AI in business – Where are we now, and what comes next?

In this special episode of the Tech on Toast podcast, we’re live from the Dojo offices with Rob Howes, SVP of Technology at Dojo. Join host Chris Fletcher as he dives into the real-world impact of AI in business – from navigating the hype to unlocking practical use cases. Rob shares how Dojo is building an AI-native future through smart integration, cross-functional buy-in, and a deep focus on solving meaningful problems. Whether you’re just starting out or scaling adoption, this episode is packed with insights for business leaders ready to turn AI curiosity into capability.

Speakers:

• Chris Fletcher, Host, Tech on Toast Podcast
• Rob Howes, SVP of Technology at Dojo

Watch the full episode now:

Podcast synopsis

1. AI adoption: POC to production: Many businesses are experimenting with AI, but scaling from proof of concept to full deployment is a challenge. Rob explains how data governance, ethics, and clear problem statements are key to making AI stick.
2. Automation, augmentation, and amplification: Rob shares a framework for understanding AI’s potential, starting with automation of repetitive tasks, followed by augmenting human capabilities, and ultimately enabling full-scale transformation.
3. Real-world use cases with ROI: From chargeback automation to customer service enhancements, Dojo is seeing real returns. Rob highlights the impact of intelligent agents, fraud prevention, and hyper-personalised customer experiences.
4. Balancing innovation with trust: With examples like Opoly and generative pricing, Rob unpacks the tension between personalisation and transparency, and why explainability is critical for customer trust.
5. How to get started: Rob outlines Dojo’s practical roadmap: start with a mission, identify high-impact problems (not AI solutions), build data literacy, and empower cross-functional teams to test and scale.

Transcript: AI in business – Where are we now, and what comes next? 

Introduction

Chris Fletcher: Welcome to the next episode of the Tech On Toast podcast. It's a special today. We're joined live in Dojo's offices with Rob Howes, Senior Vice President of Technology. 

You've picked my specialist subject – not! But everybody's interested in AI right now, and we're going to have a chat around AI and business. 

AI at Dojo and Rob Howes' background

Tell us a little bit about your role at Dojo and what you do here.

Rob Howes: I've been with the business a little over two years now. As SVP of Technology, I work with and look after the broader engineering and data communities here. I’ve had a career that has been FinTech-focused throughout. Before Dojo, I was in one of Jan and George’s other payment businesses as their CTO. Before that, I spent 15+ years in the capital market space. 

I've always been drawn to industries which have high amounts of data. The opportunity for AI presented itself throughout that journey, and I have fallen in love with technical problems that have a big impact. That’s been a recurring theme throughout and before payments, in that capital market space.  I co-founded a couple of businesses building proprietary trading platforms, so innovating with data has been something that’s been part of my career story so far.

Understanding AI in business

Chris F:  AI in business, particularly in hospitality, has suddenly started to become a talking point over this past year –  maybe not so much action yet. 

Understanding AI in business

What does ‘AI in business’  actually mean?

Rob H: It’s so much more than thinking about computers mimicking humans. Over the last couple of years, there's been a lot of attention focused on that element. It's a spectrum of possibility where we're using this technology in places like Dojo to expand our art of the possible. 

The way we're starting to think about it more at Dojo is across three levels: automation, augmentation, and amplification. 

Automation, augmentation and amplification

Rob H: Automation with AI  has been something Dojo and the industry have been doing for a while, which is looking at repetitive and menial tasks. What's been more exciting recently is augmentation – that's the AI and human combination. 

How do we start to do things faster and better than before? That sits on the in the amplification piece. How does that AI tooling, without human input, start to elevate and reinvent the ways that we do things? At the moment, we're moving from the automation to the augmentation phase – doing things better and at a greater scale than before. But I'm really excited about how we start to eliminate some of those barriers, whether they're technical expertise or processes that get in the way of our people doing their jobs.

Chris F: The ONS reported recently that around one in six UK businesses are now using at least one part of AI technology, and 70% are using it to improve cybersecurity. 

Why do you think uptake is still relatively low?

Rob H: The adoption numbers have probably moved on since that ONS report, which was from a couple of years ago. Whilst adoption of the technology has probably moved in a more positive direction since that report, I’m still seeing a gap between businesses that are still using AI at a proof-of-concept (POC) stage versus those that have been able to convert those POCs into production-grade instances.

