[INET'99] [ Up ][Prev][Next]

Brokering Automated Enterprises

Jerry FOSS <jerry.foss@marconicomms.com>
Marconi Communications Ltd.
United Kingdom


Information brokerage provides intermediary services between consumers and information sources. However, it seems that intermediary businesses are set to evolve beyond mere brokerage: To serve their clients in the new online economy infomediaries need to offer logistics brokerage, marketing consultancy services, and relationship management. In this scenario, automated online businesses emerge and continually re-assess their market and self-organize to achieve their optimal business development. This process is likely to involve increasingly active intelligent elements in the Net. Further, the interactions of global masses of information may bring about some interesting consequences.


From the global info-jukebox...

For a short while it seemed that the free-spirited evolving Internet allowed all users access to all information sources and suppliers: The dis-intermediated online info-bazaar needed no middlemen. However, the Net is still in its early stages of commercial operation and yet forecasts for millions of surfers spending billions of cyber-dollars show the undying requirement of those who broker deals between service and customer. Further, the human providers and customers may become well outnumbered by automated enterprises trading information as a vital commodity around the global Net.

Most aspects and commodities in society are becoming represented as information stored and accessible in the world's networks: business, entertainment, human activities, organizational activities; activities of the world's communication and information networks themselves; information about information. There appear to be two main families of information services evolving: (i) information trading and processing, and (ii) group activities. As more people are empowered with computing and communications capabilities, new and unforeseen services and applications will emerge [1].

The standard trading model (supplier via intermediary to customer) has been with us for centuries and seems unlikely to disappear, even (or especially) with the introduction of the Internet (see diagram 1). This paper discusses the scenarios for how brokerage may organically evolve; what drives its development; its role in the developing information services; and the operation of online enterprises. It is suggested here that the concept and traditional scope of brokerage will extend to intelligent marketing and business relationship brokerage. Existing commercial trading relationships will need to adapt to the new dynamics of online life. These roles will often be provided by independent, organically evolving businesses, manned or automated.

A number of broker services are appearing on the World Wide Web. Many of these are "shopping" agents, such as Buy.com [2]. Massachusetts Institute of Technology (MIT) has an acknowledged history of agent recommendation techniques dating from HOMR, the Helpful Online Music Recommendation service, using client profiles. From this project Firefly was born and added a CD "shopping trolley service" and communal media services. This is an example of the way in which "new media" services can quickly diversify. (Firefly has been an established part of Net-culture, and has been acquired by Microsoft.) Firefly [3] is now developing relationship management software. This subject is considered later in this paper, and is considered here to be a vital intermediary service in future information trading environments. A current MIT project, Market Maker, [4] is examining the use of agents for browsing and trading.

Many websites now offer services that blur the boundaries between a number of services; news, publishing, entertainment, shopping, messaging, diaries, scheduling, travel brokerage, and other services are seamlessly bundled into entrepreneurial online malls and portals. The significant trend is the speed at which these sites can redefine their service and reorganize their site accordingly. Portal development is currently intensifying with numerous acquisitions and collaborations, and may be the focus of the converging media industries.

Brokered service construction

Information brokerage currently seems to be viewed as little more than search and retrieval functions. Brokers are not entities that can be rigidly defined; as with any enterprise, they organically evolve. There is a vast range of services required by both customer and supplier. These may be serviced by a single broker with a wide range of services, or a number of brokers who specialize in niches and preferably collaborate in a managed organization to serve their clients. Some of these services may include:

An increasing subject of concern is the administration of information rights and revenues. Many interested parties are eagerly proposing schemes for trading while protecting the rights of info-commodities, and for distributing royalties related to information trading. Brokers may be seen as "trusted third parties" who may administer rights and royalties processes; this would require that they be registered (or licensed) with appropriate governing agencies to perform these roles.

In the increasing complexity of the trading environment, service-provider enterprises will need to assimilate their clients' (and their clients' clients) markets and relationships. This is increasingly important as collaborative information-based enterprises acquire more complex networked relationships. This function will almost certainly be outsourced to intermediaries (also called infomediaries). This trading environment and client-market relationships are discussed below.

It is suggested here that future online services will be built from reconfigurable service components (applications, agents, etc.) as needed. Successful enterprises offering online services will continually reassess the ongoing business trends and requirements in the evolving commercial environment (see diagram 2). Service components acquired from other online suppliers would be networked (see diagram 3) to form a "virtual service node" -- a suitable architecture to implement the required service. The service customer perceives that the service emanates from a specific network location, regardless of the actual networked configuration of the components that supply the service. For the service-provider enterprise, the construction of services in this manner will need brokerage to identify and commercially acquire the service components from online vendors, that is, applications software products for sale/rent from other independent retail outlets.

