Phase 2: Design and create
Description
Business topics cover the overall context, roles and responsibilities, fee structures and branding.
Examples of topics are compensation mechanisms, branding, and value proposition.
Approach
Consider the all relevant business, legal and operational topics and determine which are relevant to the specific context of your AI data space and make decisions on how your design will cater to the choices made. The resources provide a comprehensive overview of topics relevant for an AI data space. In implementing the AI data space make and formalise agreements with all involved stakeholders on each topic.
Resources
Best practices
- When developing agreements, tools, processes, look for what is already out there in and outside your sector. This ensures you do not ‘reinvent the wheel’ and makes it easier to align with other initiatives in the future.
- Ensure that the interaction model is based on generic roles and not specific actors currently involved in the AI data space. This enables future changes and scalability in that the roles can easily fulfilled by various actors.
- Make the design as generic as possible and as specific as needed. This greatly increases the scalability of the design as it is easier to facilitate other use cases.
- Involve stakeholders from all participating organisations in the decision making process to ensure alignment on the results and improve buy-in from participating organisations.
- Consider all relevant BLOFT (sub) topics, but select and focus on only those topics that are especially relevant to your AI data space.
- Involve the correct experts and prepare analysis before having topic specific discussions to ensure efficient well-informed design decisions.
- Ensure all topics are considered to ensure no decisions are made which may impact future scalability.
- Double check all agreements that have a dependency on another topic to ensure that they do not conflict.
- Make use of existing building blocks from the reference implementations to reduce implementation load and increase efficiencies.
- A test-driven development approach allows design choices to be validated early to ensure the feasibility of the complete design.
Description
Legal topics cover the impact of regulation, governance structure and mutual responsibilities between stakeholders.
Examples of topics are liability, governance structure and penalties.
Approach
Consider the all relevant business, legal and operational topics and determine which are relevant to the specific context of your AI data space and make decisions on how your design will cater to the choices made. The resources provide a comprehensive overview of topics relevant for an AI data space. In implementing the AI data space make and formalise agreements with all involved stakeholders on each topic.
Resources
- DSC Blueprint
- SITRA rulebook
- Relevant AI regulation*
Best practices
- When developing agreements, tools, processes, look for what is already out there in and outside your sector. This ensures you do not ‘reinvent the wheel’ and makes it easier to align with other initiatives in the future.
- Ensure that the interaction model is based on generic roles and not specific actors currently involved in the AI data space. This enables future changes and scalability in that the roles can easily fulfilled by various actors.
- Make the design as generic as possible and as specific as needed. This greatly increases the scalability of the design as it is easier to facilitate other use cases.
- Involve stakeholders from all participating organisations in the decision making process to ensure alignment on the results and improve buy-in from participating organisations.
- Consider all relevant BLOFT (sub) topics, but select and focus on only those topics that are especially relevant to your AI data space.
- Involve the correct experts and prepare analysis before having topic specific discussions to ensure efficient well-informed design decisions.
- Ensure all topics are considered to ensure no decisions are made which may impact future scalability.
- Double check all agreements that have a dependency on another topic to ensure that they do not conflict.
- Make use of existing building blocks from the reference implementations to reduce implementation load and increase efficiencies.
- A test-driven development approach allows design choices to be validated early to ensure the feasibility of the complete design.
- Include feedback loops between the phase 2 and 3 because of alterations in the original design of the PoC.
- Have a clear party or person (product owner) in the lead of the development of the PoC to help guide the process and increase the chance of success of the PoC.
* = Relevant EU regulation might include GDPR, AI Act, Data Governance Act, Data Act, Digital Services Act, eIDAS Regulation, E-Privacy Regulation, Digital Markets Act, Database Directive, Cyber Security Act
Description
Operational topics cover how processes and (centralised) services are operated in the future ecosystem.
Examples of topics are monitoring & reporting, version management and complaint & dispute management.
Approach
Consider the all relevant business, legal and operational topics and determine which are relevant to the specific context of your AI data space and make decisions on how your design will cater to the choices made. The resources provide a comprehensive overview of topics relevant for an AI data space. In implementing the AI data space make and formalise agreements with all involved stakeholders on each topic.
Resources
- DSC Blueprint
- SITRA rulebook
- Relevant AI regulation*
Best practices
- When developing agreements, tools, processes, look for what is already out there in and outside your sector. This ensures you do not ‘reinvent the wheel’ and makes it easier to align with other initiatives in the future.
- Ensure that the interaction model is based on generic roles and not specific actors currently involved in the AI data space. This enables future changes and scalability in that the roles can easily fulfilled by various actors.
- Make the design as generic as possible and as specific as needed. This greatly increases the scalability of the design as it is easier to facilitate other use cases.
- Involve stakeholders from all participating organisations in the decision making process to ensure alignment on the results and improve buy-in from participating organisations.
- Consider all relevant BLOFT (sub) topics, but select and focus on only those topics that are especially relevant to your AI data space.
