The Role of Marketplaces for the Transformation from Robotic Process

Automation to Intelligent Process Automation

Thomas Neifer

1,2,3

, Paul Bossauer

1,2,3

, Dennis Lawo

1,2

, Robert Volkening

and Andreas Gadatsch

2,3

Information Systems, University of Siegen, Siegen, Germany

Institute for Digital Consumption, Bonn-Rhein-Sieg University of Applied Sciences, Sankt Augustin, Germany

Management Sciences, Bonn-Rhein-Sieg University of Applied Sciences, Sankt Augustin, Germany

Keywords:

Intelligent Process Automation, Robotic Process Automation, Marketplaces, Adoption Factors.

Abstract:

The corporate landscape is experiencing an increasing change in business models due to digitization. An in-

creasing availability of data along the business processes enhance the opportunities for process automation.

Technologies such as Robotic Process Automation (RPA) are widely used for business process optimization,

but as a side effect an increase in stand-alone solutions and a lack of holistic approaches can be observed. In-

telligent Process Automation (IPA) is said to support more complex processes and enable automated decision-

making, but due to the lack of connectors makes the implementation difﬁcult. RPA marketplaces can be a

bridging technology to help companies implement Intelligent Process Automation. This paper explores the

drivers and challenges for the adoption of RPA marketplaces to realize IPA. For this purpose, we conducted

ten expert interviews with decision makers and IT staff from the process automation sector.

1 INTRODUCTION

Digitalization is bringing profound changes to organi-

zations and business models (Bouwman et al., 2018;

Neifer et al., 2021b). Due to the increasing availabil-

ity and utility of data along business processes, their

automation is also undergoing a major transformation

(Chakraborti et al., 2020). Traditional approaches to

process automation focus primarily on processes that

occur frequently and repeatedly in the same ﬂow pat-

tern. However, if repetitive processes do not occur

frequently enough, their automation is usually con-

sidered too cost-intensive, especially if integrations

of different systems and their data have to take place

(Van der Aalst et al., 2018). In this case, Robotic

Process Automation (RPA) technology is often dis-

cussed in companies (Neifer et al., 2021a). For exam-

ple, Van der Aalst et al. (2018) highlight the question

”What should be automated and what should be done

by humans?” when implementing RPA technologies.

There is no uniform deﬁnition of RPA in the liter-

ature. In its original form, RPA refers to a technology

for automating manual activities in processes that are

structured, rule-based and repetitive by digital soft-

ware robots, simply called bots (Kleehaupt-Roither

and Unger, 2018). RPA offers many beneﬁts, such

as it allows accuracy, reliability, uniformity, and con-

sistency by processing routine tasks in the exact same

way without interruption, with a reduced susceptibil-

ity to error. This is ensured through regulatory com-

pliance as well as work history revision control. Fur-

thermore, productivity can be increased in two ways:

First, through faster process cycle times due to au-

tomation, and second, by shifting the focus of em-

ployees on important and value-added work. Further-

more, because it is a non-invasive technology with

low technical barriers, the burden on IT and the barri-

ers to adoption are reduced (Madakam et al., 2019).

Besides the advantages, RPA also introduces new

challenges. Among other things, RPA focuses on

highly structured routine tasks. Further, the low im-

plementation hurdles ensure the possibility of the use

of this technology by the business departments inde-

pendently of the IT department, which may promote

shadow IT structures as well as IT security violations

(Willcocks et al., 2015; Gadatsch and Mangiapane,

2017; Matthews and Greenspan, 2020).

Due to the primary reduction to highly structured

tasks, the question about the differentiation of au-

tomation and human activities is answered by this

technical limitation. Wherever unstructured data,

complex tasks and decisions occur, a human must in-

tervene. This is particularly problematic, since only

about 30% of the enterprise data is structured (e.g.,

customer IDs, addresses, account details, etc.) the re-

maining 70% represents unstructured data (e.g., PDF

documents, emails, images, etc.) (Taulli, 2020b).

To overcome this barrier, there is an increasing re-

Neifer, T., Bossauer, P., Lawo, D., Volkening, R. and Gadatsch, A.

The Role of Marketplaces for the Transformation from Robotic Process Automation to Intelligent Process Automation.

DOI: 10.5220/0011146200003280

In Proceedings of the 19th International Conference on Smart Business Technologies (ICSBT 2022), pages 15-25

ISBN: 978-989-758-587-6; ISSN: 2184-772X

liance on the use of Artiﬁcial Intelligence (AI) meth-

ods. This manifests itself in the term Intelligent

Process Automation (IPA, also Intelligent Automa-

tion or Cognitive Automation) and marks a change

from rule-based to intelligent process automation

(Chakraborti et al., 2020). This should result in self-

learning software robots that are robust and ﬂexible

in the face of complex process dynamics and have

autonomous decision-making capabilities (Czarnecki

and Auth, 2018).

However, this transformation is still at an early

stage, which is why RPA vendors are reaching their

limits with regard to the further development of their

RPA tools into an all-encompassing, intelligent holis-

tic solution, as these require a variety of technolo-

gies and functions (Taulli, 2020a) as well as an

ecosystemic approach (Neifer et al., 2021a). Further-

more, for an efﬁcient training of machine learning

(ML) methods regarding the decision making of these

bots, this requires an integration of (un)structured

data from different sources in order to beneﬁt from

synergy effects (Dong and Rekatsinas, 2018). Cur-

rently, it does not seem possible for an RPA vendor

to provide all these challenges from a single source

urkoglu, 2020). This problem gave rise to RPA

platforms, which in their function as a marketplace

offer modular components from RPA providers, third-

party vendors or individual developers (Mullakara,

2020).

