March 1,2000 Volume 3 - Number 4
The new generation of enterprise portals meshes with intelligent agents
EIP: The Second Wave
Mark M. Davydov
Today’s business — or, more accurately, a significant number of companies from all
sectors of the economy — has had to confront three new facts of management.
First, companies are now competing in a customer’s market. Second, this market requires an entire
retooling of the conventional business order — specifically, shifting from a classical view of
management activities grouped around back-office systems (planning, organizing, directing, and
controlling) to one that groups activities around producing systems to better serve the customer. Third,
the competition for these customers is no longer confined to the traditional marketing approach, never
mind a wide-coverage product advertisement; and new, clever, and powerful forms of attracting customers
through the company’s online information services are truly forces to be focused on.
This focus involves developing customized content, new community aggregation models, information
brokering capabilities, and other value-added services, as well as creating new revenue methods that
exploit core e-commerce competencies while opening the way for new information supply partnerships.
As
a result, the capabilities of various “online information-services enabling” tools — most
importantly, different types of the “enterprise portal” (also known as the enterprise
information portal or EIP) — are finding their way to the Web infrastructure in organizations of all
sizes. Although the concept of EIP was introduced only a few years ago, we are already in the midst of
the second wave of this very important technology.
Christopher Shilakes and Julie Tylman described in 1998 a “first wave” of EIP evolution, which
essentially stated that the first-wave EIP products and related technologies present a strong
decision-support and content-management emphasis.
The second wave is shifting the primary focus of EIP. Whereas EIP did emphasize broadly based and
generalized decision processing and mass dissemination of corporate information, it now targets
collaboration and highly targeted and personalized distribution of content, bundled with multiple types
of specialized expertise-oriented services. These trends, happening over the past year or so, indicate
that the EIP concept is on the verge of an explosion along four key directions:
• Enterprise Business Intelligence Portals (EBIP) provide the central launching point
for corporate decision-processing and content-management applications. Their primary focus is to connect
users with structured and unstructured content relevant to them.
• Enterprise Collaborative Processing Portals (ECPP) connect users not only with all
the information they need, but also with everyone they need. ECPPs consolidate groupware, email,
workflow, and critical desktop applications under the same gateway as decision-processing and
content-management applications. ECPPs are characterized by “virtual project areas” or
communities.
• Enterprise Mission Management Portals (EMMP) provide a “digital
expertise-oriented workplace,” a highly specialized and personalized Web site where everything a
user team needs (such as access to ERP applications, productivity and analysis tools, and relevant
internal and external content) to effectively manage mission-critical management activities such as
customer relationship management (CRM) is consolidated and made accessible via the Web.
•
Enterprise Extended Services Portals (EESP) do everything the first three types do, but
focus on providing comprehensive job support from the standpoint of “virtual enterprises” by
creating communities and “virtual service spaces” of channel partners, suppliers, distributors,
and customers.
The convergence of the first and second EIP waves will occur within one and a half years. This time will
be spent on extending architectural frameworks that guided technology from the search-based, first-wave
portals to a fully functional architecture capable of enabling expertise- and service-based, second-wave
portals.
An Architecture for the Second-Wave EIP
No one has a definite vision of a comprehensive architecture for the second-wave EIP yet. Such a
challenge can’t be left to one individual or company. It requires a strong coalition of leading
portal companies and standard driving institutions.
Fortunately, object-oriented thinking allows us to
proceed without having to solidify a comprehensive architecture. We don’t have to work with a final
master plan. We can begin with a few underlying architectural concepts and evolve into the future,
adapting and modifying the framework as we go.
The most critical concept at this stage is distributed object processing. What we need is a portal object
component model (POCM) that suppresses the specifics of directories, searching, “tunneling,”
“front-ending” of numerous applications (such as ERP, groupware, and OLAP), and many more
underlying portal technologies. We need one instead that provides a clear component structure. With POCM,
Web developers will be able to assemble a complete EIP solution (whether of a specific type or a
combination of types) using pre-built portal components. Think of an intranet or extranet running an
application such as an “EIP Director,” which consists purely of interacting component-based
objects, complete with their characteristic intranet formats. What classes of objects can be envisioned
here? The most important are the following:
• Data filtering and analysis
•
Information brokering
• Workflow management
• Data mining
• Document
management
• Mission and task management
• Simulation and gaming
• Collaborative application integration
• Personal assistance
• Risk management.
From a POCM point of view, here are the main underpinning technologies we have to play with:
•
Web site construction and browser
• Internet-based object protocol engine
• Intelligent agents.
