All Collections
AI and Innovation
How to Manage AI Innovation with Ardoq
How to Manage AI Innovation with Ardoq

7 Steps to Managing AI Innovation across your Organization

Simon Field avatar
Written by Simon Field
Updated over a week ago

This article outlines a step-by-step approach for managing and accelerating AI innovation with Ardoq.

With this approach you can:

  1. Model AI innovation and get an map of the technical capabilities and expertise which you can leverage

  2. Develop ideas on how to exploit the power of different AI capabilities in your organization

  3. Use Ardoq’s visualization capabilities to prioritize investments and define a roadmap that aligns to strategic business objectives

AI innovation management is likely to involve the adoption of new technical capabilities that are realized by new technology products, and upgrades to deployed products that have recently been enhanced with AI-powered features.

By following our 7 steps to AI innovation management, you are able to gain a comprehensive view of the opportunities, risks and impacts of your initiatives across affected people, processes, information and technologies in your organization, make well-informed decisions, and accelerate AI efforts from conception to realization.

7 Steps to AI Innovation Management

Here are the 7 steps you should follow to manage AI innovation in your organization:

Step

Ardoq Use Case Solutions

Main Component Types

1. Discover

  • Technical Capability Management & Realization

  • Technology Portfolio Management

  • Application Lifecycle Management

2. Ideate

  • Strategy to Execution

3. Describe

  • Strategy to Execution

4. Prioritize

  • Strategy to Execution

  • Application Risk Management

5. Plan

  • Strategy to Execution

6. Architect

  • Technology Portfolio Management

  • Business Capability Management & Realization

  • Application Portfolio Management

  • Application Hosting

  • Application Integration Management

  • Data Lineage

7. Deploy & Learn

  • Strategy to Execution

  • Application Risk Management

Step 1. Discover

The first step is about understanding the AI expertise you already have in your organization. By mapping AI technical capabilities, and the people with related expertise you can identify key resources that can help you reach your goals. If you don’t yet have a map of technical capabilities, including AI-related ones, make a start now:

  1. Create a Technical Capability Model that classifies and defines the different flavors of AI technology based on the Technical Capability Management and Realization Use Case. You can use the TCM - Technical Capability Detail & Realization Survey (TC->App) to create new Technical Capability components. That survey also invites you to identify experts who have knowledge of, and experience in, these capabilities. It will automatically create the corresponding references between capabilities and people for you. You can also link it to all the Applications that realize the capability.

  2. If you have a map of Technical Capabilities, but no linked Experts, consider creating a simple survey asking respondents to select all AI Technical Capabilities they have expertise in. The audience for the survey will depend on the nature of your business and the size of the organization. It might go out to all department heads, asking them to identify AI experts among their staff. It might go out to IT staff, but you should consider the possibility that AI expertise might exist anywhere in the organization. By distributing the survey, you will be crowdsourcing a map of the expertise in AI that already exists across your organization. This is the network you will leverage to generate and prioritize your AI initiatives, and which you may draw upon to form your core team.

  3. In addition to identifying new technologies and products that you wish to introduce as part of your initiative, you should look more closely at your existing product portfolio. Many commercial off-the-shelf software solutions are embedding AI into their products. Is your organization using them? The AI capabilities may be defined in your model, but you may not be aware of which of your current applications are already realizing those capabilities. And you may not yet have documented new products and what they can do for you. Extend your technology catalog and link it to your Technical Capabilities to build a picture of what you might achieve, and the products that can help you do it (see Technology Portfolio Management Metamodel for details). You can leverage your Application Ownership network and process to capture this knowledge. send the TCM - Technical Capability Realization (App->TC) Survey to all Application Owners asking them to indicate which Technical Capabilities their Application(s) realize, perhaps giving them a reminder that new versions of their applications might realize new AI capabilities.

  4. This updated knowledge may help you spot where you can quickly acquire an AI capability by upgrading to the latest version of a commercial software solution that now has embedded AI. This proposition will still have to compete with other proposed initiatives that might have a better business case.

The areas of expertise and gaps in knowledge that you and your colleagues can reveal with Ardoq will be useful in developing ideas to exploit opportunities and strengthen less mature capabilities.