For these marketplaces, the problem of user ac-

ceptance and trust in the provider and platform

emerges, which is central to the platform econ-

omy (Clement et al., 2019). Scientiﬁc research on

the transformation to intelligent process automation,

however, focuses here mainly on the differentiation

of RPA/IPA development stages (Lacity and Will-

cocks, 2018; Stoudt-Hansen et al., 2019; Ferreira

et al., 2020) and providers (Le Clair et al., 2019),

the strategic framework (Mohanty and Vyas, 2018;

Scheer, 2020), potentials and risks (Smeets et al.,

2019; Agostinelli et al., 2019), and the explanation of

IPA technology and derivation of use cases (Berruti

et al., 2017; Lacity and Willcocks, 2018; Burgess,

2017; Smeets et al., 2019; Taulli, 2020b; Ferreira

et al., 2020; Moiseeva et al., 2020; Engel et al., 2021).

Therefore, we see a research gap in the exploratory re-

search of RPA marketplaces as a bridging technology

towards IPA.

This work aims to analyze drivers and challenges

for the adoption of RPA marketplaces to realize IPA.

For this purpose, we conducted ten expert interviews

with decision-makers and IT staff from RPA and pro-

cess automation departments to address both a tech-

nical and strategic perspective. The identiﬁed factors

contribute to the scientiﬁc discourse about RPA mar-

ketplaces and can be used as a basis for companies

to decide whether to use RPA marketplaces. In the

following, we delineate the terms RPA and IPA, pro-

vide insight into the state of the research, describe

our methodological approach and then present and

discuss our ﬁndings on the drivers and challenges of

adopting RPA marketplaces to realize IPA.

2 ROBOTIC AND INTELLIGENT

PROCESS AUTOMATION

In literature, Robotic Process Automation is pre-

dominantly deﬁned by its functionality (Smeets et al.,

2019). While Van der Aalst et al. (2018) deﬁne

RPA as a collective term for tools that operate com-

puter systems via user interfaces in the way a hu-

man would, Allweyer (2016) distinguishes RPA from

physical machines as a type of personal assistant to

support or replace employees. The basic idea of RPA

is the exact imitation of manual user input by a soft-

ware bot. In doing so, the underlying system does

not recognize that it is being operated by software in-

stead of a human (Smeets et al., 2019). Furthermore,

the implementation and realization of RPA does not

require any programming skills, there is only a con-

ﬁguration of rules or recordings of manual user inter-

actions (Willcocks et al., 2015; Czarnecki and Auth,

2018). This promotes a detachment from the IT de-

partment (Czarnecki and Auth, 2018) and a tendency

towards shadow IT (Gadatsch and Mangiapane, 2017;

Matthews and Greenspan, 2020). Overall, RPA is

focused on rule-based and stable processes with re-

peated execution, a low need for change, and highly

structured input data (Allweyer, 2016).

Further, RPA follows an outside-in approach,

whereby no changes are made to existing application

systems as part of the automation process (Van der

Aalst et al., 2018). While classical automation ap-

proaches access the system via an application pro-

gramming interface (API), RPA usually uses the

graphical user interface (GUI) (Smeets et al., 2019).

This already results in the initial problem that RPA

does not directly access the integration of multiple

systems via the GUI (Taulli, 2020a). This would re-

quire major adjustments to the IT infrastructure and

the existing processes (Czarnecki and Auth, 2018).

Intelligent Process Automation on the other

hand describes the further development of RPA into a

cognitive, self-learning overall solution (Smeets et al.,

2019). This allows processes to be fully automated

and enable human decision-making behavior as well

as unstructured data (e.g. text, speech) and complex

ICSBT 2022 - 19th International Conference on Smart Business Technologies

Figure 1: Account Opening Process in Transition from RPA to IPA.

issues to be mapped in the form of forecasts and anal-

yses (Langmann and Turi, 2020). Accordingly, IPA is

an RPA-based ecosystem of AI and other digitization

technologies that take automation to the next level

through the interaction of technologies (Berruti et al.,

2017; Mohanty and Vyas, 2018; Zhang, 2019). Key

technologies to enable IPA include Smart Workﬂow

(Smeets et al., 2019), Machine Learning (ML) (Al-

paydin, 2020; Smeets et al., 2019; Mohanty and Vyas,

2018), Natural Language Processing (NLP) (Mohanty

and Vyas, 2018; Smeets et al., 2019; Sarkar, 2019),

Natural Language Generation (NLG) (Celikyilmaz

et al., 2018), Optical Character Recognition (OCR)

(Langmann and Turi, 2020), Process Mining (Scheer,

2020), and Cognitive Assistants (Allweyer, 2016).

The relevance of the transformation from RPA

to IPA can be seen in Figure 1, which shows an ex-

ample account opening process. Without RPA, not

only the process steps with a decision-making func-

tion (e.g., determining credit limit and additional of-

fers) but also repetitive steps (e.g., creating an ac-

count, creating documents) are performed manually.

The staff must perform routine activities with a pre-

dominantly high structural component.

RPA can be used to automate the process steps

with a high degree of structure and repetitive char-

acter. Process steps with a decision-making func-

tion (e.g., determine credit limit) or unstructured data

bases (e.g., validate documents), on the other hand,

must still be performed manually.

Unstructured or semistructured data, e.g., in the

form of e-mails, is a major problem in customer data

acquisition. Since an RPA bot requires structured in-

put, Natural Language Processing (NLP) in conjunc-

tion with a learning algorithm must be used to create

an intelligent model that can also process this data in a

meaningful way. In addition, clustering methods can

be used to make a decision regarding the credit limit

based on features within the customer data. Docu-

ment validation could also be addressed by an NLP

model trained on these documents. Regarding the de-

cision on additional offers, data mining methods as

well as recommendation systems can be used, which

derive customer preferences based on the customer

data and provide personalized recommendations for

additional offers. By implementing such an IPA so-

lution in conjunction with an integrated data architec-

ture (Neifer et al., 2021a), it is possible to digitally

map the account opening process in its entirety. Hu-

man intervention would only be necessary in excep-

tional cases.

