Introducing Building With AI, a 10-part AI Six podcast series
People know what you do by the work they ask you to do.
My business is built on inbound enquiries, repeat business and people discovering me through the authority content I share on my podcast, YouTube and LinkedIn. That content also helps me appear in Google and in the answer engines people increasingly use to find expertise.
When I looked at the enquiries coming into my business, I noticed something. People were asking me to train their teams in artificial intelligence, advise them on AI and speak at their events. But they were not asking me about building with AI.
That made me stop and wonder whether people knew what had been happening behind the scenes.
For the past year, my team and I have been building seriously with AI. We have developed custom assistants, voice agents, agentic workflows, apps, governance tools, local AI systems and a domain expert model for public-sector communications.
I still train people in AI, and that work matters enormously to me. But I do not simply teach people how to use tools that other organisations have built. I am an applied AI strategist and builder. My training is stronger because it is informed by what I build, test, improve and sometimes break inside my own business.
That is why I am launching Building With AI, a special 10-part series of the AI Six podcast, running every weekday from Monday 10 August to Friday 21 August 2026.
What building with AI means to me
Building with AI is not the same as opening a chatbot and asking it a question.
For me, it begins with a real problem. I decide who experiences that problem, what a useful outcome should look like and where human judgement must remain. I organise the knowledge, design the workflow, choose the tools, test the result and keep iterating.
Sometimes I build the complete product through natural language and vibe coding. Sometimes I work with my AI developer. My role is not to claim that I manually write every line of code. My role is to understand the problem, shape the solution, provide the domain expertise and decide whether the result is good enough to use.
I often begin inside my own business because it gives me the freedom to experiment. I can build something, leave it, return with a clearer perspective and improve it before asking a client to trust it.
When I build an MVP for a client problem, I will often give clients access for free. Then I watch where the value actually appears. Do they return to it? Does it save time? Does it improve a decision? Is it valuable enough to become a paid product, or does it work better as an added benefit for an existing client?
I do not believe every experiment needs to become a business. Testing with real users tells me far more than pretending I already know the commercial answer.
The 10 AI builds and lessons learned
Each episode of this series takes one part of my building journey and explains the problem, the technology, the decisions and the lesson.
1. I Don’t Just Teach AI, I Build With It
I begin with the reason for this series and the gap I noticed between how the market saw me and the work I was actually doing. I also explain why I believe AI is communication, not coding, and why people with deep domain expertise can now build solutions that were previously beyond their reach.
2. What It Really Means to Build a Custom AI Assistant
I use Claude Projects and custom GPTs to create assistants around defined roles, knowledge and standards. A useful assistant is not simply a general model with a folder of documents. It needs a purpose, boundaries, trusted material and a clear definition of what good looks like.
3. I Built a Voice Agent. Here’s What I Learned
AIJ is my voice agent, built using ElevenLabs and grounded in my own course content. She can retrieve information and answer learners’ questions across courses including AI for Professionals, Social Media for Professionals and Digital Communications for Professionals.
The build has also taught me that cloning a voice is not only about pitch and pace. Accent, identity, accuracy, disclosure and trust all matter.
4. How I Built an Agentic Workflow for My Podcast
AI Six now has an operational workflow built around Google Sheets, n8n and Claude Opus. I add the story, sources and my point of view. The workflow prepares a draft, returns it for review and keeps me in control of the editorial decision.
It saves around one hour per episode without outsourcing the judgement that makes the podcast mine.
5. AI Rulebook: Building an EU AI Act Governance Tool
AI Rulebook is a live tool I built with Lovable. It generates a draft AI policy, assesses an organisation’s AI readiness and recommends governance measures that need to be introduced or strengthened.
It turns a complex compliance and governance challenge into a practical starting point, while making clear that an automated output is not a substitute for professional legal advice.
6. Building an AI Tool to Monitor Misinformation and Disinformation
I have built an MVP for government and public-sector communications teams that need to understand misleading claims and narratives around policies, services, legislation and consultations.
Clients can sign in, search a topic and receive a report to support human assessment and evidence-based communication. Its working name was Civic Signal, but I am changing it because a product name should help people understand what the product actually does.
The tool will now be tested free by clients. Their behaviour and feedback will tell me whether its value belongs in a standalone product, a wider service or the GovComms.ai ecosystem.
