligent design and manufacturing. There’s
Design Automation long been an awkward dance that
Indeed, while leading-edge users of happens between designers and
generative design technology begin manufacturers. Designers may come
exploring the new tools to solve up with a product with little concern
specific design conundrums—“light- for how it will be made, and throw
weighting” an existing product, for it over the wall. Manufacturing will
example—they are finding that there throw it back, requesting changes to
a number of significant advantages to improve manufacturability. And so
the approach. it goes until they find some common
ground. With generative design tools,
Reinvigorating creativity and
manufacturing intelligence is built into
customer focus. Generative design
the design process from the start, with
can free up designers and engineers
considerations for how the geometric
to focus on the “what” of their work
variances in a design will impact the
rather than the “how.” Instead of
ability to injection mold, machine, or
spending the majority of their time
3-D print it, for example, built in.
doing drags and drops to iterate a
design, they can concentrate on Bringing other business functions
customer pain points. “The designer’s into the fold. Generative design
role changes from CAD jockey to tools are built for use by designers
curator and empathizer,” says Schäfer. and engineers, but they can also
benefit other areas of the business
Rather than asking, “Does this design
with a vested interest in product
meet the requirements?” says Meintjes,

4 Harvard Business Review Analytic Services Briefing Paper | The Next Wave of Intelligent Design Automation
development. “Engineers will be me,” says DeSantis. Others may think
the ones using the tools day in and the software looks too complicated,
day out,” says DeSantis. “But the or prefer working in the ways they
approach enables you to bring in sales always have.
or marketing or customer service
The key to overcoming anxiety or
sooner in the process to begin having
opposition is taking the time to
conversations about price, product
illustrate how the systems work. “Once
differentiation, or weight.”
you walk them through what it actually
Reimagining manufacturing systems. is, they begin to see that it’s another
In the future, manufacturers may tool they can use to accomplish
extend the functionality of generative their work in a quicker fashion,” says
design tools to help them design or DeSantis. “Generative design is not
redesign their manufacturing tools, going to replace engineers anytime
factories, or supply chains. The soon. It simply allows them to apply
approach can help them come up with their skills in a different way.”
a range of scenarios for how to create
a certain part, where to locate the CHANGING MINDSETS
factory, and how to supply it. Indeed, generative design requires
designers and production engineers to
work and think differently. “It’s a whole
Laying the Groundwork for the new way to design. That’s the primary
Future of Product Development challenge,” Schäfer says. “It’s not
While generative design technology
just learning a new piece of software
has the potential to transform product
or some new features. It’s reframing
design and manufacturing, successful
how you think.” In the past, engineers
introduction and adoption of these
would come up with several ideas,
tools require shifts in mindsets,
adjusting those to meet requirements.
training, processes, and infrastructure
With generative design, they begin with
in the enterprise. The biggest
a plethora of design solutions that all
challenges have to do not with the
meet those requirements, and explore
technology itself, but with the ability
the trade-offs with each. It requires
of humans and human organizations
giving some training and time to wade
to truly adopt an entirely different
in and develop new kinds of intuition
approach, says James White, director
with the tools.
of the additive manufacturing strategy
practice at CIMData. “The way we’ve One option is to develop a generative
been designing products has been design center of excellence to start,
unchanged for 5,000 years. We come as Airbus did, charged with providing
in with the idea of the end thing we generative design services to the
want to produce, we deconstruct it business with the goal of extending the
into components, we figure out how practice to the rest of the enterprise
to make those components, and we over time.
assemble the thing. It’s been the same
approach, from building the Pyramids
to producing iPhones.”
Such a significant shift will require
manufacturers to address and enable
changes on a number of fronts.

ADDRESSING FEAR, UNCERTAINTY,

AND DOUBT
This new class of systems is more
than an incremental improvement on
WITH GENERATIVE DESIGN, ENGINEERS BEGIN
CAD software. The automation and
intelligence they introduce is likely to
WITH A PLETHORA OF DESIGN SOLUTIONS THAT ALL
elicit concerns from some employees.
“The first thought is often that this
MEET THOSE REQUIREMENTS, AND EXPLORE THE
computer software is going to replace TRADE-OFFS WITH EACH.
Briefing Paper | The Next Wave of Intelligent Design Automation Harvard Business Review Analytic Services 5
Some innovative solutions that technically achieve the desired
requirements may be too unusual for customers to accept.