Holistic IPA solutions in the form of a one tool do

not yet exist. Thus, the dilemma prevails that com-

plementary technologies of different suppliers have

to be connected with each other. Thereby, the selec-

tion, as well as speciﬁc conﬁguration represent a great

challenge to the users (Taulli, 2020c). RPA mar-

ketplaces could enable more holistic IPA solutions,

which offer pre-built bots, plugins, connectors, or

even individual components and capabilities of soft-

The Role of Marketplaces for the Transformation from Robotic Process Automation to Intelligent Process Automation

ware robots (Taulli, 2020a) developed by platform

providers, third-party vendors, or even individual de-

velopers (Mullakara, 2020). In this context, Schep-

pler and Weber (2020) highlight that there is a need

for research in evaluating the ability of RPA market-

places to integrate solutions through pre-built compo-

nents and simpliﬁed development of bots. This goes

hand in hand with the evaluation need for acceptance

and trust on the part of users in both the platform and

the speciﬁc providers (Clement et al., 2019).

3 RELATED WORK

The scientiﬁc literature on the shift from rule-based

to intelligent automation as well as on RPA market-

places is still relatively scarce, but recently experi-

ences an increasing interest. Berruti et al. (2017) con-

sider IPA under a platform idea, consisting of the ﬁve

core technologies RPA, Smart Workﬂow, Machine

Learning, NLG, as well as Coginitive Assistants,

which aims at a fundamental redesign of processes.

Leading platforms in this space include UiPath, Au-

tomation Anywhere, and Blue Prism (Taulli, 2020a).

According to Mohanty and Vyas (2018), compa-

nies are increasingly recognizing the potential ben-

eﬁts of IPA and experimenting with intelligent au-

tomation in various functional areas. According to

them, RPA represents the basic building block on the

path to IPA, where considering ML and RPA sepa-

rately would be a step backward. While some soft-

ware vendors and startups are making progress in de-

veloping applications with image and speech recogni-

tion as well as information extraction, the focus here

is on complementing RPA (Burgess, 2017). Burgess

(2017) argues that combining RPA and ML can en-

able automation of end-to-end processes, and both

technologies beneﬁt from each other: If the data is

unstructured or semistructured, or if judgment is re-

quired, ML can support the RPA bots. RPA, on the

other hand, can map the extraction and assembly of

data from various sources, thereby acting as a data

provider for ML. Smeets et al. (2019) found out that

companies see more potential in combining RPA with

other technologies (such as process mining and work-

ﬂow management systems) than in moving RPA itself

toward IPA. Stoudt-Hansen et al. (2019) observed a

shift from RPA to a holistic approach of automation,

which they refer to as hyperautomation. However, the

dilemma here is that all the necessary technologies are

not easy to link, because they often come from differ-

ent vendors.

Le Clair et al. (2019) therefore see the decisive

success factor in the integration of artiﬁcial intelli-

gence (AI). For this reason, most RPA providers are

trying to develop their tools in-house or with external

partners. On some marketplaces, such as Automation

Anywhere, companies can source suitable solutions

from third-party vendors for their automation prob-

lem. Credibility and trust in the vendors therefore

play a key role. Many vendors make bold claims re-

garding the capabilities of their respective products,

which is referred to as ”RPA washing” (Lacity and

Willcocks, 2018). Further, in terms of identifying

suitable process candidates and providing the neces-

sary scale, vendors must ensure that their automation

solutions are user-friendly and easy to implement and

deploy. Thus, supporting analytics to ﬁnd the right

process candidates, scalability of the bots, and cen-

tralized coordination and openness of the IPA plat-

form have become more important (Le Clair et al.,

2019).

Taulli (2020c) recommends the use of RPA mar-

ketplaces due to a non-existent end-to-end solution, as

they can help shorten development times on the user

side and provide a new revenue stream on the vendor

side. Automation should be seen from a holistic per-

spective and be realized through end-to-end automa-

tion platforms or programs. These statements are con-

sistent with those of Le Clair et al. (2019) and Burgess

(2017). Mullakara (2020) agree that RPA market-

places have great potential as part of the evolution

of RPA toward hyperautomation, because these mar-

ketplaces enable the integration of a variety of new

technologies. However, this integration should be as

simple as possible, e.g., via drag-and-drop interfaces.

They expect higher participation in the future as well

as a wide range of prefabricated components.

4 EXPERT INTERVIEWS

To explore the drivers and challenges for RPA mar-

ketplace adoption and factors to support the transfor-

mation from RPA to IPA, we conducted ten semi-

structured expert interviews (see Table 1) within an

interpretive research stance (Collis and Hussey, 2013;

Bell et al., 2018; Themistocleous and Morabito, 2012;

Depietro et al., 1990). We chose semi-structured in-

terviews because of the exploratory nature of the re-

search question. While there is already some litera-

ture on RPA and IPA, we wanted to keep the inter-

views very open to give the experts more room to for-

mulate requirements and opinions that have not been

captured by the literature so far. In fact, while con-

ducting the interviews, new aspects kept coming up

that were incorporated in the follow-up interviews,

which is a great advantage of qualitative surveys in

ICSBT 2022 - 19th International Conference on Smart Business Technologies

Table 1: Overview of participants.

ID Sector Job Department

E01 Telecommunications Head of Department Service IT Conception & Automation Control

E02 Insurance Head of Department RPA Competence Center

E03 Insurance RPA Expert Organization & Coordination of Automation

E04 Retail Team Leader Inhouse Consulting, Process Digitization & Automation

E05 Conglomerate Product Owner Process Automation & API Strategy

E06 Consulting Team Leader Data Transparency & Efﬁciency

E07 Consulting RPA Expert Process Automation Consulting

E08 IT Head of Department Business Intelligence & Cognitive Automation

E09 IT Team Leader Process Automation

E10 IT Head of Department Business Intelligence

the form of interviews.