7. The AI ROI Calculator: Is AI Actually Delivering Value?
The AI ROI Calculator at airoi.digitaltraining.ie is inspired by OpenAI’s scorecard for the AI age.
It assesses one workflow at a time through useful work, the full cost of a successful task, dependability and value at scale. It estimates monthly ROI, net value, value per euro, payback, human hours returned and the cost of successful AI-supported work.
It is an independent tool, not an official OpenAI product. Its purpose is to move organisations beyond counting licences and logins towards measuring outcomes they can defend.
8. Voice Studio: Building My Local AI Operating System
Voice Studio is the most ambitious internal system I am currently building. It is a voice-first operating system for my business, grounded in my corpus of knowledge.
The system uses Obsidian, Markdown files, RAG and a locally hosted Qwen model. It separates confidential client knowledge from my creative business knowledge and is being designed to connect with other systems in my company.
This build has also led me towards smaller, locally hosted models. Small can be powerful, particularly when a model is given trusted knowledge, a defined role and proper governance.
9. GovComms.ai: Building a Domain Expert AI System
GovComms.ai is my flagship solution. It brings together 25 years of my domain expertise and almost a year of sustained research, technical specification, testing and iteration.
Its first iteration was PolBot, which I developed and presented during my artificial intelligence and public policy programme at Harvard in 2025. Today, GovComms.ai is a working MVP built for communications professionals and leaders across government and the public sector in the European Union.
It uses a Mistral model, fine-tuning and RAG. It supports six communications functions and 18 essential tasks, while allowing an organisation to ground the system in its own locked-down, approved material.
I also built the first GovComms.ai landing page with Lovable, using voice, vibe coding and repeated iteration to move quickly from an idea to something people could see and test.
The ambition is clear: trusted, governance-first AI for public-sector communications. AI for Public Good.
10. What Building With AI Has Taught Me About the Future of Work
The final episode brings the lessons together. Building with AI has taught me that I can scale my ideas, intellectual property and capacity, but only when human judgement remains present from the beginning.
AI is most powerful when we stay in our domain lane. My work is grounded in communications, journalism, public policy, education and 25 years of experience. AI helps me activate that expertise. It does not supply the professional conscience or responsibility behind it.
Why I am sharing the work now
I have spent years creating authority content about communications, marketing, public policy and artificial intelligence. But authority is shaped by what we choose to show.
If I only talk publicly about training, people will understandably think training is all I do.
This series is my way of opening the door to the AI ideas lab inside my business. I want to show the working products, the unfinished systems, the strategic decisions, the failed iterations and the governance questions.
I also want to challenge the belief that building with AI belongs only to coders or large technology companies.
Technical expertise matters, and I value the developers I work with. But the starting point for a valuable build is often the ability to understand a problem, communicate an outcome, structure knowledge and apply judgement. That is why I say AI is communication, not coding.
Listen to Building With AI
Building With AI begins on Monday 10 August 2026 and continues every weekday until Friday 21 August.
Each episode is approximately six minutes long and focuses on one build, one problem and one practical lesson.
Listen and subscribe to AI Six podcast
I have spent the past two years building with AI. Now I am ready to show you what I have built.
If you want to build your AI literacy and skills, I can help
📚 Grab AI Playbook on Amazon, the perfect podcast companion.
📰 Subscribe to Human in the Loop, my LinkedIn Newsletter
🎓 Take the CPD Accredited AI Playbook Diploma
📞 Book a call at digitaltraining.ie
About the Podcast Host
AI and Communications Specialist | Author | Podcaster | International Speaker
Joanne Sweeney is a recognised leader in digital marketing and digital transformation, specialising in the practical application of AI to enhance communication strategies. The founder of Digital Training Institute, Joanne, empowers leaders and communications professionals to leverage AI responsibly and effectively, ensuring transparency, trust, and tangible results.
She completed the Harvard Kennedy School Executive Programme in AI and Public Policy this summer before launching her fourth book AI Playbook, in September. Joanne has also published two editions of Public Sector Marketing Pro, which were preceded by her first title, Social Media Under Investigation Law Enforcement and the Social Web.
Her books have become go-to resources for governments, public sector and policing agencies, as well as marketing and communications professionals, all navigating the challenges of the changing Internet.
Joanne is the host of the AI SIX Podcast, a daily six-minute podcast that provides actionable insights, expert interviews, and practical advice to help professionals navigate the new world of work.