- Involve the correct experts and prepare analysis before having topic specific discussions to ensure efficient well-informed design decisions.
- Ensure all topics are considered to ensure no decisions are made which may impact future scalability.
- Double check all agreements that have a dependency on another topic to ensure that they do not conflict.
- Make use of existing building blocks from the reference implementations to reduce implementation load and increase efficiencies.
- A test-driven development approach allows design choices to be validated early to ensure the feasibility of the complete design.
- Include feedback loops between the phase 2 and 3 because of alterations in the original design of the PoC.
- Have a clear party or person (product owner) in the lead of the development of the PoC to help guide the process and increase the chance of success of the PoC.
Description
Functional topics describe the functions and services that will be offered to facilitate the goal use case.
Examples of topics are functional components of AI systems and customer control.
Approach
There are different functional implementations in an AI data space. The resources give examples of functional implementations for your AI data space. These implementations could be necessary for your AI applications. Towards a federation of AI data spaces and DSC UCIG give examples of functional implementations. Analyse the reference implementations in the NL AIC GitLab and IDSA GitHub to determine if they can be applied to your situation or make use of existing building blocks to improve interoperability and future scalability.
Resources
- Towards a federation of AI data spaces
- IDS RAM
- DSC UCIG
- SITRA rulebook
- NL AIC GitLab
- IDSA GitHub
- Vantage 6
- Inter data space reference guide
- Intra data space reference guide
- Reference implementation
- Design Principles for Data Spaces – Position Paper
- Go-Fair
- DSC Use case playbook
Best practices
- When developing agreements, tools, processes, look for what is already out there in and outside your sector. This ensures you do not ‘reinvent the wheel’ and makes it easier to align with other initiatives in the future.
- Ensure that the interaction model is based on generic roles and not specific actors currently involved in the AI data space. This enables future changes and scalability in that the roles can easily fulfilled by various actors.
- Make the design as generic as possible and as specific as needed. This greatly increases the scalability of the design as it is easier to facilitate other use cases.
- Involve stakeholders from all participating organisations in the decision making process to ensure alignment on the results and improve buy-in from participating organisations.
- Consider all relevant BLOFT (sub) topics, but select and focus on only those topics that are especially relevant to your AI data space.
- Involve the correct experts and prepare analysis before having topic specific discussions to ensure efficient well-informed design decisions.
- Ensure all topics are considered to ensure no decisions are made which may impact future scalability.
- Double check all agreements that have a dependency on another topic to ensure that they do not conflict.
- Make use of existing building blocks from the reference implementations to reduce implementation load and increase efficiencies.
- A test-driven development approach allows design choices to be validated early to ensure the feasibility of the complete design.
- Include feedback loops between the phase 2 and 3 because of alterations in the original design of the PoC.
- Have a clear party or person (product owner) in the lead of the development of the PoC to help guide the process and increase the chance of success of the PoC.
Description
Technical topics describe the technical requirements to provide and control the functional components.
Examples of topics are protocols/standards, message formats, audit trails, the use of Privacy Enhancing Technologies, semantic agreements and AI technologies.
Approach
Designing your technical implementation based on existing architecture and reference implementations allows for an efficient and scalable implementation. Therefore analyse existing resources and determine whether they can be re-used or (partially) applied to your AI data space context. Build, test and validate the technical specifications and iterate on design based on practical learnings.
Resources
- Towards a federation of AI data spaces
- IDS RAM
- DSC UCIG
- SITRA rulebook
- NL AIC GitLab
- IDSA GitHub
- Vantage 6
- Inter data space reference guide
- Intra data space reference guide
- Reference implementation
- GAIA-X: Technical Architecture
Best practices
- When developing agreements, tools, processes, look for what is already out there in and outside your sector. This ensures you do not ‘reinvent the wheel’ and makes it easier to align with other initiatives in the future.
- Ensure that the interaction model is based on generic roles and not specific actors currently involved in the AI data space. This enables future changes and scalability in that the roles can easily fulfilled by various actors.
- Make the design as generic as possible and as specific as needed. This greatly increases the scalability of the design as it is easier to facilitate other use cases.
- Involve stakeholders from all participating organisations in the decision making process to ensure alignment on the results and improve buy-in from participating organisations.
- Consider all relevant BLOFT (sub) topics, but select and focus on only those topics that are especially relevant to your AI data space.
- Involve the correct experts and prepare analysis before having topic specific discussions to ensure efficient well-informed design decisions.
- Ensure all topics are considered to ensure no decisions are made which may impact future scalability.
- Double check all agreements that have a dependency on another topic to ensure that they do not conflict.
- Make use of existing building blocks from the reference implementations to reduce implementation load and increase efficiencies.
- A test-driven development approach allows design choices to be validated early to ensure the feasibility of the complete design.
- Include feedback loops between the phase 2 and 3 because of alterations in the original design of the PoC.
- Have a clear party or person (product owner) in the lead of the development of the PoC to help guide the process and increase the chance of success of the PoC.
Check out related data sharing downloads
Share with