The experts were selected primarily based on their

function in the company. The interviewees were RPA

or process automation managers and IT staff from

different large companies in Germany in order to in-

clude both a strategic and technical perspective. Fur-

thermore, the aim was to ensure an across-industry

overview. Therefore, the management consultancies

provided an additional perspective through their ex-

perience with the requirements of companies for RPA

marketplaces resulting from their consultancy ser-

vices. The interviews lasted on average 49 minutes

and followed a semi-structured guideline with the fol-

lowing topics:

• Practices and Experiences with RPA marketplaces

to derive insights about drivers and challenges as

well as to develop a common understanding of

RPA, IPA and RPA marketplaces.

• Acceptance and usage of RPA marketplaces with

respect to the development of IPA.

• Perception of the future impact of RPA market-

places on IPA.

The interviews were transcribed with MAXQDA and

analyzed using the inductive approach of thematic

analysis according to Braun and Clarke (2006). Based

on the previous experiences with RPA marketplaces,

both the drivers and the challenges with regard to the

use as well as the feature supporting the transforma-

tion from RPA to IPA were focused on. Two authors

undertook the coding of the interview material inde-

pendently and then combined the resulting code sys-

tem (Berends and Johnston, 2005).

5 RESULTS

The following chapters describe the results of the in-

terviews conducted, which are summarized as drivers

and challenges for the adoption of RPA marketplaces

to realize IPA in Figure 2. The results were differen-

tiated into organizational and technical factors.

5.1 Organizational Drivers

Nine out of ten experts stated that RPA marketplaces

can contribute to a democratization of development

of RPA or IPA solutions. In principle, RPA market-

places can simplify the development of RPA skills and

lower the barrier to entry. RPA marketplaces can also

provide insights into the topic of IPA, which can be

used as orientation. For example, according to E06,

“standard modules or functionalities for data extrac-

tion and data classiﬁcation, in particular OCR func-

tionalities, are very interesting”.

”[...] smaller and medium-sized companies

can beneﬁt from RPA marketplaces, especially

if they ﬁnd it more difﬁcult to get started with

RPA due to a lack of resources.” – [E05]

Further potential strategic beneﬁts are identiﬁed in

a cost reduction (eight experts) and time savings

(six experts). The more ready-to-use components

are reused to develop the bot solutions, the more

the development costs decrease. According to E02,

”through standardized development, the bots can be

scaled with low costs”. However, E04 and E05 refer

to the high training effort of AI models. According

to E04, the training effort and thus the resulting costs

could be ”reduced by ready-to-use AI models” and

”IPA solutions could be implemented faster”. The

reuse of ready-to-use automation modules via plug &

play can also signiﬁcantly reduce development time.

Thus, a signiﬁcant increase in efﬁciency in bot devel-

The Role of Marketplaces for the Transformation from Robotic Process Automation to Intelligent Process Automation

opment is achieved through reuse and avoidance of

multiple developments.

”If I apply a meta-bot in 25 bots, that’s just

great, [...] maybe an hour of development time

[...] and you can plug & play that meta-bot

back in and you saved a lot of time.” – [E05]

Five experts ﬁnd RPA marketplaces useful for idea

generation. For example, E03 is interested in basic

frameworks of objects that provide ideas for develop-

ing solutions for company-speciﬁc systems and can

be adapted accordingly. According to E05, business

departments often ﬁnd it difﬁcult to identify suitable

use cases, which is why ”they can use an RPA mar-

ketplace to generate ideas for suitable process candi-

dates”.

5.2 Technical Drivers

According to nine out of ten experts, RPA market-

places can be useful for the development of IPA so-

lutions through the provision of ready-to-use intelli-

gent automation solutions. For example, E02 states

that RPA marketplaces could ”offer additional fea-

tures, such as OCR, that are not included in standard

RPA solutions”. Contrary to the promises of some

RPA vendors, according to E03, E06 and E10, RPA

software products have so far been only partially suit-

able for the development of intelligent bots. There-

fore, according to E06, extending the capabilities of

RPA with cognitive components is difﬁcult, and they

”use a dedicated OCR system where the process al-

lows it. However, this does not apply well in practice

because we would then have to switch to a completely

different system in a process”. Accordingly, standard-

ized intelligent components provided via RPA mar-

ketplaces are perceived as an improvement.

”If there were OCR functionalities that could

be integrated into workﬂow modeling under

UiPath, for example, that would be much more

attractive.” -– [E06]

Six experts see an opportunity for quality improve-

ments in the development of IPA solutions on a tech-

nical level. The reuse of ready-to-use components

makes it easier to identify error sources and thus

reduce them during bot development. Further, ac-

cording to E02 and E07, the participation of various

providers in RPA marketplaces within the framework

of direct network effects ensures an increase in the

quality of solutions and of IPA in particular, since ”a

broad positioning increases the pressure on the in-

dividual providers and thus the quality of the solu-

tions”.

5.3 Organizational Challenges

All experts agree that integration into the IT in-

frastructure has a signiﬁcant inﬂuence on the use-

fulness of RPA marketplaces. In this context, E01,

E03, E04 and E08 criticize the fact that the range of

marketplaces is currently ”too focused on standard

software”, but the “system landscape of companies is

characterized by a large number of in-house devel-

oped systems” and ”[...] the modules are not suitable

for automation on in-house developed systems”. Ac-

cording to the experts, this is also the main reason for

the low use of external RPA marketplaces and the de-

cision to build an internal RPA marketplace. E03 sees

the capabilities of AI as crucial for successful integra-

tion, explicitly to what extent ”intelligent components

are customizable for speciﬁc companies”. E04 also

sees the integration of off-the-shelf AI solutions, pro-

cess mining, and chatbots from RPA marketplaces as

promising, but it depends on how ”standardized the

use cases need to be”. It appears difﬁcult to use in-

telligent ready-to-use modules in a company-speciﬁc

way, as the modules have to be trained on a use-case-

speciﬁc basis and are correspondingly costly.