Are Web site and browser
formalization and an object protocol engine important at this stage? Taking different combinations of
today’s competing site technologies (Windows and Unix) and object protocols (DCOM, CORBA, and RMI)
into consideration, the choice is far from straightforward, even if we choose mixed platforms and object
protocols.
As I see it, the only way to approach POCM now is to think ahead to an advanced state of affected
technologies. In a year or two, all the competing technologies involved will be forced to converge toward
a unified Web-site technology with a versatile Internet object-protocol language. In other words,
let’s not get preoccupied today with decisions that are going to end up being much different in the
near future.
Therefore, my conclusion is that, at this stage, a formal architectural definition of Web site
construction and object protocol elements is not critical. What is the most critical issue today from an
architectural perspective? It is working out how to use intelligent agents on the Web within the scope of
EIP.
A Class of Intelligent Agents to Start With
Intelligent agents are first and foremost tools that can be applied in numerous ways to make different
types of EIP a reality. Although the current agent applications are rather experimental and ad hoc,
primarily targeting information searching and collaborative filtering aspects, it is only a matter of
time before intelligent agents will play a decisive role in all aspects of EIP. An application that
immediately springs to mind is to use intelligent agents in EMMPs such as CRM portals, specifically in
areas of competition.
In the digital economy, you gain competitive advantage by exploring and
exploiting the decentralized points of control under conditions of abundant resources and scarce human
attention. Here, the most powerful technologies are those that extend, augment, and develop
relationships. As the relationships between enterprises and their customers become more complex, the
enterprises need more information and advice on what these relationships mean and how to exploit them.
Intelligent-agent technology offers some very interesting options for addressing such needs.
Consider just one example: Customers set certain priorities in their demands for products and services
that lead to purchase-decision rankings based on price, service, delivery time, and quality. Intelligent
agents can master individual customers’ or customer groups’ demand priorities by learning from
experience with them, and can quantitatively and qualitatively analyze those priorities. Several
commercial products already provide such functionality; for example, PersonaLogic uses an intelligent
agent that considers priorities in searching for a car.
Agents are software entities that are able to
execute a wide range of functional tasks (such as searching, comparing, learning, negotiating, and
collaborating) in an autonomous, proactive, social, and adaptive manner. The underlying technique of the
agent — a C/C++ program, a VB script, or a Java program, for example — is irrelevant as long as
the agent is capable of displaying intelligent behavior. “Intelligence” and the related term
“intelligent agent” are difficult to define. This matter has been the subject of many
discussions in the field of artificial intelligence. At a minimum, an intelligent agent has to be able to
accept the user’s statement of goals and underlying preferences (rules) and carry out the task
delegated to it, applying an inference engine or some other reasoning mechanism to act on these
preferences.
There is a vast range of services customers require that intelligent agents can address. Some of these
services may include:
• Customized customer assistance with online services: news filtering, messaging, scheduling, making
arrangements for gatherings, ordering, and so on
• Customer profiling, including inferring information about customer behavior based on business
experiences with the particular customer
• Integrating profiles of customers into a group of marketing activities
• Predicting customer requirements
• Negotiating prices and payment schedules
• Executing financial transactions on the
customer’s behalf.
These examples represent a spectrum of applications from the somewhat modest, low-level news filtering
applications to the more advanced and complicated customer relationship management applications that
focus on predicting customer requirements. The main point is that an intelligent agent is an intermediary
between the enterprise and its customer, and a source of effective, utilitarian information encountered
at different virtual destinations.
In terms of the state of the technology, the only obstacle for mass adaptation of intelligent agents is
the lack of generally agreed-upon standards, especially for agent communication language. A first step in
this direction has been made with the development of Knowledge Query Manipulation Language (KQML). A lot
of work has to be done in this area, as most of the current agent systems do not comply with KQML.
Organizations such as the Internet Engineering Task Force (IETF) and its working groups are addressing
the agent communication language issue.
Predicting tomorrow’s Internet developments depends strongly on what is a leading development today.
It is the portal. Furthermore, agent-empowerment of portal technology will make a difference. It will
allow enterprises to help their user communities (existing and prospective) understand what the
agent-enabled applications could do. In some mission-critical activities, such as customer relationship
management, you need intelligent agents in order to be a contender.
Mark M. Davydov, Ph.D, (mark.davydov@
den.galileo.com) specializes in advanced systems architecture and data management
solutions. He has planned and implemented enterprisewide systems-architecture
initiatives for more than 30 Fortune 500 companies.
