Our Approach

Yoyo's approach to AI

A coral dot with the word AI in the centre on a black circuit board

Our Yoyo mission is to help Good Brands succeed, so that they can deliver an even greater impact to society and the planet. As a B Corp, Yoyo is committed to ethical business practices and our values remain consistent - people, community and the environment will always be central to our decision-making.

AI will help Yoyo to achieve our goals through:

Increasing the value that we give to our clients through greater insight and additional time created for higher impact activities.
Decreasing time spent on repetitive tasks. This allows us to focus on more impactful and meaningful activities and/or gives us more space to strategise and be creative.

1. Be thoughtful: Every AI-related decision should be grounded in our agency’s strategy, purpose, ethics, and values. Always consider whether AI is the right tool for the task. Be intentional in how you experiment - whether starting with human thinking or exploring an AI-led approach.
2. Human first: AI is only as good as the people using it. These tools augment human creativity and strategic thinking, they shouldn’t replace them. We value creativity, vision, innovation, and sound judgment. Ultimately, people remain accountable for the work we deliver.
3. Continuous learning: The AI landscape is evolving rapidly. We are committed to building a culture of ongoing learning, experimentation, and adaptation. We encourage everyone to stay curious and open to new tools and techniques and the ways in which they can best be utilised to bring the greatest value to our team and our clients.
4. Embrace the tech: AI tools are already being used across all departments and roles. This will only increase. Every team member has a role to play in embracing the technology. We share our progress and celebrate AI successes together.
5. Trust is everything: Trust takes time to build and can be lost quickly. If our work loses credibility through poor use of AI, that trust can be damaged. Our processes must be designed to protect the quality and integrity of our work at all times.
6. Quality and creativity first: Speed is not a substitute for standards. Every piece of work must include human input to ensure it meets our expectations for quality, originality, and creativity. AI is not always accurate. Always verify the sources.
7. Be transparent and ethical: We must always strive to minimise bias and act with integrity. Data privacy and security are non-negotiable. We will always be clear with clients about when and how AI is used in their projects.

Transparency with Clients

Why it matters:

Our clients have a right to know when and how AI is used in the creation or delivery of their work.

How we do this:

Disclose AI use in contracts, SoWs, or onboarding docs
Send updates & host sessions with our clients to discuss AI usage
Our Yoyo approach to AI visible to everyone

Security, Data Privacy and Confidentiality

Why it matters:

To protect the privacy of our customers and our business, client or internal data must not be uploaded or shared through AI tools that retain or train on user inputs. This includes data related to customers, employees or partners.

How we do this:

Use closed, private systems, when using company and client materials. Internally we use Gemini, as part of our Google Business Workspace. Gemini does not share data outside of Yoyo, it brings the same enterprise-grade security as the rest of the Google Workspace.
Use third-party tools (ChatGPT, Perplexity, Claude, etc.) if preferred for non confidential research (e.g. anonymous, with no actual data) but opt out of training in the settings.
Use reputable tools that have been developed by reputable companies. Research must be conducted about the history, the reputation and the customer reviews on every AI tool used. Employees must evaluate the security of any AI tool before using it. If there is any doubt, discuss with Yoyo’s Technical Director.
Apply the same security best practices for all company and customer data. This includes strong passwords, keeping software up-to-date and following other internal policies around data and security.
Regular education on data privacy across the team, as well as during onboarding.

Intellectual Property and Licensing

Why it matters:

AI tools that generate content, code, or imagery may raise questions about copyright ownership.

How we do this:

Verify and document the licensing terms of all AI tools before use, ensuring outputs are permitted for commercial purposes.
Avoid using AI-generated content (especially images or code snippets) in client-facing work without review and documentation of sources.
Disclose all AI generated content to clients.
Clarify that AI-generated work may not always be eligible for copyright protection in certain jurisdictions.

Accountability and Attribution

Why it matters:

AI is a tool, not a creator. The responsibility for decisions and outputs lies with humans. This applies to internal work and the work that we produce for our clients.

How we do this:

Understand the steps AI has used for its analysis, insight or decision-making. Check sources, use prompts to dig deeper and in some cases, use different Large Language Models (LLMs) to check work.
Ensure human oversight, sign-off and accountability at each stage, especially in client deliverables, legal docs, and proposals.

Tool Overreliance and De-skilling

Why it matters:

Excessive use of AI may reduce critical thinking, creativity, and skill development.

How we do this:

Encourage AI as a co-pilot, not an autopilot. Human reasoning must lead decision-making.
Experiment with doing some tasks with AI, and some without, so that we are continuing our curiosity and learning around the opportunities and drawbacks of using the tools.
Promote internal training that upskills team members in areas beyond AI. Continue attending conferences and events that showcase creativity and innovation.
Integrate reflection questions into project retros, like ‘What did we learn by using AI on this?’ and ‘How?’.

Accuracy and Verification

Why it matters:

AI is known to "hallucinate" facts, misinterpret nuance, or oversimplify complex topics.