See also the Process Playbook: How to Automate Application Ownership for more details of how to ensure that your applications always have owners.

Step 2. Ideate

Develop a set of ideas to exploit the power of different AI capabilities in your organization. Use Ardoq’s Strategy to Execution Use Case to document and quantify the ideas, creating a collection of Initiative components. Consider the information you will want to analyze to compare and contrast the different ideas. Add these as fields to the Initiative component type. Candidates include:

  • Development effort

  • Business impact

  • Feasibility

  • Value

  • Implementation Horizon

You might organize a face-to-face brainstorming session, or use an Ardoq survey to solicit ideas from across the organization and automatically generate Initiative components. Bringing together a set of ideas and focusing on value helps to keep resources focused.

“Not all AI use cases are equal. Separate everyday AI from game-changing ones by distinguishing AI that is within boundaries, pushing boundaries and breaking boundaries. Then, identify the key business KPIs from these cases that change management needs to tackle.”

Step 3. Describe

For each Initiative, consider the changes to business capabilities that will occur as a consequence of the inevitable disruption. Introduction of a new technology is not just about a technical change: people and their skills, processes, products, services and information will all need to change if you want to reap its potential rewards. Use the S2E Capability Delta Survey to create and link one or more Capability Delta to each Initiative to help you think through the implications. In addition to the introduction of new technology components, what other applications will have to be changed or integrated? Will there be new processes, or changes to existing ones? Are new skills or organizational units required? What information will be used? Perhaps there will be completely new information sources? Viewing such disruption from a capability perspective helps you to see change from every angle, rather than purely the technology one.

And whilst the focus of attention will inevitably be on direct outcomes (the impact on customers of changed products or services, or the impact on staff of AI assistance in their work), you should also consider the capability deltas needed to support AI: training and education for staff, development of AI policies and compliance processes, changes to data management policies and processes, reviews of and improvements to data quality, legal services and customer contracts.

Step 4. Prioritize

Once you have a set of possible Initiatives, use Ardoq’s visualization capabilities to drive investment decisions, leveraging the descriptive fields you included with each Initiative. Ardoq Bubble Chart Visualization is a good way to bring multiple dimensions together in a single chart:

You need to understand each potential initiative sufficiently to conduct your prioritization exercise: what are the potential gains, costs, barriers, risks. One possible extension of the Strategy to Execution Use Case would be to add a Disruption component type that helps you visualize the disruptions your organization is likely to face in the future. Overlaying the AI Initiative ideas you’ve collected against the disruptions may help you spot ideas that would represent a kind of “pre-emptive strike”:

To consider an end-to-end view of your operating model, analyze your organization’s value streams. Whilst these do not feature yet in Ardoq Use Cases, they are available in the BizBok assets if you are a member of the Business Architecture Guild.

Your thoughts on these considerations from different perspectives may lead you to revise the initial values given to some of the descriptive fields for a given Initiative. Perhaps it is less feasible than originally thought? Or offers greater potential business value?

Many organizations choose to have a multi-stage approach to managing their innovation funnel, funding many initial short experimental initiatives, then providing more development funds to refine only the most promising among them, as more is learnt about the risks, challenges and opportunities presented by each one. Levels of detail and the types of analysis you can conduct at different stages will vary according to your approach.

Your aim is to select a limited number of feasible initiatives with high strategic value that you can pursue with sufficient resources and purpose to see through from conception to realization.

Step 5. Plan

You’ve now got a set of funded initiatives. The others may be reclassified as either Rejected or Pending for further consideration in a later phase. The funded ones need to be fleshed out. Build on your initial understanding of your funded initiatives: having earlier identified the Capability Deltas associated with each of your initiatives, you now need to consider how they will be accomplished. Follow the Strategy to Execution Use Case to tell the whole story of your planned AI journey:

  1. Are these initiatives collectively pursuing a new Strategic Objective, or one that is already in place? Do new OKR’s need to be defined? You may need to add or update Objective and Key Result components using the S2E - Objective Survey or S2E - Key Result Survey in recognition of your strategic initiatives. Many organizations are struggling to show a return on their investments in AI. Your initiatives should be linked to business objectives to ensure that your investments will deliver a business outcome and not just a technology.

