Introduction to AdaptivEnterprise Solution Suite

Document Sample
Introduction to AdaptivEnterprise Solution Suite Powered By Docstoc
					Introduction to AdaptivEnterprise TM
Solution Suite
A paradigm shift is occurring…….

    “To build business services, firms should invest in composite application
              technology from players like Agentis Software and SAP”
                             - Forrester Research -

 Gartner calls our category of services the Integrated Services Environment and
                        have labeled us “the most visionary”
                                - Gartner Group -

The way to successfully implement a Web-Services Architecture is thru the use
                         of goal directed agent technology
                                     - W3C -

  Agentis is leading a major paradigm shift in the way companies approach the
      entire application lifecycle process. We have projects around the world
     leveraging our platform to gain competitive advantage in their respective
An extensive body of theory

                  From the Space Shuttle to the Adaptive Enterprise
                        • Shuttle’s Real-Time Fault Detection and Management System
                        • Adapted to business problems by Australian Research Institute
                        • In production today for enterprise mission critical systems

 •   “Belief-Desire-Intention (BDI)” agent theory
      –   Research project initiated by NASA in 1986 at Stanford Research Institute
      –   Objective: real-time decision making in unplanned situations
 •   The “Procedural Reasoning System”
      –   Developed at the Australian Artificial Intelligence Institute (co-owned by SRI and Australian
      –   Delivered in 1993
 •   Theory extended and applied to military and business situations by AAII
      –   Air defense, air traffic control, etc…
      –   real time decision making in business environment
Agentis Corporate Overview

•   Corporate Facts
     –   Operations in US, Europe, Asia and Australia (Development Centre in Melbourne)
     –   Sales offices in San Francisco, NYC, Chicago, Paris, London and Melbourne
     –   Customers and engagements in US, France, Germany, UK, Japan, and Australia
     –   Key partnerships in Europe (Cap Gemini)
     –   Key Investors: Bear Sterns, Partech (BNP Paribas), Draper Fischer Jurvetson
     –   Senior executive team led major publicly traded software companies (Informix, IBM, Oracle,
         Unisys, CA, …) including former Informix CEO
•   Company successes and history of the technology
     –   Initially developed for space shuttle (real time fault detection and management system) and
         later extended to business applications (IP fully owned by Agentis)
     –   Core mission-critical systems at DHL running on Agentis Platform (billing, call center tracking,
         alert system, online pricing …)
     –   Mission-critical solutions for Banking, Insurance, Transportation and Logistics
     –   Endorsed by Industry analysts for adaptive enterprise space (Gartner’s most visionary,
         Forrester’s leading provider)
Agentis Approach to Adaptive Applications
•   Drive development through a high level application model where the “model is
    the code”
     – Business owners participate in development
     – Process centric development
     – Automate low level code writing
•   Put intelligence into systems so that they can autonomously handle variations in
    business rules, process context, and exceptions as they arise
     – Isolate independent tasks
     – Define task purpose and usage rules and let system find proper sequence
•   Deploy highly scalable, high performance solutions
     – Architects need control over system design
     – System architecture decoupled from business logic

    • Radically (up to 75%) reduces time to market and cost of development
    • Application is easier to maintain and adapt to changes in requirements
    • Automation of highly complex problems with many rules and combinations
Why most systems are not adaptive

•   Business logic, processes, and rules are locked up in low level, programmers
     – Hand crafted Java, C#, VB, or COBOL
     – Business users cannot validate functionality in implementation code
•   Statically linked code
     – Analysts and programmers must explicitly define all permutations
     – All sequences of steps must be coded at design time
     – Class models difficult to change once deployed
•   Purposely limit sophistication of process to avoid system complexity
•   Not an event driven technology (Batch systems)
Agentis AdaptivEnterprise Solution Suite
                AdaptivEnterpriseTM Solution Suite is the only adaptive,
               Integrated Service Environment for the automated creation,
                  deployment and real-time maintenance of applications.

