Redesign how you work with AI while strengthening your skills, judgement and confidence
If you’re already using AI for individual tasks, this course will help you change the way you work with it. You will move beyond asking AI for one-off outputs and start designing how people and AI work together through structured workflows.
You will remain actively involved in interpreting information, making decisions, challenging outputs and deciding what good work looks like. The aim is not simply to use more AI, but to use it in ways that strengthen your skills, judgement and confidence.
Across four practical sessions, we’ll take a marketing campaign from source material through to messaging across channels. This shared example will enable you to follow a workflow and develop it into a repeatable AI-supported workflow. You’ll learn how to connect different stages, give AI the right context, build in checks, test the quality of the results and explore how automations and agents could take the workflow further.
Taking the course
Responsible AI Workflows for Cultural Marketing comprises of four weekly two-hour live online sessions on Zoom.
You will not just learn about workflows and agents, but will also build and test something you can use. Accessibility, inclusion, privacy, copyright and environmental impact will be built into the process, helping you create a workflow that is reliable and responsible. You will leave with a documented workflow and practical plan that you can introduce to your team.
If you have any access requirements, please contact sophie@a-m-a.co.uk
Session dates
- Session 1: From AI tasks to responsible workflows
10am-12pm on Wednesday 20 January 2027 on Zoom - Session 2: Building an AI workflow you can defend
10am-12pm on Wednesday 27 January 2027 on Zoom - Session 3: Testing and improving workflows
10am-12pm on Wednesday 3 February 2027 on Zoom - Session 4: From workflows to systems
10am-12pm on Wednesday 10 February 2027 on Zoom
Why do the Responsible AI Workflows for Cultural Marketing course?
Build something real
Build and test a complete responsible AI-supported workflow. You’ll leave with reusable templates and a practical implementation plan for applying the method to your own work.
Change how you work with AI
Create structured workflows that keep you involved in the thinking while strengthening your skills and confidence.
Learn with arts and culture sector peers
Develop and share workflows with your colleagues across the sector, gaining fresh ideas and valuable connections.
Who is it for?
This course is for people whose work involves marketing, communications, audience development and visitor engagement across arts, cultural and heritage organisations.
You will have some experience of using AI in tools such as ChatGPT, Copilot, Claude or Gemini. You don’t need technical or coding skills, but you would like to move beyond one-off prompts and explore a more structured and reliable way of working with AI.
Cost
Standard Member rate: £250 + VAT
Small Organisation / Between Jobs / Freelance Member rate: £150 + VAT
Non-member rate: £395 + VAT
Book now
Session schedule
Session 1: From AI tasks to responsible workflows
10am-12pm on Wednesday 20 January 2027 on Zoom
This session will examine different parts of the shared campaign process and help create a responsible workflow map.
It begins by looking at what we mean by an AI workflow and how this differs from using chat, projects, research tools, automations and agents. Through short examples, you will explore how AI can support connected stages of work rather than simply completing individual tasks.
Working with the same realistic process, you’ll begin mapping what information is needed, what happens at each stage and where decisions and responsibilities sit.
This session covers:
- The differences between AI tasks, projects, workflows, automations and agents
- Moving beyond individual prompts towards connected, repeatable processes
- Examples of workflows across arts marketing
- Identifying repetition, friction, information handovers and judgement
- Deciding whether a task is suitable for AI support
- Mapping inputs, stages, decisions, people and outputs
- Deciding what AI could support and what should remain with us
- Identifying initial risks, checks and approval points
- Choosing an appropriate way of working
By the end of this session you will have: a shared responsible workflow map and a practical method you can apply to processes in your own work.
Session 2: Building an AI workflow you can defend
10am-12pm on Wednesday 27 January 2027 on Zoom
Turn the shared process map into a responsible AI-supported workflow that is ready to test.
This session will examine different stages of the workflow, including what information each stage needs, what it should produce and where our judgement and checks should sit. We’ll then bring these stages together into one complete specification.
You will turn the campaign process map from Session 1 into a clear workflow specification that someone else could understand, follow and question. We’ll then build an initial prototype ready to test in Session 3.