Barriers to scaling AI adoption

Rob H: There are a variety of reasons why that uptake is still not as great as we’d like it to be. One is looking at the data quality and governance that businesses have. To utilise this technology to its fullest potential, you need clean, structured data to make it easier for AI to consume that data. That’s still something many businesses aren’t yet set up to do.

You've got ethical and trust concerns, from people being more fearful about things that they don’t understand and worrying about how ethical and trustworthy the data is that’s feeding this. 

If I look at businesses like Dojo, we’re a regulated business. Those regulations and our compliance with them are there to protect people and our customers. Making sure that this new technology is clearly complying with that is also something that can stop many businesses from going beyond the lab. 

There’s also a skill gap. This is relatively new, and there are not many proficient people. There are no real rule books on how to do this at scale – that’s another barrier. One of the biggest things I see tripping up businesses at that POC stage, when they’re trying to go beyond, is what I talked about at the beginning. One narrative from my career has always been trying to fall in love with problems. 

In this case, there are a lot of examples where people have fallen in love with the solution. They're looking for a problem it's going to fix. That means you get these underwhelming moments where POCs don't have the impact that they could or should have. That’s not because the technology doesn't have promise, but because we’ve probably not picked the right use case to pursue.

Chris F: I did a talk the other night with about a hundred restaurant operators and asked them who was using ChatGPT in their business – everybody. If I'd asked them that even 18 months ago, you'd probably have 10%. Deloitte said there were 7 million people in the UK who have used some kind of generative AI at work. 

Bottom-up use vs top-down strategy

Do you think it's a ground-level movement or a top-down strategy coming from the people using it? Are people doing self-discovery with ChatGPT, or is it the other way around?

Rob H: It's a bit of both. Like many trends, you're going to have your portion of early adopters that start to form grassroots movements. They start to increase pressure from top-level leadership to be more intentional about what their strategy should or could be. It’s almost like a feedback loop. 

Here at Dojo, there’s a combination of both. Tools like OpenAI, ChatGPT, with their incredibly friendly user interfaces, are much more ubiquitous now. It's also quite easy to get going with, which has helped the grassroots adoption become quite widespread. I'm not surprised to hear that you're getting that kind of response from it. 

What we're doing here is both. From the top, we have senior leadership buy-in that this is something we must get behind and are investing in. We're also tapping into those grassroots communities, like an AI champions group. How do we tap into that to make sure that our communication is more effective? Ensuring that any gaps, skills, and feedback are addressed in a tight manner, leveraging that top-down endorsement with that bottom-up evangelism.

Chris F: It's down to solution versus problem. People are trying to find the problem, not have the problem first. 

Common AI use cases in business

What are the common use cases when it comes to implementing AI or solving problems with it? Can you give us some examples of what's happening?

Rob H: There are three core dimensions that people are pursuing with this. You have an efficiency gain – how can we save money? How could we be more cost-effective, save time? Free up people's time? 

You’ve got effectiveness – how can we use the technology to help people do a better job? Not just faster, but get more accurate results and have a bigger impact. The third one would be looking at the innovation. Can we come up with new products or new services that we can offer or sell as a business? That last one is happening, but on a smaller scale. We're seeing quite a lot of success in the first two camps – efficiency and effectiveness. 

There’s now an increasingly wide variety of industries that are using this technology, with a few established examples. Customer service is a huge one for things like AI chatbots – being able to do things like intelligent routing of queries and calls for the right department. Sentiment analysis can enrich the customer service agents’ experience by providing insight into the quality of support they’re giving. 

You've also got opportunities in marketing and sales. How can we better predict lead scoring, for example, particularly looking at personalised content generation? How can we tailor campaigns and optimise ad campaigns to reach the right audience and use relative language? 

Operations functions in general. If I look at Dojo, this can range from our onboarding to our KYB process. How do we streamline that? How do we make it more effective as a regulated business? There are scheme and regulatory compliance use cases. There are a lot of procedural operational divisions within a company that lend themselves well to this.

Cybersecurity is another one, whether that's fraud prevention, threat detection, or anomaly detection. A big one is software engineering, and I'm not talking about vibe coding – that's almost a forbidden term here. It's great for rapid prototyping, but it's not how you build engineering solutions at scale. It’s looking at how we complement our engineering force with this kind of tooling. We talk about AI native engineering – from the code it generates, the test cases that it produces, to the documentation that can accompany that code. Those are some pretty well-established areas where we're finding this technology being used successfully and at scale.