• AdaptivEnterprise Solution Suite is a single, integrated software
  infrastructure platform where application services and interfaces are
  defined, and then automatically compiled and deployed.
   AdaptivEnterprise Solution Suite provides an application
    development environment aimed at productivity and

   AdaptivEnterprise Solution Suite is open, supporting XML and
    Web Services for re-use of existing internal and external
    services, and deployment for enterprise integration through
    standard adapters.
Goal-Directed Agents


                      Goals               Plans
 • A Goal-Directed Agent:
    – Is an independent software program acting intelligently and autonomously
      to deliver services
    – Is knowledge-based and operates in a somewhat unpredictable
    – Achieves goals in response to requests or events
    – Determines how to achieve goals by using pre-defined plans
Goal-Directed Agents: how they operate
     Agents use a set of pre-defined plans which they execute to achieve
                           goals or react to events
  A PLAN is a simple sequence of a few “activities” designed
  to achieve a GOAL under specific conditions (RULE)              GOAL
  Activities are of three types:

  • Set new GOAL or                Activity
    call external function
  • Action (Update) agent

     Agents track the current situation with AGENT DATA
                                                          DATA:   xxxx    Environment
                                                          DATA:   xxxx
 Agent data represents the environment or the internal    DATA:   xxxx
 state of the agent                                       DATA:   xxxx
                                                          DATA:   xxxx
 Agents use agent data to determine whether a rule is     DATA:   xxxx
 satisfied                                                DATA:   xxxx
                                                          DATA:   xxxx
How Agents Work

      Goal Hierarchy                 Current State          Plan Library

                                                            Condition 1          Condition 1             Condition 1
                                                                Condition 1          Logic
                                                                                       Condition 1         Logic
                                                                                                             Condition 1
                                                                     Condition 1         Logic
                                                                                           Condition 1           Logic
                                                                                                                   Condition 1
                                                                         Condition 1           Logic
                                                                                                 Condition 1           Logic
                                                                                                                         Condition 1

                                                                              Condition 1            Logic
                                                                                                       Condition 1         Logic
                                                                                                                             Condition 1

                                                                                 Logic                   Logic                   Logic

                                   Plan Selection Engine

  Process Execution
How Agents Work

   Goal Hierarchy                             Current State          Plan Library

                                                                     Condition 1          Condition 1             Condition 1
                                                                         Condition 1          Logic
                                                                                                Condition 1         Logic
                                                                                                                      Condition 1
                                                                              Condition 1         Logic
                                                                                                    Condition 1           Logic
                                                                                                                            Condition 1
                                                                                  Condition 1           Logic
                                                                                                          Condition 1           Logic
                                                                                                                                  Condition 1

         Sub-goal                                                                    Logic
                                                                                       Condition 1

                                                                                                                Condition 1

                                                                                                                                      Condition 1


                                            Plan Selection Engine

 Process Execution
                    Objective   Objective
                    Condition   Condition
                      Logic       Logic
How Agents Work

   Goal Hierarchy                             Current State          Plan Library

                                                                     Condition 1          Condition 1             Condition 1
                                                                         Condition 1          Logic
                                                                                                Condition 1         Logic
                                                                                                                      Condition 1
                                                                              Condition 1         Logic
                                                                                                    Condition 1           Logic
                                                                                                                            Condition 1
                                                                                  Condition 1           Logic
                                                                                                          Condition 1           Logic
                                                                                                                                  Condition 1

         Sub-goal                                                                    Logic
                                                                                       Condition 1

                                                                                                                Condition 1

                                                                                                                                      Condition 1


                                            Plan Selection Engine

 Process Execution
                    Objective   Objective       Objective
                    Condition   Condition       Condition
                      Logic       Logic           Logic
Implicit Programming:

                                                  Implicit Programming
 Explicit Programming

                                                            DATA:   xxxx
                                                            DATA:   xxxx
                                                            DATA:   xxxx
                                                            DATA:   xxxx
                                                            DATA:   xxxx
                                                            DATA:   xxxx
                                                            DATA:   xxxx

• Program flow must be explicitly coded at   • Design only Plans and information
  design                                       model
• Complex and rigid                          • Independent components easily
• Involves tedious programming tasks           added or modified
• Either becomes unmanageable or limits      • Agent manages program flow control
  possible variations                        • Agent handles error and recovery
An Integrated Solution Suite
      ARCHITECT                         ANALYZER
    Design/Development                 Analyze/Optimize
• Intuitive graphical modeling   • Run-time and off-line
• For business analysts and        execution analysis
  developers                     • Visualization for business
• Design and develop agent,        analysts and developers
  agent data and set of plans

        BUILDER                           SERVER
      Compile/Deploy                   Run-time Engine
• Instant code generation        • Where the agent resides
• Instant deployment in the      • The agent dynamically
  selected application server      assemble the application
• “The model is the code”        • Deployed in application
AdaptivEnterprise Architect


    • Intuitive graphical modeling
    • For business analysts and
    • Design and develop agent, agent
      data and set of plans
Development using graphical modeling
    Common environment for business analysts and IT developers
•    Goal-directed modeling
•    Business modeling
     (services, business rules,
     process flow)
•    Extensible modeling
•    Multi-user repository
•    Easy to learn and use