”That’s where I think you run into limits very

quickly in these marketplaces, because it’s re-

ally difﬁcult to offer ready-to-use models, or

if they’re trained models, they’re utopian in

price.” -– [E05]

Additionally, eight out of ten experts classify data

protection and security as relevant factors inﬂuenc-

ing the perceived beneﬁts of RPA marketplaces. E04

and E05 accordingly describe that the companies de-

cided against using external RPA marketplaces due to

”internal regulations and security checks and the as-

sociated costs”. According to the experts, security

must be guaranteed by the operators of the market-

place and could be conﬁrmed, for example, by com-

pliance of industry-standard IT security certiﬁcations

for the offered services.

”The listings on the marketplace should have

gone through security checks and be certiﬁed

by the operator, so you can be sure there’s no

malicious code in there.” -– [E04]

Four experts highlight the guarantee of support as a

further inﬂuencing factor for the acceptance and on-

going use of the marketplace. The providers are seen

as having a responsibility to adapt the modules and

keep them up to date.

”[...] is there support at all the same as with

ofﬁcial packages from UiPath or not, because

that would be a problem if we unleash a lot

ICSBT 2022 - 19th International Conference on Smart Business Technologies

of robots on different activities and then those

activities don’t work anymore.” -– [E02]

Proﬁtability is another factor that must be given ac-

cording to four experts. E01 expects in this regard that

the ”offer on the marketplaces must be more favor-

able than in-house development and its operation”.

Currently, this does not yet seem to be the case, since

adaptation and testing of the components involves a

very high level of effort compared with in-house de-

velopment.

E02 and E09 further emphasize that the num-

ber of users has a direct inﬂuence on the usefulness

of RPA marketplaces. With an increasing number

of users, ”technical and innovation limits could be

lifted”.

Another aspect represents for four experts is the

risk of dependency. Thus, the increased purchase

of prefabricated components brings knowledge losses

with itself. However, the maintenance effort remains,

so there is more dependence on module suppliers, es-

pecially for intelligent modules.

5.4 Technical Challenges

All ten experts interviewed express expectations for

the usability of the RPA marketplaces in order to pos-

itively inﬂuence the perceived ease of use. For E04, a

”very good documentation of the solutions by means

of process ﬂow charts and videos” is important in or-

der to enable a quick and simple basis for decision-

making for the specialist department and the devel-

opers. Thus, it is basically possible to transfer IPA

development to the business departments if the solu-

tions from the RPA marketplaces are very well docu-

mented and easy to adapt. For E09, too, ”good and

detailed descriptions of the modules are important, as

well as easy downloading and integration of the solu-

tions into the internally used RPA tool”. For E05, the

most important factors in achieving a high user expe-

rience in the company’s internal RPA marketplace are

”transparency and create easy searchability and list

contacts”. Thus, ﬁltering by bot solutions should be

possible ”technology-based by system and technical

by functional area”. Furthermore, a detailed text de-

scription and ﬂowcharts should explain the use case

in more detail and the corresponding process experts

should be speciﬁed as contact persons. Furthermore,

according to E08, a ”possibility to categorize the so-

lutions according to industries on the external RPA

marketplaces” has been missing so far.

Five experts also agree that the quality and matu-

rity of solutions on RPA marketplaces can still be im-

proved in order to increase their usefulness. Although

the range of products on offer in the RPA market-

places is extensive, some solutions are of insufﬁcient

quality, which means that the maturity level of the so-

lutions needs to be increased. For example, the cur-

rently offered services are not attractive enough and

plug & play modules are not yet mature.

”I think the construct behind the marketplaces

and the reputation is very good. I see a per-

spective there, but we are still quite far away

from usable modules or bots that can be down-

loaded and used plug & play.” -– [E03]

5.5 Adoption of RPA Marketplaces to

Realize IPA

Only three of the ten companies interviewed currently

use an external RPA marketplace to develop IPA solu-

tions. According to E02, this is used to source ”com-

ponents for data recognition and preparation, smaller

loops, automatism or algorithms for speciﬁc process-

ing” which can also interact with older systems. One

company plans to expand further, as it sees RPA mar-

ketplaces as very promising in terms of IPA. In Fig. 2

we have summarized the most important drivers and

challenges from the interviews and sorted them ac-

cording to the assigned relevance in descending order

in the associated category. According to the experts,

these factors play a signiﬁcant role in the adoption of

RPA marketplaces.

Six experts are open to using an external RPA mar-

ketplace to develop IPA solutions in the future. At the

same time, the establishment of an internal market-

place for the development of classic RPA solutions

is being pursued. E04 sees great potential in build-

ing modules that can be ﬂexibly incorporated into

bots and states that especially ”standard AI solutions,

OCR capabilities, process mining, chatbots and pre-

trained algorithms are of interest”. Thus, a solution is

developed once for a department and ”made available

to the departments across countries in each case”.

For E06, in the context of IPA, standardized building

modules for decision making, data recognition, and

extraction are of interest, but only to gain insight. In

the long term, the goal is to develop or enhance the

modules themselves. E10 also sees the future use of

RPA marketplaces as useful, especially with regard to

the development of IPA.

”The world is kind of crying out to be able to

buy in these kinds of solutions and reuse them,

and that’s [...] where we’ve put the priority

now.” – [E05]

The Role of Marketplaces for the Transformation from Robotic Process Automation to Intelligent Process Automation

Figure 2: Summary of Drivers and Challenges for the Adoption of RPA Marketplaces.