Chris Fletcher: Consistency in operations, especially in hospitality, has been hard to achieve for many years. 

Improving consistency and plugging skill gaps

Do you think unlocking AI in that field in areas like payments and PCI, will support these industries going forward? 

Rob H: It can create a level playing field. How do we use this technology to better eliminate the barriers that get in the way of our people? A lot of what you've described are domains of expertise, but they're quite established – there's a lot of documentation and material on them. All of that feeds incredibly well into the world of AI tooling. How do we augment those kinds of industries and skill sets without having to become PCI experts?

Balancing innovation with customer trust

Where's the line? How should businesses balance innovation versus customer trust?

Chris F: Oh Polly's use of AI is a really clever use for discount delivery, but also shows the tension between personalisation and fairness. 

Rob H: I'm not surprised at the mixed feedback on this. On the one hand, I do see the innovation – changing the paradigm of how one does discounting. A big issue is that businesses are not fully considering transparency. 

We talked about how the perceived lack of trust or worry about the ethics can be a barrier to adoption. It's a barrier to a seamless experience or a good feeling from a consumer when you're interacting with AI in this way. To get that balance right, the key is to really identify and focus on that transparency – being able to explain how and why the solution arrived at the conclusion that it did. 

If we want to look at it from a discounting point of view, you don't want the user to feel the element of profit driving: “I'm only going to offer you a discount if you think to ask for one.” It can lead to cynicism. In this case, they were genuinely looking to positively engage with their customers and truly innovate. But, there is a danger that without explanation, it can come across as potentially exploitative of those who are less savvy with the technology or less inclined to ask these questions. 

“Why am I getting a discount? Was there just a fat margin that you were hoping I wouldn't get to?” Thinking about the decision process that frames the AI: “Is it because this is a repeat customer? As a repeat customer, we see that you buy from us regularly. We'd like to offer you a loyalty discount on top of that.” Or: “Because there's an opportunity to buy more things, we'll give you an additional discount.” Otherwise, it feels opaque, and that black box can be a real barrier to trust.

Chris F: There’s quite a bit of education still to be done around AI, because ChatGPT is a commonly used tool now. But the reality of what you're talking about is just doing something and putting it out – not fully explaining the decision-making. That education piece is quite big, isn't it? 

The need for AI education and literacy

Rob H: Huge. You provide education in general to the masses. Tomorrow, I'm going to Imperial College as part of their advisory board to talk about the future of computer education and how AI might shape that. 

You've also got it within companies as well. A big part of what we're doing here at Dojo is starting to think more intentionally about how we drive AI literacy. I believe in the power of this tool, it's only getting more capable – still not great at everything! That's the education piece, being honest and vocal: “Yes, it's really strong at lots of things, but it's also not yet good at a lot of things.” Sharing this is so important from an educational point of view. Otherwise, you're getting these warped points of view depending on what you see. If you only see the bad, then you walk away disillusioned with what it can do. Equally, you potentially succumb to the hype if you’re only seeing the amazing success stories.

Why invest in AI now?

Chris F: What are the biggest reasons to invest in AI? Why would you get into it now and not wait another five years?

Rob H: As a business, we're in an incredible sweet spot right now. It's still new. We've had enough success, interest and investment in this that it's gone beyond being niche and a fad. There's a growing consensus that this is here to stay. As a business, you've got the benefit of more tooling. There's more infrastructure available with which you can start to use it. 

The barriers to entry are lower, but you've still got that sweet spot where you're early enough to be considered an early adopter. Particularly if you set up the groundwork so that you're not just spinning up your next POC – a lab project that doesn't go anywhere. If you're intentional about these things, you have an opportunity as a business to create the playbook for what great looks like and establish yourself as a thought leader. It’s going to help not just your business, customers or the product that you're offering, but also your talent. It's going to be a powerful tool for attracting and retaining talent. 

There is also the opportunity cost of not getting involved now – it’s rapidly growing. If you're not doing it, your competition almost certainly are, and I mentioned that talent point as well. You're going to struggle to retain. People are increasingly looking to work in roles and within companies that give them access to this tooling for upscaling. There's a real danger between your employees versus your competitors. Customer expectations are also growing. They're seeing this in other parts of their lives – like Netflix or Amazon – where AI is increasingly used to personalise their experience. That expectation is broadening. 