    Describe the agent
    and its components

    Design the plan
    using drag & drop

    Declare usage rule
    and priority
Components of a Goal-Directed Agent
                 Architecting the agent is simplified with the explorer
Reusable libraries
the agent may use to achieve                                 Agent data that the agent will
some goals                                                   monitor at run-time

Service requests and external
events the agent is designed to                           Library of plans the agent may use
respond to                                                internally

Published services and internal services have a goal (    if published,             if internal)
                     and one or several plans to achieve the goal

       Published service with 1 plan                Internal service with 3 plans
Integration: Extending the modelling language

• Facilities
    – The models used to define the business logic can be extended by importing user-
      defined or Agentis provided java beans for integration to external systems
• Agentis-provided facilities
    – JDBC for databases
      (Oracle, SQL Server, DB2, etc…)
    – Java Messaging Service (JMS)
      (MQ Series, Tibco, WebMethods, etc…)
    – Java Connector Architecture (JCA)
      for packaged applications
      (SAP, Siebel, Peoplesoft, etc…)
    – Web Services (SOAP / WSDL)
    – Web User Interface (WebUI)
      for “conversational” interaction with a web
    – Java / EJB
AdaptivEnterprise Builder

    • Instant code generation
    • Instant deployment in the
      selected application server
    • “The model is the code”
Configuring Application Deployment
  Where the IT professional controls the actual deployment of the application

                                                   Selecting the application server

                                                Selecting transactional control
Model-to-Deployment in Minutes

 Once deployment parameters have been set by the IT professional, IT
 developers may compile, deploy and execute agent-based applications

  With a single command: “BUILD”

  Total time “compile and deploy”
                                          Ready to run the application
Development Phase Benefits

• Time To Market Reduction
   – Implicit programming reduces
     the amount of design and
     development activities
   – Coding phase is eliminated with
     instant code generation
   – Plans are functionally complete
     and independently validated
   – Continuous application
     evolution allows quick initial
     deployment, then incremental
• Qualitative benefits
   – “The model is the code”  application always in sync with business requirements
   – Continuous refinement of business requirements during design and development is
     immediately replicated in the application
   – Optimal segregation of duties between developers and business analysts
   – Easily learned by both developers and business analysts
AdaptivEnterprise Server

                                   Run-time Engine
                           • Where the agent resides
                           • The agent dynamically assemble
                             the application
                           • Mission-critical performance
Architecture of a compiled agent
                                                      App Server
                                                   App Server
                                               App Server
                 Web Proxy
                                         Web              EJB                               Database         Replication
    Client      Or Hardware                                                                                                Database
               Load Balancer           Container        Container

             Struts - based Servlet                                           Agent Runtime                                       Common
                     Servlet session                                           Agent                                              Services
                   Servlet session                                           Agent
                 Servlet session                                        Agent                                                          Security
             Custom            Agent
Requests      Struts
                                                              Session Bean                    Entity Bean
                                API         RMI                                                                                       Transaction
                                                           Services   Agent Data              Agent Data

                                                             Plans     Facilities          Message Beans                               Logging
Responses     Custom
               JSPs                                                                            Events
                                                                                                                                  Monitoring &

                                                                                    Integration Facilities                            Connection
                          JMS                                             JDBC               JMS             SOAP                      Pooling
                  Web Service                           Engine                                                                        Connection
                                                                         WebUI            Bean Shell     Java Bean
                 (SOAP/HTTP)                                                                                                           Pooling
Run-time: An enterprise environment

 • Parallel processing
    Multiple instances of the same agent can be deployed on the same CPU and take
      advantage of multiple CPU’s

 • AdaptivEnterprise Server and application deployed as EJBs
   in J2EE application server
      inherit the characteristics of the application server
        •   Scalability
        •   Clustering

    leverages application server common services
        •   Security
        •   Transaction
        •   Monitoring and Management

 • Agents can be deployed on major J2EE application servers
    Weblogic, Websphere, Oracle AS, JBoss
AdaptivEnterprise Analyzer

                         • Run-time and off-line
                           execution analysis
                         • Visualization for
                           business analysts and
Tracing & Visualization

 Run-time full trace capabilities allow
  presentation and reporting of the
           agent behavior
  The Agentis Value Proposition
                 LOWER TIME AND COST OF DELIVERY BY 50% TO 80%