How we do this:

Establish a mandatory review step for any AI-generated content before it's sent to clients or published.
Build a checklist or prompt repository that includes: "Have you fact-checked the claims?", "Do the numbers have cited sources?", etc.
Make it the norm to always verify stats from tools like Perplexity.

Inclusive AI Adoption

Why it matters:

AI can be empowering, but not everyone adopts it at the same pace or comfort level. We have always held ‘One team, One Yoyo’ and this applies to integrating AI - it’s important that everyone is supported during this process of change.

How we do this:

Provide everyone time for AI experimentation and provide accessible training.
Celebrate small AI wins (via newsletter or Slack) to demystify adoption and reduce resistance.
Provide 1:1 support for less confident users: AI taskforce and AI champions are assigned.

Bias and Fairness

Why it matters:

AI tools may reinforce societal, cultural, or algorithmic biases - especially in copy, imagery, hiring language, or persona creation.

How we do this:

Encourage bias checks for all AI-generated outputs, especially content related to people, demographics, or industries.
Build checks into content review processes

Are we representing diverse perspectives?
Does this language unintentionally stereotype?

Test prompts against varied inputs to ensure consistency and fairness.

Ethics Reviews and Governance

Why it matters:

Having a framework for ongoing review builds trust, especially as tools evolve.

How we do this:

Use the existing AI taskforce to hold regular “ethics check-ins” or tool audits
Keep a register of approved tools, their use cases, and any known risks/limitations
Create a simple AI Ethics Checklist that can be applied at the start of any major project - see below.

1. Is it fair?: Are the sources diverse and inclusive?

Could the output reinforce harmful bias or stereotypes?
2. Is it clear?: Can you explain how the AI reached this result?

Would this be understandable to someone unfamiliar with the tool?
3. Is someone responsible?: Who is accountable for reviewing and signing off the output?

Have you taken ownership of the result, not just the tool?
4. Does it respect privacy?: Are you using confidential or client data responsibly?

Is this being processed within a secure or approved system?

Below is an insight into how AI is used across the organisation. Other tools are being tested but below are the key ones that are used across the agency. This is updated every quarter.

Sales & Marketing: We use AI to accelerate content creation, improve new business research, and efficiently manage client communications.

Tools & Use Cases:

Jasper.ai - trained on Yoyo's brand guidelines and is used to support Yoyo's marketing content creation.

ChatGPT / Gemini - researching a company when developing a proposal.

Fireflies - recording all new biz calls. Captures details, summarises calls and is used to quickly bring other team members up to speed.

Gemini - used on briefs and documents to quickly locate information and to summarise documents. This is a closed system - information remains confidential.
Operations: We use AI to ensure the accuracy of legal and contractual documents, and to develop and refine internal processes.

Tools & Use Cases:

ChatGPT / Gemini / Claude - checking clauses in contracts and Statements of Work (SoWs), defining new processes, creating agendas, and sense-checking proposals. Gemini is used for confidential information.

Built-in AI in Scoro (Yoyo project management system) - assistance with how a process or feature should work.
Client Services & Project Management: We use AI to help streamline document creation, client communication, notetaking and research wherever possible.

Tools & Use Cases:

ChatGPT / Gemini - improve client communication, summarise large email chains, creating quote breakdown descriptions for Scoro, build project briefs from templates and onboarding info, summarise whole Figjam boards after workshops.

Copilot - creating Excel documents & converting Project Framework Agreements into spreadsheet format for the development team estimates and Jira tickets.

Fireflies - meeting notes for most client calls.

Claude & Google LLM Notebook - useful for client insight and discovery phase research and helpful for creating quick overviews when time is limited.
Design: We use AI to research and ideate, acting as a creative partner to support our work in content, design strategy and media creation.

Tools & Use Cases:

Claude / ChatGPT / Perplexity - used for initial content generation, design strategy, and research support.

Figma - supports ideation, wireframing, and collaborative design creation. Used to rapidly prototype ideas before spending time and resource refining them.

Photoshop - applied with the image editing, adjustment, and enhancements tooling within the program.

Freepik - source for AI-generated imagery, video, and audio assets.
Development: We use AI to accelerate coding, simplify complex logic, and improve documentation. AI acts as a pair programmer for debugging and research.

Tools & Use Cases:

Gemini - debugging, interpreting error messages, and researching technology options.

VS Code: Copilot & Cursor - coding support, refactoring, and suggesting implementations. Simplifying and improving performance of complicated logic, generating user and developer documentation, and mapping out code approaches.

Cursor: ChatGPT - fact-checking code generated by Cursor, simplifying complex coding scenarios for non-technical managers, and summarising time entries from Scoro.
Finance: We use AI to primarily research and clarify of complex information.

Tools & Use Cases:

ChatGPT - mainly for explanations/breakdown of complex legal/financial information.

Gemini - searching historic data and emails on the Yoyo Google workspace.

Yoyo's approach to AI

Contents