There's a real danger that businesses become irrelevant if they don't get involved. I’ve had a quote shared with me from an author called Peter Hinssen. He wrote a book called ‘The New Normal’ a number of years ago. He said, “BC no longer stands for ‘Before Christ’, it stands for ‘Before ChatGPT”. It's a great illustration of the fact that this is a material moment in time, much in the way the internet or smartphones changed the paradigm in how we're engaging with our consumers. Businesses won't be relevant if they're not acknowledging it.

Chris F: I'm old enough to remember the internet and phones coming in. I remember people saying, “Why would I put my company on the internet?” Obviously, we've moved on since then. Same with social media. This is the first time since then that I feel a big shift in business and our day-to-day lives.

Rob H: Hugely transformative. Businesses have an opportunity. We're still early enough to be that thought leader to help shape what this is. That's tremendously exciting for businesses.

Chris F: We talked about some efficiency gains, particularly in supply chain and cybersecurity, which I spoke to Naveed about, which was eye-opening in terms of what Dojo are doing. 

Quick wins: Where AI delivers fast ROI

Where are you getting the quickest ROI?

Rob H: Cybersecurity is a great one – demonstrating or talking about the value proposition in terms of protecting our customers and preventing fraudulent transactions. At the scale we're at, this can be tens of millions of pounds. It's things that I've touched on already – customer service is a fantastic area from an ROI point of view. 

We have 150,000 businesses, and we pride ourselves on first-class service, which requires high-touch engagement from our customer service agents. It's the customer operations that are fantastic from an ROI point of view, because it's not just what a customer might directly experience – it's what happens behind the scenes. How do we empower our customer service agents to have more free time, or the time that they do spend with our customers, to be that much richer?

Getting started with AI: Dojo’s roadmap

Chris F: How do they get started on that journey and build a roadmap for AI?

Rob H: My advice to listeners would be to take a step back and think about what you want AI to unlock or achieve for your business. I really believe in the transformative effect it's already having with businesses, and there's more potential out there. 

It was the recognition that this is going to transform every role and business. That led us to crafting a vision statement for AI – to be the model workplace where AI is unlocking the full potential of our people. To see them be the happiest, most innovative, most productive professionals shaping the future of in-person payments. We're talking about transforming our business into an AI-native business. That was hugely important in defining and framing the vision for our business.

As part of identifying that vision, we moved on to the next step: “What's going to be our next 12-month mission towards that vision?” For Dojo, that was looking at establishing foundations on which we were going to build an AI-native business. This included AI literacy. This was with our exec team through to the rest of the company, making sure we understand what this is and what its potential is. 

It's making sure that we have the right data foundations, data assets and governance in place. This is still a relatively nascent field. From a security point of view, we need to make sure that we are bold but grounded and informed. A big part of that is not doing this as a tech silo initiative. This is us working with Naveed's security and legal team to make sure that we’re aligned and working together to set us up for success. We didn't want to hit the proof of concept ‘fancy demos’ that can't go anywhere because it doesn't comply, from a regulatory, legal or security standpoint.

Going back to ‘don't be that solution, finding a problem’ situation. We started canvassing across the business for our top problem statements. We framed it as: “We're asking this question because we're excited to see where AI could play a role.” The brief told us not to offer an AI solution – not to assume AI can or can't fix your problem. “What is your number one problem?” I often frame this as: “What would you get your next ten people to do? What are you looking to hire for?” That's a good indication of where your biggest problems are.  

In a matter of weeks, we had over a hundred use cases submitted across our business, which we curated. That's the next step, which is to be intentional about which ones you start with – it's not starting with a hundred. It's great to get so many, but not all of them were great AI fits.

It’s about finding the high-impact ones and picking out two or three, making sure that they're achievable while thinking about success metrics. What does great look like if we work on this? Then, getting a team of people working on those. Those steps have worked really well. They could really be useful for framing how another business might want to get started if they haven't already.

Internal capability vs external partners

Chris F: If you’re a smaller business or an even bigger enterprise, how would you balance internal capability versus an external partner? How much do you want to keep in-house?