6 DISCUSSION

6.1 The Role of RPA Marketplaces as

Bridging Technology

According to the experts, RPA marketplaces can con-

tribute to the transformation from RPA to IPA in the

future. This will be made possible in particular by the

provision of ready-to-use intelligent automation so-

lutions, as RPA marketplaces can provide additional

functionalities that are not included in an RPA stan-

dard solution and there will probably also be no holis-

tic approaches (Taulli, 2020a). In addition, RPA mar-

ketplaces can in principle lead to a democratization

of development by enabling the use of AI compo-

nents. Developing these components in-house is seen

as complex, which is why a broad range of know-

how is required for this purpose (Burgess, 2017). By

making them available via a marketplace, the barrier

to entry into the topic of IPA can thus also be low-

ered. Given that IPA solutions should also be imple-

mentable by users without an IT background (Le Clair

et al., 2019; Ito et al., 2020), the aspect of lowering the

barrier to entry seems particularly relevant. However,

the application of IPA is associated with a high level

of required know-how (Scheer, 2020; Ito et al., 2020).

The experts point out that RPA marketplaces address

this dilemma as a bridge technology and that IPA de-

velopments could come from the specialist depart-

ments. In this regard, the business departments can

beneﬁt from the attributed ability to generate ideas

via RPA marketplaces to derive meaningful use cases.

Furthermore, companies with a lack of touchpoints to

RPA can ﬁnd an easier entry into both RPA and IPA

via the use of external marketplaces (Mohanty and

Vyas, 2018).

The experts believe that competition between

providers on a marketplace can improve the quality of

IPA solutions due to the high degree of transparency

and comparability of different solutions. In addition,

RPA marketplaces offer potential for cost reduction

and time savings in the development, which can be

achieved through the possibility of using existing au-

tomation modules by searching for the best ﬁtting au-

tomation modules on the marketplace.

6.2 Drivers and Challenges of RPA

Adoption

Nevertheless, there is a discrepancy between the pos-

itive attitude towards and the actual use of RPA mar-

ketplaces. While the use of external marketplaces is

essentially limited to the implementation of intelli-

gent building modules for data recognition and prepa-

ration, the (planned) development and use of internal

marketplaces is focused on the use of classic RPA so-

lutions. This is consistent with Taulli’s view that AI

is currently limited to form interpretation and intelli-

gent data transfer (Taulli, 2020b). Furthermore, it can

be assumed that the development of an internal RPA

marketplace does not appear lucrative for small and

medium-sized enterprises due to a lack of application

ICSBT 2022 - 19th International Conference on Smart Business Technologies

scenarios for automation as well as a lack of ﬁnancial

and human resources.

The low level of actual use is explained by the ex-

perts as a result of the existing challenges in regard

to RPA marketplaces. The integration of the solutions

into the IT infrastructure is a key factor in the deci-

sion to adopt RPA marketplaces. This results from

the extent to which a company-speciﬁc adaptation of

the modules can take place and how high this adapta-

tion effort turns out to be. Here, a dissent between the

experts’ opinion and the literature can be observed.

Mullakara (2020), for example, refers to the simple

conﬁguration of RPA components, while the expert

whose company already uses an external RPA mar-

ketplace point out a high development effort. This is

consistent with the literature on integration platforms,

according to which data model matching is an essen-

tial problem for a holistic intelligent process automa-

tion approach (Neifer et al., 2021a). In addition, the

integration of solutions is inﬂuenced by the quality

and maturity of the solutions. These factors are rated

as insufﬁcient in RPA marketplaces by the experts.

Other factors that represent a hurdle to adoption

are data privacy and data security of the components

offered on the marketplaces and the guarantee of sup-

port. The experts have judged the guarantee of ade-

quate support to be questionable, particularly in the

case of services from third-party providers. In gen-

eral, cost-effectiveness must also be ensured com-

pared to in-house development, which is associated

with additional costs due to the challenges mentioned.

In addition, the risk of dependence on the module

manufacturers also plays a role for the providers, es-

pecially when it comes to intelligent components,

which are difﬁcult to understand for inexperienced

departments in particular. There is a fear that pro-

cess failures could occur due to a lack of know-how

about the functionality of the purchased components

and poor support. To address the required data pro-

tection and data security aspects, RPA marketplaces

should continue to certify and verify the offering to

further promote acceptance among companies and re-

duce the fear of process failures by ensuring appropri-

ate support.

The experts would like RPA marketplaces to be

intuitive to use and clearly presented. This should be

supported by suitable text and video descriptions of

the components as well as categorization and ﬁltering

options. Furthermore, the IPA components should be

easy to integrate into existing RPA tools. A broad po-

sitioning of the RPA marketplaces and their presenta-

tion through descriptions of the solutions and ﬁltering

by applications, processes as well as categories addi-

tionally support and simplify the generation of ideas.

According to one expert, this contributes to the fact

that with good documentation and simple adaptabil-

ity, IPA development could also be transferred to the

specialist departments. A possibility of categorization

according to the speciﬁc application scenarios can ad-

ditionally support the generation of ideas in the spe-

cialist departments. Furthermore, the ’Cognitive Au-

tomation Use Case Assessment Model’ developed by

Engel et al. (2021), for example, offers the possibil-

ity to ”make more informed decisions about selecting

use cases for cognitive automation or planning their

implementation.”

Overall, it can be stated that the experts perceive

RPA marketplaces as a promising solution for the

transformation from RPA to IPA. This is also reﬂected

in the ﬁnding that three companies were open to a fu-

ture use of external RPA marketplaces for the devel-

opment of IPA solutions and one company wants to

increase future use strongly.

6.3 Limitations

With regard to the limitations, no cross-industry state-

ment can be made due to the sample size of ten ex-

perts, as different industries may also have additional

requirements for such RPA marketplaces. The experts

were also selected by contacting them via existing

networks, which can certainly have an inﬂuence on

the ﬁndings obtained. Nevertheless, care was taken to

ensure high quality in the selection of the experts and

their suitability, as the selected companies and experts

deal intensively with the topics of RPA and IPA.

