IA4AI

Space shortcuts

Page tree

Date


Attendees

 Rosie Wilson Hugo Vaughan (TM Forum) Dave Milham  Stuart Dunn Luca Franco Varvello Yuval Stein Gaetano Biancardi Lingli Deng Emmanuel A. Otchere Benoit Radier  Yvonne Kuimba  Manoj Nair 

12 Participants 

Apologies




Input Material

Meeting Agenda

  • Meeting Recording / AI Companion / IPR Declaration
  • Align teams on ODA and AI activities
  • Agree handshakes between the different teams
  • Agree who will lead the group
  • Agree definition of “AI Ready Component” 
  • Create task force and set agenda (decide on cadence of the calls)

Call for IPR Declaration – IPR Statements for Use in Meetings

Call for IPR Declaration

The statements below are to be declared at the beginning of each Project meeting and/or Informational meeting.  Meetings can be conducted via conference calls, online webinar, or face-to-face. Select the appropriate statement for your meeting and minute the results.

  • Cross-Project Meetings (Open to Eligible Persons Only: Eligible persons must be members of one or both projects in order to attend)

“This meeting is being conducted under the terms of IPR Mode RAND as defined in the By-laws Annex 1: Policy on Intellectual Property Rights.  All IPR should be declared.” 

Meeting Type

Cross-Project Meeting

IPR Claims

No IPR Declared


Meeting Recording

Meeting Notes

  • Align teams on ODA and AI activities
    • The AI and ODA Sync Task Force was initiated to align the teams on ODA and AI activity, with the aim of identifying overlapping projects and assets.
    • Emmanuel A. Otchere shared details about the AI Closed Loop Automation Project, which involves formalizing key concepts and ideas to leverage AI in automation.
      • See GB1077 AI for ODA Components v1.0.0 CLA-341
      • This document provides clarity on how intelligence can be realized through TMF standards including business processes, information frameworks, functions and open APIs.
    • The project has crystallized into a set of use cases and topics, with a focus on closed-loop management.
    • The team agreed on the need for a task force to lead these discussions and set an agenda for future meetings.
    • Emmanuel A. Otchere discussed the development of software assets to support AI automation targets.
    • Emmanuel A. Otchere explains that the project has evolved from formalizing key concepts into a reference architecture based on member use cases spanning network, utility, and business support areas.
    • The team has developed several components including anomaly management, AI model manager, knowledge manager, close loop manager, inference manager, simulation manager and natural language processor, with Emmanuel A. Otchere noting that they consolidated originally separate anomaly-related components (detection, prediction, mitigation) into a single suite based on vendor and service provider feedback.
  • Classification of Business Processes and AI
    • Yvonne Kuimba  and Emmanuel A. Otchere discussed the classification of business processes and AI management functions.
    • They agreed to group these into business functions and AI management functions, with the possibility of further discussion to find the optimal grouping.
    • Emmanuel A. Otchere emphasized the importance of clarity on requirements, scenarios, and supporting assets in the component work.
    • They also discussed the need to understand use cases for intelligence services, with anomaly management being one example.
    • Emmanuel A. Otchere presented a document providing insights on the concept of intelligence and how it gets realized based on foundational assets.
    • Emmanuel A. Otchere discussed the role of the intelligence management block in ODA, emphasizing its function as a cross-functional context area.
    • He also highlighted the importance of understanding that intelligence is embedded in various components of ODA.
    • AI could also be implemented in ODA as a service or a side-car 
    • Gaetano Biancardi agreed with Emmanuel A. Otchere's points and suggested that the industry's evolution would provide clarity on the best practices for implementing AI.
    • Yvonne Kuimba  proposed grouping the work into two streams:
      • Outside in: Embedding AI use cases into ODA components 
      • Inside out: ODA foundations needed to manage AI 
    • Emmanuel A. Otchere and Gaetano Biancardi supported this direction, emphasizing the need for collective agreement on cases where embedding makes sense and leveraging extra context.
  • Integrating AI Into ODA Operations
    • The team discussed the potential of integrating AI into their operations, specifically in the context of the ODA component. 
    • Gaetano Biancardi proposed to review the existing ODA use cases and considered the possibility of adding AI to existing use cases and the implications of human involvement in these processes.
