ROI from Generative AI projects is closer and less risky than you might think –
but you need to know where to look
Gen AI has hit the world of business fast, and with a bang. And it’s causing plenty of uncertainty among organisations that don’t want to miss the boat, but don’t want to take on far-out innovation risk either. The good news: they may not have to.
Has there ever been a bigger buzzword than “Generative AI”? Well, I guess “Cloud” came pretty close at one point. With all the hype and promises of disruption being made around it, it’s no surprise that many organisations are keen to leverage its power sooner rather than later.
But let’s be honest – even for the businesses keen to get stuck in, brand new technology like Gen AI comes with two huge question marks:
- How do we figure out where to start?
- How do we de-risk this move? We’re doing this to improve our business – and ideally gain some competitive advantage – after all. We can’t afford to compromise our data, or reputation
Some common Gen AI misconceptions
They’re the right questions to ask. But since few organisations have the required skills in-house, these concerns can easily stop businesses adopting AI in the near future. For the organisations that feel this pain it’s important to know that things haven’t just been moving fast in the world of Gen AI – they’ve also been pretty quick in Gen AI solutions and services. So some of the problems people associate with AI adoption may already have been solved by now. For instance:
- The risk may be lower than you think. In a tech hype, what makes the news are far-out, futuristic applications, usually associated with business/category/market disruption. This ignores that Gen AI already drives predictable returns in many common business processes (eg in Compliance, HR or Finance).
- Gen AI isn’t just for the front office. You may have similar opportunities in areas you don’t immediately associate with AI (which has been pioneered in customer-facing applications such as chatbots). Looking at Gen AI applications in back-office processes could be a better place to start the AI journey for your business – especially if you experience pushback on AI for customer comms.
- Robust consulting methodologies already exist around Gen AI adoption. BPOs have done a lot of the heavy lifting here, so rather than figure things out for yourself, it might make sense to tap into the insights they’ve gathered from their various engagements.
The library of common quick wins is growing day by day
In fact, hundreds of GenAI use cases already exist, some of them complete with pre-built pilots. This is true not only for horizontal applications (e.g. job application screening for HR; invoice reconciliation in Finance); BPOs are also working on their libraries of AI use cases for industry-specific processes (such as application processing and underwriting for mortgages; learner communication and testing operations in Education; and so much more).
Minimising risk and time investment in AI adoption
But identifying a use case is just one part of the AI journey. You’ll also want to make sure it can deliver returns, be implemented quickly, and ideally win support for further investment. To help businesses with all of the above, we’ve developed an AI starter kit here at Firstsource. The kit uses the experience (including ROI data) from all our past Gen AI engagements and projects to accelerate and de-risk such projects for all our customers, and make their investment predictable.
Here’s what that usually looks like:
- A workshop: Together with key stakeholders, we identify a key business process that might benefit from AI. We’ve developed a number of horizontal and industry-specific heatmaps that can guide this process. Thanks to the experience and data we’ve gathered across a number of industries, we can classify the technical complexity, domain complexity and expected returns (eg productivity improvement) for each:
- Assessment: We consider infrastructure and security guidelines for the chosen use case. Then we evaluate the size, quality and suitability of your existing data for training the model and testing it. In many cases, we can work with as little as a week’s worth of data (eg around HR offboarding).
- POC: A suite of pre-configured proofs of concept for a number of different domains help accelerate the project. They include capabilities such as summarisation, translation, categorisation, or entity extraction (i.e. pulling key values such as addresses or dates from chats, emails, or calls). This is all about demonstrating value fast.
- The business case: This is a comprehensive financial model detailing the costs, benefits, and potential savings of implementing the Gen AI solution. We’re able to build this thanks to our rate cards for each base model and platform, integration cost benchmarks, assessment of operational overheads (eg maintenance, updates) and calculators for total ROI. All of this is based on real-life data from past engagements.
- An implementation playbook: a plan to turn the theoretical model into a production-ready solution. This includes a prompt library, post-deployment analysis of the model’s performance against expected returns and a change management framework, among other things.
The opposite of disruption: a pragmatic approach to AI adoption
By the end of the starter process (which we don’t charge for, by the way), our clients have a set of completed POCs customized to their data and systems, that demonstrate the value of transforming one or more of their business processes using Gen AI, along with a detailed implementation plan and business case.
There are obviously a few more details to this, but I hope that with this quick overview I’ve been able to clear up a few myths around Gen AI and show three things:
- While Gen AI is still relatively new, there are already hundreds of use cases for it. Back-office applications of AI may not be as well-known but they have predictable returns. I’ve yet to come across a business that can’t realise significant productivity gains by adopting Gen AI somewhere.
- AI may have disrupted our world to some extent, but adopting it doesn’t have to disrupt your business. Robust best practice exists already and it’s available to organisations worried about adopting tech that isn’t yet established.
- Starting with a pre-built POC that delivers ROI fast is a low-risk way to build your organisation’s confidence in Gen AI. And it might well open the door for a whole suite of transformation initiatives and continuous innovation.
If you have any questions about our starter kit, use cases relevant to your industry, rallying the stakeholders needed for an initial workshop, or anything else I’ve touched on in this post, please feel free to get in touch, we’re always happy to chat.