EMBRACING THE CLOUD MASTERING DESIGN SELECTION

Cloud computing provides cost-effective Users may also be overwhelmed by the
access to the computer power required number of choices generative design
to quickly generate multiple solutions to software can produce. The possibilities
complex design problems. are 10 or 100 times more than they’re
used to considering. However, the
OVERCOMING DESIGN BIAS makers of generative design systems
Designers using generative design tools have worked to refine user interfaces so
for the first time may be taken aback that designers can easily scroll through
by what the computer spits out. Unlike the trade-offs embedded in each design
a human designer, the machine isn’t and use a visualization tool to sort and
basing designs on what’s come before. rank them to narrow down choices.
“The designs can look very foreign,”
says DeSantis. “The first time we ran it, PICKING THE BEST PILOTS
the computer spit out 100 options and, Stanley Black & Decker picked a small-
for about 80% of them, we were like, scale, low-volume part to start. “It
‘What the heck is that?’ There’s a little didn’t distract us from what was going
shock and awe.” Some of these new on in other areas of the business,”
forms may be impossible to make with says DeSantis, “and we learned a
traditional manufacturing methods, lot about where we might apply it
and must be built using additive elsewhere.” It was a part that would
manufacturing methods such as 3-D clearly benefit from software with the
printing. A generative design solution computing power to figure out how to
might produce a skateboard that looks decrease its weight without impacting
like a boomerang with three wheels or a performance. Based on the positive
grand piano the resembles an oversized results, DeSantis’ team has begun
box. The designs may yield better educating the rest of the manufacturing
performance; the oblong skateboard may and operations organization about the
ride better than the symmetrical version, benefits of the generative design. The
and there’s no real reason a grand piano company is evaluating how it might use
has to look like a sideways harp on legs the systems to redesign some tooling
other than that’s the iterative way in to make it more cost-effective, user-
which it was first designed. Still, each friendly, and efficient.
may be a bit too unfamiliar to sell. “The
issue is human acceptance,” says White.
“There’s a barrier there when you’re
manufacturing products for consumers.”

6 Harvard Business Review Analytic Services Briefing Paper | The Next Wave of Intelligent Design Automation
THE BIGGEST CHALLENGES HAVE TO DO
NOT WITH THE TECHNOLOGY ITSELF, BUT
WITH THE ABILITY OF HUMANS AND HUMAN
ORGANIZATIONS TO TRULY ADOPT AN
ENTIRELY DIFFERENT APPROACH.
JAMES WHITE, DIRECTOR OF THE ADDITIVE
MANUFACTURING STRATEGY PRACTICE, CIMDATA
 Toward an Intelligent
Design Future
There’s little doubt that, over time,
generative design systems will be
able to tackle increasingly nuanced
conditions and considerations and
become a more important implement
THE MACHINES WILL BECOME BETTER AT UNDERSTANDING in the product development toolkit.
The machines will become better at
THE USERS’ NEEDS AND THE USERS WILL BECOME MORE understanding the users’ needs and
the users will become more adept
ADEPT AT UTILIZING THE TOOLS. at utilizing the tools. The bigger
challenges will be those with a human
face—making the fundamental
changes in product development roles,
relationships, and organizations.
EVOLVING PROCESSES
“If you look at the design process in Now is the best time for manufacturers
most manufacturing companies, it will to begin addressing those issues.
need to change so that the generative Companies that start down this path
design activity gets moved way up now will also have the opportunity to
to where they’re establishing design make sure the tools being developed
concepts and product architectures,” will work for their sector, because
says Meintjes. Using generative design product development use cases will
tools may mean the need for fewer vary wildly by company. “Pick a part or
meetings later on in the process, but process in your industry and dive right
much more interaction among more in,” advises DeSantis. “If you wait for
stakeholders throughout. Generative the technology to get there, you will get
design will benefit from “borderless, passed by.”
dynamic collaboration” among those
charged not only with product design
and engineering, but with pricing,
marketing, sales, and customer service,
says White.

UNDERSTANDING GENERATIVE
DESIGN’S LIMITS
Knowing what these tools can’t do
may be as important as knowing
what they can do in these early days.
The future vision of these systems
and the integration of more artificial
intelligence holds incredible promise,
but there’s the risk of overblown
expectations. It would be difficult to
create an algorithm that enables a
generative design system to consider
a designer’s aesthetic sensitivities, for
example. That’s a really sophisticated
problem to try to deconstruct and
have a computer help with. Right now,
it’s really good at weighing 10 design
variables against each other—and that’s
more than a human brain can hold. But
it’s still early days for this technology.

8 Harvard Business Review Analytic Services Briefing Paper | The Next Wave of Intelligent Design Automation
 hbr.org/hbr-analytic-services

CONTACT US
[email protected] Copyright © 2018 Harvard Business School Publishing.
MC207970518