    • The importance of managing control loops was highlighted, with a focus on the need for regulation and management as they move towards closed-loop practices.
    • The team also referenced previous discussions about partners for UI implementation and the importance of understanding the objectives of these implementations.
    • Gaetano Biancardi proposed to identify where AI can be applied to existing Components or where new Components are needed. 
    • The team also identified the need to identify the top AI uses cases and where needed to create new use cases. 
  • AI Component Standardization and Integration
    • Dave Milham discussed the need for additional components, particularly in relation to sidecars and component suites, suggesting further investigation into component granularity.
    • He emphasizes that AI will be embedded in ODA production, requiring different APIs for components.
    • Dave Milham also highlights the need to refine the intelligence management block and rethink component grouping.
    • The team agrees that intelligence is pervasive and that standardized component specifications and APIs are necessary to accommodate various models.
    • Dave Milham mentions the value of the AI closed loop team's work and suggests reviewing Kevin McDonnell's compilation of AI-related work across the organization as a starting point for the task force.
  • AI Technology for Traditional Techniques
    • Dave Milham discussed the overlaps and extensions in different concepts and work done by various teams.
    • Lingli Deng suggested that AI technology could be used to upgrade traditional techniques and models, and that a common basis could be provided by ODA to address different use cases.
    • Dave Milham agreed and mentioned a contribution to multi-agent from Ericsson to accelerate.
    • Lingli Deng also highlighted the importance of multi-agent collaboration, which Dave Milham agreed with, noting that it could involve different internal technologies optimized for specific use cases.
  • Integrating AI Agents Into ODA
    • The group discusses the progress on integrating AI agents into the ODA framework.
    • Emmanuel A. Otchere presents a new AI agent specification template that helps articulate a set of agents from ODA components.
    • Dave Milham notes this has implications for component specifications.
    • Yuval Stein highlighted that this work is related to the presentation by Lester Thomas and Gaetano Biancardi, shared at Accelerate on extending components to agents.
      • See presentation here
  • Two ongoing workstreams:
    • Yvonne Kuimba  summarizes two ongoing streams: 
      • A task force developing a position paper on realizing agents in ODA
      • An innovation hub/catalyst project extending canvas, with the goal of maintaining architectural alignment across different projects.
  • Forming AI Task Force
    • The team discussed the formation of a task force to address AI-related issues. 
    • They agreed to meet weekly and decided to define the AI agent and AI-ready components in a future discussion.
    • Space created for members to add agenda items to the sync call
    • The team also decided to set up a page for populating discussion topics and to give everyone access to it.
    • They discussed the need for co-chairs from different teams and agreed to invite volunteers for this role.
    • The team also discussed the need for strong representation from CSPs in the conversation.

Action List

  • Emmanuel A. Otchere Continue developing and finalizing component specifications for AI model manager, knowledge manager, close loop manager, inference manager, natural language processor, and simulation management components
  • Emmanuel A. Otchere Consolidate requirements and create component specifications for the anomaly management suite
  • Yvonne Kuimba & Emmanuel A. Otchere Organize work streams into two categories - 'outside-in' and 'inside-out'
  • Dave Milham Review and refine the intelligence management block structure and component grouping strategy
  • Yvonne Kaps to give task force members access to the working page.
  • Yvonne Kuimba to issue a call for volunteers to chair the task force.
  • Yvonne Kuimba to follow up with potential CSP representatives, including Benoit Radier and others from Cisco and China Mobile, for involvement in the task force.
  • Task force to develop a position paper on how agents can be realized in ODA.
  • Task force to discuss and define AI agents and AI-ready components in the next meeting.
  • Task force members to bring potential conflicts or crossovers in assets they're working on to future meetings for discussion.

Next Meeting details

Scheduled Date:         

Link to Project Calendar:

Apologies for next week’s meeting: 

Requested topics for next meeting agenda: 

  • Agree handshakes between different teams
  • Need to agree definitions in future calls.  
    • Definition of “AI Ready Component” 
    • What is an "AI Agent"?