)
)
Ensures that your AI initiatives are aligned with the company’s broader goals, combining automation, efficiency, and customer satisfaction for a balanced outcome.
CDI Standards Framework
The CDI Standards Framework is a comprehensive, technology-agnostic approach that defines what it takes to get AI conversations right.
Watch video
CDI Standards Framework
Unlocking value in Conversational AI
Drawing on years of experience with Fortune 500 companies, the framework captures the patterns and practices of high-performing organisations, ensuring they achieve higher automation rates, better customer satisfaction, and lower costs to serve. This framework systemises best practices to drive enterprise success by aligning strategic, operational, and technical components across your AI programs.
CDI Standards Framework is a systematic approach to developing a conversational AI capability by ensuring Mindset, Skillset, Culture, and Systems are all developed in the right pace.
A workflow, not a one-off The framework turns assessment into a cycle. We measure your AI against the standards to set a baseline, fix what matters most, then re-assess to prove the score moved, and repeat. Standards don't just tell you where you stand once: they give you a way to keep improving against a fixed definition of good.
Six domains, thirty-one standards
Getting AI conversations right takes more than good design. The CDI Standards Framework comprises 31 standards across six domains, each assessed through four lenses: mindset, skillset, culture, and systems.
Meta Strategy
Enterprise mandate and contact strategy. Standards covering human-centered strategy; enterprise AI vision and business case; resourcing and capability planning; operating model and governance; and omnichannel contact strategy and economics.
Without it: AI programmes that stall because no one owns them, funds them, or can say why they exist.
5 Standards
Design
Conversation, tone, prompt, knowledge, multimodal, and safety. Standards covering conversation design; tone, personality and empathy; prompt and output design; knowledge and memory design; multimodal design; and safety UX and disclosures.
Without it: Assistants that technically answer but leave customers frustrated, confused, or routed around.
6 standards
Improve
Analytics, scalability, ethics, continuous learning, and incident response. Standards covering analytics and feedback loops; scalability and change management; ethical and responsible AI; continuous learning and benchmarking; and incident management and corrective action.
Without it: Assistants that launch well and quietly decay because nobody is watching the numbers.
5 standards
Operational Strategy
Programme vision, scoping, and guardrails: outcome alignment, customer research and scoping, use-case prioritisation, technology and architecture selection, guidelines and guardrails, autonomy and escalation policy for agentic AI, and data stewardship and privacy by design.
Without it: Teams building the wrong use cases, with no guardrails, and finding out too late.
7 standards
Build
Data, training, orchestration, operations, and vendor risk: data and annotation standards, training, testing and optimisation, integration and orchestration, safety and guardrails in operations, LLMOps and release management, and vendor and model risk management.
Without it: Assistants that pass the demo and fail in production: hallucinations, broken handovers, brittle releases.
6 standards
Cross-cutting
Culture, talent, compliance, and risk: culture and talent development, and compliance and risk management, aligned with the EU AI Act, NIST AI RMF, and ISO/IEC frameworks.
Without it: Capability that leaves when people do, and compliance exposure you discover during an audit.
2 standards
A Roadmap to Scalable AI Success
The CDI Standards Framework is more than just a guide—it’s a roadmap to help your enterprise unlock the full potential of conversational AI while maintaining strong ethical standards and creating scalable, sustainable solutions. By adopting this framework, you’re not only investing in technology but also ensuring that your business can deliver exceptional customer experiences through AI at scale.
What You Get from the CDI Standards Framework
Strategic Vision and Business Alignment
Robust Business Plans
Provides clear, actionable plans that integrate financial, technical, and customer experience goals, ensuring every department is on the same page.
Technology and Scalability
Offers guidance on adopting adaptable, future-proof technologies that can grow with your organization’s needs.
People and Culture
Establishes frameworks for hiring, training, and fostering a culture of innovation, collaboration, and continuous learning.
Governance and Risk Management
Implements governance models and quality controls that reduce the risk of compliance, business, or customer-critical errors.
Ethics and Accountability
Provides clear ethical guidelines to ensure your AI is trustworthy, explainable, and free of bias.
Trusted by global brands
Why the CDI Standards Framework Is Essential for Enterprises
Proven Success Patterns
The framework systemises the proven strategies and methodologies observed in the most successful AI programs worldwide.
Holistic Approach
The CDI Standards Framework covers both the AI Assistant itself—how it’s designed, trained, and maintained—and the organizational structure behind it. It touches on all critical aspects: from strategic planning and business alignment to people management and technological integration.
Enterprise-Wide Alignment
By aligning the vision of your AI Assistant with broader business objectives, the framework ensures that automation goals support, rather than compete with, other departments like customer support or IT. This alignment prevents missteps and ensures seamless integration across all business functions.