     Time To Market                                Total Cost of Ownership
                                                   • Short TTM
• Initial deployment: 3-4 months                   • Low maintenance costs
• Continuous evolution of the                      • Application robustness reduces
  application                                        need for manual interventions
• Application always in sync                       • Application robustness reduces
  with most current business                         risk of human error and its
  requirements                                       adverse financial impact
• Maximum productivity from                        • High scalability/performance
  business requirement to                            lowers cost of hardware
  deployment                                       • Extensively leverages existing
• Extensively leverages                              IT assets
  existing IT assets

Our Experience
What is not an adaptive application

• Takes 1-2 years to develop or to customize package
   – Requirements will change in the time since development started
• Takes several months to update
   – Missed opportunities or prolonged problems
• Requires a team of people to handle exceptions or perform
  the “intelligent” work
   – Inconsistent
   – No audit log of rules and processes used
   – Actual process logic and rules not automated and not documented
• Doesn’t reflect all desired preferences
   – the old Ford adage, “you can have it any color as long as it’s black”
   – Too many possibilities
   – Too little time to consider all possibilities
When Adaptive Applications are Needed

• Fast time to market, or lowered cost of
• Intelligent integration
• High volume decision making
• Myriad of changing rules and regulations
• Frequent exceptions requiring manual
Gaining Adaptivity – DHL’s Billing System

• Moving from Effective
   – 600 billing clerks to manually calculate bill at end of month
   – 50,000 data aggregation COBOL programs to retrieve transaction
     information from 7 regional billing systems
   – Limited options in pricing rules and approaches
   – Conflicts in pricing rules across regions handled inconsistantly
• To Adaptive
   –   Automated bill calculation in real-time, at any time
   –   New pricing promotions can be added by end-users
   –   High level, Goal-Directed model enables rapid maintenance
   –   Single global system that rationalizes pricing rules, but considers
       regional differences
Lowering Time and Cost of Delivery
  DHL – Global Billing Solution
                             Development                            Remarks                           System   Time to          System size
                             Strategy                                                                  cost    market
   Traditional programming

                             Deploy a SW package     Discarded due to long TTM and difficulty to        NA         NA                   NA

                             Outsourcing             Risk of delays due to the fluid nature of the     $25M    18 months            10,000 days

                             Internal development   Concerns about availability & continuity of the    $12M    18 months            10,000 days
                                                    development team; no protection against cost
                             Off Shoring             Concerns about lack of integration between        $6M     18 months            10,000 days
                                                       architecture and offshore development

                             Internal development   Internal resources from the Shipping company       $3M      9 months            2,000 days

                                                               with support from Agentis

All of the explicit programming solutions have maintenance work based on programmers changing application code and doing regression testing.
Maintenance of the agent based implicit programming is done by creating or updating a plan using a graphic user interface. Changes can be easily and
quickly implemented.
Intelligent Integration

 ADSL Order Management
  (Primus Telecommunications)

• Provide the CSR with a single web browser interface that
  allows them to intuitively and effectively perform inbound
  ADSL sales.
   – Inbound sales process flow for a New Customer wanting to sign up
     for Broadband ADSL
   – Integration to Telstra for Service Qualification (SQ) Check, C-PAY
     for credit card payment, Vantive, 3 custom applications, and to IVR
   – CSR’s have one stop access to all customer, order, and history
     data even though information is stored across multiple systems
Topographical View of Primus system

                               CSR (user)

                       Business Process Services
                   Qualify         Advise       Finalize   Provision

                      Integration Data Services

     Vantive                 Helpdesk   Tulip      IVR     C-Pay
High Volume Decision Making

 Event Management and Alerts (DHL)

• Subscription based system to provide notifications
  of business events
  – Multiple messages coordinated into business events
  – Missing messages create business event
  – Users dynamically subscribe to events and specify
    notification channel (email, SMS, etc.)
  – Handles millions of events per hour / billions per year
  – Core infrastructure for multiple applications requiring
    event notification
Handling a Myriad of Rules and Regulations

• Customer Lending Portal
  (Retail Bank)

• Product applications performed on-line over
  Internet, Kiosks, or in branch
   – Products include Credit Cards, Auto loans, Mortgages,
     Personal loans, etc.
• Business rules and policies used to make
  approve/decline decision
• Additional products qualified for offered
Exception Management

 Package Tracking and Tracing

• CRM system to manage call center requests to
  track packages
  • Single global system
  • Work flow routing, escalations, management reports
  • Automated re-routing of packages when exception
    occurs so as to meet SLA with client

Shared By:
fanzhongqing fanzhongqing http://