One option for a service provider is to offer customized services per customer: The provider needs to identify the client's requirements (who the client is, what he wants); identify the components and functions needed to perform the task (some of these may already be owned; others are additional and need to be acquired); initiate the search for the additional required components; negotiate and procure these components, including payments; integrate components into suitable architecture to perform the service for the client; perform the service; and allocate charges for the service performed. Along the market spectrum from mass-customization to general provision we have certain niche markets in which service providers will aim to attract and serve as many customers as possible, according to its business strategy. The strategy should derive the manner in which service components are planned for and procured. In this fully mature Net-centric service environment, intermediaries are essential to assist not only the instantaneous brokerage of the components, but also the marketing to derive the future business strategies for the service provider.

Brokerage businesses themselves are enterprises that need to constantly trawl their market environment and update their business practices and their configurations. They may also be collaborative organizations -- collections of business partnerships that collectively perform intermediary brokerage functions for online enterprises. For any online enterprise the continual reassessment may result in new business partnerships and new services to be offered, and therefore new networks of service components to be constructed.

The service formed by a network of components links with other services to form a platform to support services for a "higher layer," and so on. Consequently, services at any specific layer are composites of services of lower layers; therefore, service platforms are recursive. This interdependency of services supported by lower-level service platforms should be a consideration of any service provision enterprise.

Collaborative online enterprises

The trading environment is shown in Diagram 4. The various parties involved are:

This is a simplistic summary of the parties in the trading environment. Some of these relationships can become very protracted and interrelated because of the recursive nature of trading entities. For example, the providers of service components are themselves enterprises. The agent interactions may also be much more complex than indicated in the diagram.

Information gathering and exchange in the "open market" may involve other third parties -- independent / freelance agents, probably contracted by brokers. "Info-sleuths" are now appearing, and their credibility is the quality of information they commercially trade with their clients, although in this free market, quality is not standardized.

The manner of operation of the various types of agents is debatable, and in the early stages of networked interactive dialogue, compatibility can't be guaranteed. Traditional alliances of partners may attempt "local" standardization within their own groupings. As alliances re-form across these traditional boundaries, these incompatibilities would have to be faced: Either a further (voluntary) restandardization takes place or interactions occur through interpreters (also agents?). In any case we should not necessarily rely on formal standardization of these processes.

Intelligent marketing, relationship management

One of the characteristics of the online economy will be the proliferation of the use of agents representing all parties -- customers, suppliers, and intermediaries. It is quite conceivable that billions of agents will flood the Net, many of which are performing "shopping" functions -- seeking and trading merchandise. For example, in the scenarios described in this paper we are specifically interested in the merchandising of service components. In the physical market, human-based strategies are tempered by human inertia, sophisticated experiential reasoning, and (social) interaction of human traders. There is also an (extrinsic) cost of updating services, which brings a traditional degree of stability. In comparison, early agent-ware may use naive rule bases that may be too simplistic and methodical. Their obvious fast computational capabilities may accelerate a market condition into instability.

Studies into agent-based brokerage have examined the impact such brokerage will have on market dynamics, profile, and so forth, when different strategies are employed for business management and pricing. Hanson and colleagues [5] indicate that the possibility of a highly dynamic market with price wars are one very likely result, and also discuss pricing strategies, specifically myopic policies that lead to periods of calm interspersed with periods of instability. The resulting "profit landscape" (a market with profitable peaks and lower-profit troughs) is open to exploitation by well-informed agents; lesser-informed agents are unable to respond so effectively. Quality of information to active trading services is therefore a decisive factor in their effectiveness (market information gathering is discussed below).

Hanson and Kephardt [6] describe the characteristics of agent-based economies where niches are emerging. Agent economies would normally tend to self-organize into noncompetitive niches, but if they are allowed to optimize both the price and categories offered, agents tend to compete in the same narrow market, probably leaving some areas of the market unattended. Extrinsic costs and overheads may be easier to absorb in the online market than in the equivalent physical market, resulting in adaptable brokerages readily opting for easier niches. Competition may also drive automated brokerages into safer profit-earning categories of goods and services. Very low extrinsic costs may encourage brokerages to offer a larger range of categories.

When applied to the service component market, pricing and niche strategies clearly directly affect the availability of components and consequently the immediacy of service design. Such a market-responsive approach may introduce some inertia into service construction, and will certainly require intermediaries to track new components that may be outside popular niches. However, marketing the information economy will become more sophisticated. Component availability can be pre-empted by intelligence gained by the marketing activities of intermediaries, specifically those who can accurately determine a forecast of future customer-service requirements. The intermediary must then decide whether this information accelerates business (to mutual benefit of itself, supplier, and customer) and therefore can be given freely, or whether the market information itself is a chargeable commodity.