Rob H: The size of your organisation will play a role in this. If I'm agnostic about the company size, I’m thinking how precious you might want to be about it. It depends on the nature of your business. If I’m building AI LLM models, then I'm probably going to be fairly precious about who I have working on that. That’s a very proprietary IP that I want to protect. For a business like Dojo, that's not our layer or USP that I'm seeking to expand our thought leadership in. 

There was a concern about how much energy we invest in low levels at that stage. These are areas that are increasingly being commoditised. If I think of AI infrastructure tools or AI platform services, I'm mindful that we are very limited in how much we want to build at that layer, rather than consume. I'm more interested in how we use that infrastructure and capability to unlock better payments and customer service – the bits that make Dojo special. There isn't a playbook for how to do this well. 

There is a USP in businesses that know how to unlock this potential, and I'm certainly keen to invest in that in-house. But even on that journey, we don't need to go it alone. So, how do we stand on the shoulders of giants? 

We have a number of technology partners and collaborators, but we're working very closely with Google. Whether that's their Gemini model or the Vertex AI platform on which we build the agents. If I take Gemini, for example, we've rolled it out across our entire real estate. It's adoption, training, and putting it in front of people, and the tooling is more available now. That helps you do certain things in-house more readily, but we are also working with partners to help us validate the way we're thinking about some of these solutions. 

We're a very big Salesforce partner and have invested a lot in preparing data to be consumed across the various Salesforce offerings. This made it a natural place for us to start exploring synergies with Agentforce. They are an AI agent platform built on top of Salesforce. 

We're looking at various partnerships. A really interesting one that I'm quite proud of is our collaboration with Imperial College. We have 10 Imperial image computing AI students with us, and they're part of a newly formed AI trailblazer program. We've got investment from Google for an intensive, bootcamp-like training.  Not just on the technology, but on what Dojo is, what it means to work in customer service – what are those problem statements? Or in risk and underwriting. It’s really immersive, deep, in-house knowledge, which we used to see those use cases I mentioned earlier and start to explore different streams in a rapid, iterative, incremental way. 

How do we start to test things and validate whether or not we're getting the results we expect? We will stumble. We will hit a few dead ends, but that's okay if the feedback cycle is relatively short. It's an in-house curated program, but it's augmented and supported by partnerships.

Partnerships and grassroots involvement

Chris F: Bringing it back to payments, how is it adapting to AI and is it keeping up?

Rob H: Definitely more to go. I'm aware that we’re in a highly competitive industry. You're seeing that across our competition, and Dojo is certainly not resting on that front. We talk about customer service – Dojo’s been using AI for quite some time in that space. Over a year ago, we launched a call summarisation piece that tries to provide our customer service agents with an automated way of generating their call summaries and actions, freeing them up to take the next call. 

More recently, with Salesforce, it's the speed with which things are moving. That was a UK English-based solution. Over the last year, we’ve become a more international business with operations in Spain and Italy. We’re always looking to expand this to other languages. With the evolution of the AI tech stack, we don't need to do more tech investments. It’s that art of the possible now – it’s just another language.

AI’s role in transforming payments

Chris F: It’s almost a barrier removal as you discussed before. A normal business two years ago would probably struggle. 

Rob H: Absolutely, it would be a huge investment to make a multi-language capability. That AI trailblazer group is looking at a customer service agent chatbot. They've got so many information sources with which to try and offer support to our customers. How do we consolidate lots of disparate information sources, in real time, for our customer service agents – so that they have that context-aware, real-time information about the customer that they're speaking to, and a much richer history of that customer? 

As we become multi-geographic and also multi-product, you've now got more cognitive load that those customer support service agents need to bear in mind to do an effective job. How do we turbocharge that?  It’s an area we're looking at, and I'm sure other businesses will be as well.

Fraud is a big one. We use AI tooling to help us with more accurate detection – fewer false positives. The speed with which fraudsters move means that traditional rule-based systems simply can't keep up. It’s them fighting that game on a more equal playing field.

Chargebacks are another area, typically very bureaucratic and process-driven. Not necessarily the first domain you might think of when it comes to innovation. We're looking at how AI can automate large aspects, particularly the information-gathering process. For the readers and listeners, chargeback is a dispute mechanism. This is where a consumer disputes whether a transaction was really made by them. That's an involved process in terms of backwards and forwards. Do I need CCTV evidence that the merchant supplies? It can be quite an expensive, laborious process for us and our customers. We’ve got a stream at the moment, building an automated chargeback flow powered by AI. That massively speeds up the summarisation, the analysis of that information. It tries to solve this quite painful problem for our customers.