As far as the transferability of the results is con-

cerned, there will be a wide variety of requirements

in the corporate landscape with regard to compliance

and IT security. In SMEs, other requirements might

arise, especially with regard to the development of

IPA solutions against the background of scarce ﬁnan-

cial and human resources and the implementation of

internal RPA marketplaces. In areas with sensitive

information, in-house developments are likely to re-

main dominant. However, it was also not the goal

of this work to develop generally applicable recom-

mendations for action for the adoption of RPA mar-

ketplaces. The purpose of this work was to identify

drivers and challenges in order to better understand

the adoption of RPA to realize IPA. Nonetheless, an

overlap of the ﬁndings with the existing literature can

be observed, suggesting that the ﬁndings are not lim-

ited only to the sample and that implications for SMEs

have also been incorporated through the participating

management consultancies. For this reason, it would

be useful to validate the drivers and challenges iden-

tiﬁed in this paper for a follow-up study in a larger-

The Role of Marketplaces for the Transformation from Robotic Process Automation to Intelligent Process Automation

scale quantitative study.

7 CONCLUSION

This paper has dealt with the exploration of drivers

and challenges for and the investigation of the po-

tential of RPA marketplaces to support the transfor-

mation from rule-based to intelligent process automa-

tion. It is shown that RPA marketplaces can drive the

transformation of RPA and IPA primarily by provid-

ing intelligent modules, since it can be assumed that

there will be no holistic IPA standard solution. Fur-

thermore, the development of IPA solutions can be de-

mocratized by lowering the barriers to entry through

a simpliﬁed and user-friendly development environ-

ment. On the business side, this is also supported

by the fact that RPA marketplaces can contribute to

the generation of ideas, since, according to the ex-

perts, many small and medium-sized companies in

particular lack suitable ideas for use cases. In addi-

tion, RPA marketplaces can contribute to cost reduc-

tion, time savings and increased quality of solutions

through their platform structure and associated scal-

ing effects.

However, there are also some challenges, which

are mainly determined by the integration of the com-

ponents into the existing IT infrastructure of the

companies, the user-friendliness of the platform and

its guarantee of data security and data protection.

This results in an area of conﬂict between compo-

nents that are as individual as possible and there-

fore not standardized, the simple development and

use of these components, and corresponding cost-

effectiveness and security for the companies. In their

statements, the experts refer to a number of design

suggestions to meet some of the challenges. For ex-

ample, detailed documentation consisting of text and

video descriptions as well as adequate categoriza-

tion and ﬁltering options for the components on offer

should be provided for simple and user-friendly de-

velopment. This would also positively inﬂuence the

aspect of democratization of development for com-

panies by reducing the complexity of implementing

RPA and IPA. Furthermore, RPA marketplaces should

ensure good support with regard to integration into the

existing IT landscape of companies, as well as meet

the demands for data protection and data security via

trustworthy certiﬁcates and veriﬁcation of their offer-

ings.

The drivers and challenges identiﬁed can thus

serve as a basis for decision-making in the adoption

of RPA marketplaces by companies on the one hand,

and as a guide for RPA marketplace operators in the

design of their platforms on the other. Nevertheless,

our results are limited by the sample size consisting of

ten experts and the business composition. Therefore,

future research should on the one hand deal with the

quantitative validation of the identiﬁed factors and on

the other hand investigate them for speciﬁc sectors as

well as in small and medium-sized enterprises.

ACKNOWLEDGEMENTS

This research was funded by Federal Ministry

for Economic Affairs and Energy; grant number

01MT20003D. The authors declare no conﬂict of in-

terest. The authors would like to thank the reviewers

whose comments and suggestions helped to improve

this work.

REFERENCES

Agostinelli, S., Marrella, A., and Mecella, M. (2019). Re-

search challenges for intelligent robotic process au-

tomation. In International Conference on Business

Process Management, pages 12–18. Springer.

Allweyer, T. (2016). Robotic process automation–

neue perspektiven f

ur die prozessautoma-

tisierung. Kaiserslautern: Hochschule

Kaiserslautern. https://www. kurze-prozesse.

de/blog/wpcontent/uploads/2016/11/Neue-

Perspektiven-durch-Robotic-Process-Automation.

pdf. Zugegriffen am, 3:2019.

Alpaydin, E. (2020). Introduction to machine learning.

MIT press.

Bell, E., Bryman, A., and Harley, B. (2018). Business re-

search methods. Oxford university press.

Berends, L. and Johnston, J. (2005). Using multiple coders

to enhance qualitative analysis: The case of interviews

with consumers of drug treatment. Addiction Research

& Theory, 13(4):373–381.

Berruti, F., Nixon, G., Taglioni, G., and Whiteman, R.

(2017). Intelligent process automation: The engine at

the core of the next-generation operating model. Dig-

ital McKinsey, page 9.

Bouwman, H., Nikou, S., Molina-Castillo, F. J., and

de Reuver, M. (2018). The impact of digitalization

on business models. Digital Policy, Regulation and

Governance.

Braun, V. and Clarke, V. (2006). Using thematic analysis

in psychology. Qualitative research in psychology,

3(2):77–101.

Burgess, A. (2017). The Executive Guide to Artiﬁcial Intel-

ligence: How to identify and implement applications

for AI in your organization. Springer.

Celikyilmaz, A., Deng, L., and Hakkani-T

ur, D. (2018).

Deep learning in spoken and text-based dialog sys-

tems. In Deep learning in natural language process-

ing, pages 49–78. Springer.

ICSBT 2022 - 19th International Conference on Smart Business Technologies

Chakraborti, T., Isahagian, V., Khalaf, R., Khazaeni, Y.,

Muthusamy, V., Rizk, Y., and Unuvar, M. (2020).

From robotic process automation to intelligent process

automation. In International Conference on Business

Process Management, pages 215–228. Springer.

Clement, R., Schreiber, D., Bossauer, P., and Pakusch, C.

(2019). Mehrseitige m

arkte. In Internet-

Okonomie,

pages 261–278. Springer.