Kelly [7] advocates that in the information economy, competitive advantage is gained through ability to explore and exploit the decentralized points of control; resources (and arguably lower prices) become more abundant, but human attention becomes more scarce; as physical proximity is replaced by multiple interactions between networked entities, opportunities for intermediaries expand greatly; and the most powerful technologies are those that extend, augment, and develop relationships. As the relationships between enterprises and their customers become more complex (e.g., customers with increasingly complex requirements), the enterprises need information and advice on what these relationships mean in the evolving market -- what are the threats and opportunities; who are competitors, who are potentially useful collaborators; what is a likely profit landscape; and which peaks can be realistically exploited. This can be thought of as logistics brokerage -- to relate the market dynamics to a client's business activities. Consultant intermediation services can advise on potential strategic partnerships. This environmental relationship management is set to become a major activity for online enterprises.

Intelligent agent technology offers some implementation options for these activities. Consider the information network model in diagram 5. Four communities --that is, four enterprises -- are represented here. One or more agents take the responsibility to break out and browse ad hoc with other agents representing other infonets, for a number of purposes: for example, trawling for information of potential interest to the user of the infonet represented by this agent or promoting the agent's infonet business to other infonets. Agents may locate and contact other agents, or even have unscheduled contacts. In these liaisons, agents can compare experiences and swap or trade information of other associates for future reference, subject to previous contractual agreements and security.

At this level, agents may use other agents that assist locating potential target-agents, assist interagent functions, and so forth. Agents performing basic tasks may evolve and develop a capability to perform more complex functions, that is, display emergent behavior. Some agents may take on a nomadic life, returning periodically to update their host. This type of mobile agent may be used for policing and monitoring activities in future information environments, checking for abuses of rights infringements, and so on. Intelligent online marketing activities can be used to determine the degree and nature of self-organization of an online business.

With the information gathered by this agent activity, information businesses can fine-tune their operations and services. New services require new components; hence there is a burgeoning underworld of traded service components. There are inevitable issues about allegiances and loyalty in federated multi-agent systems. Some of the management issues are discussed by Decker and colleagues [8]. The service components (including agents) are tradable commodities in the lower layers of this service/agent hierarchy. From these complex informational activities brokerage businesses will most likely evolve to provide relationship management services between businesses, their partners, their clients, and their customers. Full-time professional intermediation is needed with the ability to manage and contextualize vast interrelated fields of information swamping online enterprises.

Project: Brokerage in an information economy

To investigate the enterprise brokerage, Marconi Communications is undertaking a research project [9] with the Distributed Artificial Intelligence Research Unit at Queen Mary Westfield College (University of London).

The project aims to study intermediary services as implemented by intelligent agent technology. The initial aim is to build a brokerage model and implement some trading services. This involves building customers with different service requirements; a number of providers of content and services; and brokerage businesses to operate between these entities. The implementation considers the use of component-ware -- elements of software that integrate and build the functionality to be offered by the brokerage business. Much of the component-ware is agent-based. These agents perform tasks such as information filtering, search, and contractual negotiation [10].

The project application scenario then considers a business evolving from an initial range of services to a new range of services as the business considers the market offering new opportunities. To prove the effectiveness and characteristics of large-scale markets, the simulated trading environment will then be "scaled-up," comprising more customers, several service providers, and more brokerages.

The project aims to deal with brokerage and online commerce generically, that is, an abstraction regardless of specific application. However, an example is needed for a demonstration, and the media industry is chosen as a sample trading domain because:

It is assumed here that the media industry covers music, films, TV/WebTV, news, journals, Internet and its services, and "portals" (i.e., customer access to the "services-mall"). It seems that for the next few years at least the media industry will be typified by enterprises breaking away from their traditional market and offering service products in adjacent areas of the media industry. Service offerings will thus be constantly built and revised.

A variety of billing strategies and customer types is to be emulated in the model that will then be extended to emulate a (relatively) simple online business evolving to provide more adventurous services.