Deep Expertise
Behind the framework is a vast depth of knowledge and experience, covering everything from developing a robust technical strategy to managing ethical considerations. It provides clear guidance on creating business plans, setting performance metrics, and driving continuous improvement. The framework ensures that your conversational AI meets both automation and customer experience goals simultaneously.
Scalable, Global Solutions
Whether your AI assistant supports a single country or spans multiple languages and regions, the CDI Standards Framework is adaptable to your needs. Its scalable design ensures long-term viability, supporting multi-channel deployments, LLMs and backend integrations.
Ethical and Responsible AI
We focus on responsible AI design, helping enterprises create AI Assistants that are not only effective but also ethical, with systems in place to manage risks, avoid critical errors, and ensure inclusivity.
Find out what’s wrong. Then put it right.
)
)
)
)
Work with us
Our experts help brands design, build, and maintain AI assistants that work. If you want to hit the ground running, or need help scaling your team, get in touch.
Related services
Assessments
)
Find out what's wrong as the first step to putting it right. We assess your team, technology, and organisation against the CDI Standards Framework.
Consulting
)
Put it right, with our consultants alongside your team. Project-based or ongoing, on the programme you already have running.
Team training and certification
)
Build the capability to keep it right. Certified training for the people who design, build, and run your AI conversations.
Discover our case studies
Assessing, improving and implementing the building process for a cosmetics chatbot
Conversation Design Institute (CDI) was asked to perform a fast track audit. This improved the inter…
Honeybook explores Agentic Experience Design
CDI enjoyed hosting a group of talented content and design experts from Honeybook for a one day work…
Boosting Voice AI Performance for Deutsche Telekom
CDI equipped Deutsche Telekom's mixed team of engineers, business stakeholders, and designers with t…
Leading Global HR Services Provider
Conversation Design Institute (CDI) was asked to help redesign the chatbot of a global leader in wor…
IND embraces Conversational AI as a channel
The IND is the Immigration and Naturalisation Service (Immigratie- en Naturalisatiedienst). Their 'c…
Global Sports Retail Chatbot
A Forbes 500 sports retail brand launched their customer service chatbot in the US market in 2020. I…
Migrating Hansgrohe's Chatbot, Hans
The Hansgrohe Group, a leading German sanitation company, is known for its premium brands Hansgrohe…
Migrating Trulet’s Chatbot, Trudi
Trulet, an Australia-based company, created Trudi—an innovative AI chatbot for real estate that enha…
Expanding multilingual IVR support
A leading global provider of cutting-edge software and services, embarked on a transformative journe…
Goodcall’s AI Phone Assistants
Goodcall is a California-based startup that specializes in delivering AI-powered phone assistants th…
Healthcare HR Chatbot
Over the course of two years, CDI worked with one of the world’s leading pharmaceutical companies to…
Vandebron’s Chatbot, Mick
Conversation Design Institute worked with Dutch energy company Vandebron to build and launch their c…
Vodafone’s Virtual Assistant, TOBi
Vodafone Group Plc, Europe’s largest Mobile and Fixed Network Operator, serves over 300 million mobi…
HP’s Virtual Assistant
HP Inc. is a technology company that believes one thoughtful idea has the power to change the world.…
FAQ
What is the CDI Standards Framework?
A technology-agnostic definition of what it takes to get AI conversations right: 31 standards across six domains, each assessed through four lenses; mindset, skillset, culture, and systems.
What do we get at the end of a CDI Standards assessment?
A scored read against the standards: what is working, what is broken, and what to do next. Scoring is evidence-based, so you get the reasoning behind each score rather than a number on its own.
Does it assess the assistant, or the organisation behind it?
Both. Design, Build and Improve cover the assistant; Meta Strategy, Operational Strategy and Cross-Cutting cover the mandate, funding, operating model, talent and governance around it. Assistants usually fail for reasons that sit in the second group.
Do we have to be assessed against all 31 standards?
No. Our Light Assessment reads across the priority standards; Focused scores the domains that matter most to you in full, with evidence; A Comprehensive Assessment scores every standard and adds a prioritised roadmap.
Is the framework tied to a particular vendor or platform?
No; it is technology-agnostic by design. It describes practices and outcomes, not products, which is why it survives a platform migration. Technology and architecture selection is itself one of the standards, in Operational Strategy.
How is this different from the EU AI Act or NIST AI RMF?
Those tell you what you must not get wrong; the framework tells you what good looks like. Compliance and risk sit in the cross-cutting domain and align to the EU AI Act, NIST AI RMF and ISO/IEC; the other five domains cover the strategy, design, build and improvement work that regulation does not describe.