Collis, J. and Hussey, R. (2013). Business research: A prac-

tical guide for undergraduate and postgraduate stu-

dents. Macmillan International Higher Education.

Czarnecki, C. and Auth, G. (2018). Prozessdigitalisierung

durch robotic process automation. In Digitalisierung

in Unternehmen, pages 113–131. Springer.

Depietro, R., Wiarda, E., and Fleischer, M. (1990). The

context for change: Organization, technology and en-

vironment. The processes of technological innovation,

199(0):151–175.

Dong, X. L. and Rekatsinas, T. (2018). Data integration and

machine learning: A natural synergy. In Proceedings

of the 2018 international conference on management

of data, pages 1645–1650.

Engel, C., Elshan, E., and Ebel, P. (2021). Deploying a

model for assessing cognitive automation use cases:

insights from action research with a leading european

manufacturing company. In Proceedings of the 54th

Hawaii International Conference on System Sciences,

page 6253.

Ferreira, D., Rozanova, J., Dubba, K., Zhang, D., and Fre-

itas, A. (2020). On the evaluation of intelligent pro-

cess automation. arXiv preprint arXiv:2001.02639.

Gadatsch, A. and Mangiapane, M. (2017). Vier dimensio-

nen der it-sicherheit. In IT-Sicherheit, pages 17–22.

Springer.

Ito, N., Suzuki, Y., and Aizawa, A. (2020). From natu-

ral language instructions to complex processes: is-

sues in chaining trigger action rules. arXiv preprint

arXiv:2001.02462.

Kleehaupt-Roither, B. and Unger, T. (2018). Von rpa-

mythen zur automatisierungsstrategie. Controlling &

Management Review, 62(8):48–56.

Lacity, M. and Willcocks, L. P. (2018). Robotic process and

cognitive automation: The next phase. SB Publishing.

Langmann, C. and Turi, D. (2020). Robotic Process Au-

tomation (RPA)-Digitalisierung und Automatisierung

von Prozessen: Voraussetzungen, Funktionsweise und

Implementierung am Beispiel des Controllings und

Rechnungswesens. Springer-Verlag.

Le Clair, C., Cullen, A., and King, M. (2019). The for-

rester wave™: Robotic process automation, q4 2019.

Forrester Research.

Madakam, S., Holmukhe, R. M., and Jaiswal, D. K. (2019).

The future digital work force: robotic process automa-

tion (rpa). JISTEM-Journal of Information Systems

and Technology Management, 16.

Matthews, P. and Greenspan, S. (2020). Automation and

Collaborative Robotics. Springer.

Mohanty, S. and Vyas, S. (2018). How to compete in

the age of artiﬁcial intelligence: Implementing a col-

laborative human-machine strategy for your business.

Apress.

Moiseeva, A., Trautmann, D., Heimann, M., and Sch

utze,

H. (2020). Multipurpose intelligent process automa-

tion via conversational assistant. arXiv preprint

arXiv:2001.02284.

Mullakara, N. (2020). Robotic Process Automation

Projects: Build Real-world RPA Solutions Using

UiPath and Automation Anywhere. Packt Publishing

Limited.

Neifer, T., Lawo, D., Bossauer, P., and Gadatsch, A.

(2021a). Decoding ipaas: Investigation of user re-

quirements for integration platforms as a service. In

Wijnhoven, van Sinderen (Eds.): Proceedings of the

18th International Conference on e-Business, ICE-B

2021, July 7-9, 2021, pages 47–55. SCITEPRESS-

Science and Technology Publications.

Neifer, T., Lawo, D., and Esau, M. (2021b). Data science

canvas: Evaluation of a tool to manage data science

projects. In Proceedings of the 54th Hawaii Interna-

tional Conference on System Sciences, January 4-8,

2021, pages 5399 – 5408. ScholarSpace.

Sarkar, D. (2019). Text analytics with Python: a

practitioner’s guide to natural language processing.

Springer.

Scheer, A.-W. (2020). Process mining. In Unternehmung

4.0, pages 85–102. Springer.

Scheppler, B. and Weber, C. (2020). Robotic process au-

tomation. Informatik Spektrum, 43(2):152–156.

Smeets, M., Erhard, R., Kaußler, T., et al. (2019). Robotic

process automation (rpa) in der ﬁnanzwirtschaft.

Robotic Process Automation (RPA) in der Fi-

nanzwirtschaft.

Stoudt-Hansen, S., Karamouzis, F., Villa, A., Ray, S.,

Dunie, R., Sturgill, N., Shotton, L., Miers, D., and

Biscotti, F. (2019). Predicts 2020: Rpa renaissance

driven by morphing offerings and zeal for operational

excellence. Gartner, available at https://www. gartner.

com/doc/reprints.

Taulli, T. (2020a). RPA Foundations, pages 1–25. Apress,

Berkeley, CA.

Taulli, T. (2020b). RPA Skills, pages 27–51. Apress, Berke-

ley, CA.

Taulli, T. (2020c). RPA Vendor Evaluation, pages 101–114.

Apress, Berkeley, CA.

Themistocleous, M. and Morabito, V. (2012). How can

user-centred design affect the acceptance and adop-

tion of service oriented healthcare information sys-

tems? International journal of healthcare technology

and management, 13(5-6):321–344.

urkoglu, C. (2020). Robotik-Prozessautomatisierung mit

SAP Intelligent RPA. BoD–Books on Demand.

Van der Aalst, W. M., Bichler, M., and Heinzl, A. (2018).

Robotic process automation.

Willcocks, L. P., Lacity, M., and Craig, A. (2015). The

it function and robotic process automation. The Out-

sourcing Unit Working Paper Series (May): 1–39.

Zhang, C. (2019). Intelligent process automation in au-

dit. Journal of emerging technologies in accounting,

16(2):69–88.

The Role of Marketplaces for the Transformation from Robotic Process Automation to Intelligent Process Automation