This session covers:
- Defining the purpose of the workflow and what a good outcome should look like
- Deciding what should stay, change or be removed before adding AI
- Breaking the work into clear stages and information handovers
- Assigning work between people, AI and existing systems
- Identifying the context, evidence and source material each stage needs
- Separating facts, evidence, assumptions, unknowns and red lines
- Keeping people actively involved in judgement, decisions, approvals and quality checks
- Building in privacy, security, data minimisation, copyright, bias, accessibility and environmental considerations
- Recording key decisions and risks so that we can explain how the workflow works and answer the question: “Can we trust this?”
- Creating an initial prototype that can be run and tested
By the end of this session you will have: a responsible specification and initial prototype for the shared campaign workflow, plus a reusable structure you can apply to workflows in your own work.
Session 3: Testing and improving workflows
10am-12pm on Wednesday 3 February 2027 on Zoom
Run, challenge and improve the shared campaign workflow.
Building a workflow is only the beginning. Before you try to make it agentic, you need to find out whether it works, whether information moves accurately between stages and whether it genuinely saves time or improves quality.
This session will work through the complete campaign workflow. Starting with a realistic source pack, we will use AI to develop a campaign brief, audience approach, message pillars and channel plan.
We will run the prototype developed in Session 2, introduce some realistic problems including unsupported claims, overlooked contradictions and bias, and examine how those problems travel through the workflow. This will show how a reliable workflow is built through testing, judgement and iteration.
This session covers:
- Running a workflow manually before trying to make it agentic
- Using projects, files and reusable instructions
- Checking whether information survives accurately between stages
- Identifying weak instructions, missing context and duplicated work
- Building in approvals, checkpoints, stopping points and escalation
- Testing normal cases, edge cases and failure cases
- Comparing outputs produced through different ways of working
- Deciding whether the workflow genuinely saves time, improves quality or reduces risk
By the end of this session you will have: a tested and improved AI-supported campaign workflow and validation record.
Session 4: From workflows to systems
10am-12pm on Wednesday 10 February 2027 on Zoom
Turn a tested workflow into an agent-ready system and plan how it could be introduced responsibly.
In this session we will return to the campaign workflow tested in Session 3 and explore how several stages could be carried out by an agent. We’ll decide what information and tools it needs, where people remain involved and what should happen when it encounters a problem.
Together, we’ll turn the campaign workflow into an agent-ready specification and consider how the same approach could be applied within your own organisation. You’ll also develop a practical plan for testing, measuring and introducing an AI-supported workflow without making claims that have not yet been demonstrated.
We’ll explore what changes when AI moves from responding to individual prompts to working towards a goal across instructions, files, tools and outputs. We’ll also consider what needs to be in place before an agentic workflow can be used by a team.
This session covers:
- What an AI agent is and what makes a workflow agent-ready
- Understanding agent modes, goals, loops and reusable instructions or skills
- How agents work across instructions, files, tools and outputs
- The difference between supervised AI work and more autonomous execution
- Deciding what an agent can access, change or produce
- Setting clear permissions, boundaries and approval points
- Handling missing, sensitive or contradictory information
- Defining when an agent should pause, stop or escalate
- Monitoring outputs and keeping appropriate records
- Maintaining accountability for decisions and outcomes
- Identifying measurable benefits
- Introducing an AI-supported workflow to your team
By the end of this session you will have: an agent-ready workflow specification, a method you can apply to your own work and a practical implementation plan.
By the end of the course, you will be able to:
- Tell the difference between chat, workflows, automations and agents – and spot where AI can actually help in your own work
- Map a real process into a responsible AI workflow, with the right context and safeguards
- Build your own judgement, checks and approvals into each stage, so you stay the one making the calls
- Test and judge whether the workflow is accurate, unbiased and worth the time it saves
- Prepare a workflow for agentic use, with clear permissions and boundaries
- Leave with a practical implementation plan that you can introduce to your team
Any questions?
Please contact one of our team at membership@a-m-a.co.uk, they’ll be more than happy to help