  2. Given the amount of change you are planning, consider whether some or all of your initiatives should be decomposed into child initiatives, with their own timelines, impacts and budgets.

  3. Identify which applications, technology services, organization units or other enterprise assets will be impacted by the initiatives, recording the connections with Impacts references.

  4. Record budget, status and timelines against each initiative.

  5. Use your understanding of the initiatives to allocate and record owners.

  6. You can also allocate resources, such as organizational units and people, to initiatives. This can help you confirm that you have the capacity to deliver each initiative on time.

  7. Consider linking the initiatives to a task management tool such as Jira to support detailed tracking of progress.

Step 6. Architect

Each Initiative will have an incremental effect on your enterprise architecture, which you should document using Ardoq Scenarios. Your future-state architecture will need to reflect these changes:

  1. Add newly introduced technologies to your technology catalog with the addition of new Technology Product components (see Technology Portfolio Management Use Case for details);

  2. Show the future state of technical and business capabilities. Some will be completely new, others will have improved levels of maturity. The Capability Delta components you’ve already created will act as your guide.

  3. You are likely to be deploying new technology. This may be adoption of cloud-hosted services, upgrades to existing platforms or applications, or the deployment of new applications, application modules or technology services (or a mix of all of these). Understanding their access to, and use of, data will be critical to demonstrating compliance with your AI and data policies and any corresponding regulatory requirements. New Technical Capabilities will be added, and the maturity of existing Business Capabilities will improve. Your Scenarios should show how these new capabilities will work - how the technologies are linked to data and processes and the people who will operate and consume them, along with information about when they will begin operation.

  4. Beyond a contextual view of how people, processes, technologies and data come together to form new or improved capabilities, you might choose to show in detail your new solution architectures, decomposing applications into application modules and showing the specific technologies they employ and how they connect. See What is an Application? and How to Adopt the C4 model with Ardoq for more information.

  5. AI solutions will need access to data; existing data sources and information systems will require new integrations; planned deployments must cater for essential quality characteristics, such as security, scalability and availability; new organizational units may be created; new processes introduced; new skills learned or new staff recruited. Understanding and protecting the privacy and confidentiality of data is essential, and the architecture will need to show how the solution will comply with your organization’s AI policies and regulatory requirements. In future articles, we plan to take a detailed look at how to define and use Quality Models like ISO 25010, and regulatory requirements such as proposed in the EU’s AI Act. Use the Application Integration Management Use Case and the Data Lineage Use Case to show how data is used by your new Applications and their constituent Data Store and Application Modules.

  6. There is usually more than one possible solution architecture that can address the desired outcomes: consider creating multiple scenarios and conducting an architecture tradeoff analysis to select a preferred approach.

Step 7. Deploy and Learn

It is easy to lose sight of your objectives in the excitement of exploring new technologies. Linking your Ardoq Initiatives to projects and tasks in Jira allows you to monitor the progress of your portfolio of initiatives that make up your programme of work. Also, be aware of the conscious decisions you may be making to cut corners in your implementations in the dash to exploit opportunities and learn as quickly as possible. Undocumented technical debt can quickly become forgotten until it reappears further down the road as a massive liability. Make sure you document technical debt as you acquire it so that it can be remedied when the time is right.

Ardoq dashboards enable you to visualize your progress, providing you with the right data presented in the right way to support course correcting decisions or responses to new situations. Your desired outcomes, and ways of measuring them, should be built into the data you record against Ardoq components. Consider creating a set of Metrics components that can be maintained by all those involved in the programme and develop a programme dashboard. Projected benefits for your initiatives may well prove to have been exaggerated, given the excitement and hype surrounding this emerging technology. Make sure you capture actual benefits once you begin to deploy solutions, so that you can rapidly evolve your evaluation methodologies. These feedback loops need to be planned up front by each initiative so that outcomes can be measured and reported as your AI programme scales and your organization’s experience matures. And your initiatives will need to consider the importance of building mechanisms to continuously gather feedback, learn and refine, and monitor for and mitigate biases in your AI solutions.

Did this answer your question?