Information network (infonet) model: Generic characteristics

Informational processing seems to conform to a generic information network model, whether it is at component level (agents, code); human level (neural activity to personal/group interactions); or enterprise, business, and organizational levels. An information network -- or infonet -- is a community of information-processing nodes; these nodes receive information, process it, and pass it on to other nodes. Infonets appear to display common properties (arguable, and not claimed to be comprehensive!):


What are the implications of the new dynamic online service economy? What is the future status and role of the nation state as the global economy becomes increasingly online? How will national legal systems affect multinational virtual organizations? How will global commerce accept cultural differences in trading, bartering, and so forth? In this volatile info-underworld, will the latency of traditional monetary currencies be sufficient, or will information become its own global currency?

Customers are becoming more sophisticated and information products are becoming increasingly complex. The access and selection of information sources to create the customized info-product is a major service-sector. The intermediary services offered by brokers are a necessary part of the online trading environment, and they are in a suitable position in the trading model to administer the legal and commercial requirements of information trading. In dynamic markets, successful businesses need to re-assess their performance and reconfigure their operation. Brokerage businesses are thus likely to evolve from their current (relatively simple) services to provide support for next-generation online enterprises. These likely developments are explained below.

First, enterprises (including brokers themselves) are composed of virtual organizations of component-ware: Software components would be acquired and disposed of as necessary, allowing the enterprise to easily reconfigure itself. Commercially this means that component acquisition, like any other tradable commodity, is subject to contractual negotiation, copyrights, royalties, and licenses in an increasingly volatile service market.

Second, brokerage businesses naturally evolve to offering "logistics brokerage" or "relationship management" -- research and consultancy relating a client's enterprise to its position in the market. This includes tracking the changing requirements of its customers, intelligent marketing, the threats and opportunities posed by other enterprises in the market, potential alliances, and so forth. Intelligent agents seem to offer an essential enabling technology for these processes, both in the search and acquisition of components, and for browsing for information from other agents who represent other partners and enterprises.

...To the info-climate

Technologies, like mobile agents, are the proxies of enterprises and customers: They serve to promote freer communication between information "clusters" in the global Net. Effectively, information can become "aware" of other information. "Spatial" relationships between information clusters can be formed corresponding to various criteria, such as similarities of content, potential interests, user groups, and customer and market information. The meta-information can derive information groupings that form affinities between each other, leading to natural clustering. However, there will always be unpredictable stimuli upsetting the equilibrium forming in any of the groupings. On a global scale this is an info-climate -- "weather systems" of information constantly clustering and reforming. The movement of the information, like the information itself, has a meaning to be interpreted, which therefore increases the need for relationship management.

Information processing in a networked environment will evolve to be noncentralized, self-organized, and integrated into overall activity -- similar to human neural activity. More efficient and effective interfaces of the future (e.g., prosthetics, "wearable computers") will allow a seamless continuum between human neural activity and the mechanized networked information processes. Will the future global information network consist of humans and all connected machines, seamless and with little or no barrier between them -- i.e., cyborg-nets? Or will an automated global information network become very discerning about which humans, if any, are allowed access?

On a global scale, interacting information masses have significant (if unknown) effects. In the global information-trading environment, the world is your hard disk; information is the most marketable commodity: Information trading is the biggest market of the future.


  1. Foss, J D, Atkin, B C M, Ackroyd, E, The Global Information Trading Environment, International Symposium of Subscribers Loops and Services, Vancouver, BC, Canada, 1993.
  2. Buy.com, http://www.buy.com
  3. Firefly, http://www.firefly.com
  4. Market Maker, http://maker.media.mit.edu
  5. Hanson, J E, Kephart, J O, Jakka Sairamesh, Price-War Dynamics in a Free-Market Economy of Software Agents, Proceedings of ALIFE VI, Los Angeles, 1998, http://www.research.ibm.com/infoecon/paps/html/alife6/alife6_public.html
  6. Hanson, J E, Kephart, J O, Spontaneous Specialization in a Free Market Economy of Agents, Artificial Societies and Computational Markets Workshop, 2nd International Conference on Autonomous Agents '98, St. Paul, Minneapolis, 1998, http://www.research.ibm.com/infoecon/paps/html/aa98/aa98_public.html
  7. Kelly, K, New Rules For The New Economy, Fourth Estate, 1998
  8. Decker, K, Sycara, K, Zeng, D: Designing a Multi-Agent Portfolio Management System, http://www.cs.cmu.edu/~softagents/papers/cikm95.pdf
  9. Brokerage in an Information Economy, http://www.elec.qmw.ac.uk/dai/projects/brokerage/
  10. Jennings, N, Faratin, P, Johnson, M, O'Brien, P, Wiegand, M, Using Intelligent Agents to Manage Business Processes, Proc. First Int. Conf -- Practical Application of Intelligent Agents and Multi-Agent Technology (PAAM96), London, UK, 1996

[INET'99] [ Up ][Prev][Next]