Chris F: That's a great example of a lengthy process. When you were talking about a hundred common use cases – with lots of backwards and forward, maybe involving three different parties: the customer, your customer and you. Bridging all that together with AI – that’s a shortcut to success.

Rob H: When we saw that use case, it shot up for the reasons you just described. It was a clear pain point for those three parties involved. One where we saw AI could be a really great fit – early results certainly indicate it.

Future trends in AI: What to watch next

Chris F: Can you give me three emerging AI tech trends that we should watch out for in the next 12 months?

Rob H: One that comes to mind would be hyper-personalisation at scale – providing truly unique, relevant, personalised feedback to a customer in real time. Loyalty is a great standout – doing that well and with consent. It comes back to trust. Where are you coming at this from a decision process? Why are you offering the services? It's a really exciting opportunity to delight the end-consumer, with things that are relevant to them. They're expecting more of these kinds of things. It's an opportunity to really wow and delight. 

For example, “it's great that you got an offer on burgers – but I'm a vegan and I’m not interested in your beef burger offer.” It’s about recognising what's right for each individual consumer. Maybe it’s an early riser – get a pound of their coffee because they’re coming in super early. You can really scale it with this technology. It was computationally, labour-intensively, and infeasible to look at before. 

Another one is the rise of agentic AI. By agentic AI, we’re talking about systems where you have multiple agents – think of them as digital workers. The difference is starting to join those workers together so that they can start to tackle more complex workflows and use cases. Rather than having a summariser or an analyser, it's being able to group those things together to be more proactive. From a fraud point of view, it's not just detecting or flagging that a transaction might be fraudulent. It might also be starting to initiate actions to warn the customer that there's a growing threat in other respects, or starting to automate gathering information or even automate the chargeback dispute process.

Chris F: Almost like triggering behaviour, like you would in a marketing campaign – like a nurture campaign.

Rob H: Yes, you can start to do far more complex tasks. Customer support is a recurring theme when we talk about ROI. How can you have multiple agents looking on behalf of a customer account – maybe identifying opportunities? Maybe there's an agent that tracks our terminal Telemetry data and recognises that one of your terminals is old now and the battery health isn't as good, and orders your replacement terminal in advance. It’s starting to get more sophisticated, so I'm very excited about the potential of taking a step up and looking at the composition of multiple agents. You're seeing a lot of technology trends supporting things like Google's agent protocol – how do we make it easier for these agents to talk to each other, irrespective of how those individual agents have been built? 

The third would be the explainability of AI – that last use case on agentic is very powerful. You also need to have those guardrails to make sure that they are working as intended within their appropriately bounded contexts. Even on the hyper-personalisation piece – making sure that we're using information we're allowed to use, that's appropriate, and not perpetuating bias. So that we can explain why it is that we're reaching out to you with a particular offering or service.

All of that is becoming very relevant. You have legislation that's looking at this – the EU AI Act for transparency and fairness. The technology needs to start evolving in much the same way the agent-to-agent protocol does. You will start to see standards form around how a model and LLM can prove how it's arrived at its decision process and standardising how we do that –  it’s the direction that I see the AI tech enablers investing some energy in.

Chris F: It's fascinating. Anybody listening has been on a journey for the last 45 minutes, across all the different factions of what AI can do, and how it could be treated in business. That was Rob Howes, everybody. You can find out more about Rob or Dojo at dojo.tech, and we shall see you all next week. 

Wrapping up: The potential of AI in business

This episode explores what it really means to put AI into action – not just as a buzzword, but as a tool to drive meaningful change. From automating chargebacks to detecting fraud faster, Rob shared how Dojo is using AI to solve real business problems at scale. It’s not about adopting every new tool – it’s about embedding AI where it adds value, supporting teams, improving consistency, and powering smarter decision-making. As more businesses move from proof-of-concept to production, it’s clear that success depends on strong foundations – and technology that’s built with purpose. 

Curious about how other operators are building stronger tech foundations? Catch up on Episode 1 – all about scaling a hospitality business – or dive into Episode 2, where we unpack how to manage a fractured tech stack.