How to Choose the Right AI & Automation Partner for Your Business
AI and Automation Poses a New Challenge in Business Process Outsourcing
Business leaders are pursuing AI and automation with a mixture of excitement at the opportunity unfolding and fear of being left behind. Given the high stakes and level of expertise required, AI outsourcing is in high demand. Yet the business practices by which companies outsource non-core operations, are not the ones that will deliver success when automating core businesses.
Mordor Intelligence expects 37% a year growth in AI-as-a-Service, with the market reaching almost $100 billion by 2030. This is a small part of the global outsourcing market. Precedence Research pegs global IT outsourcing at US$612 billion and rising 8% a year to US$1.35 trillion by 2034.
The primary purpose of outsourcing is cost saving. This has led to the rise of professional procurement departments, specialising in cost control and operating as a silo. As a result, business objectives are one step removed from the vendor relationship, which poses a challenge when automating core business functions. Add to that the novelty and uncertainty over AI and it is no surprise that ISG reports generative AI has the lowest satisfaction score among services from technology providers.
As a result, business leaders must take direct control of the AI automation decision. They are moving from outsourcing of peripheral operations to changes at the heart of a company. While the opportunity for cost control is considerable, the main goal is to maintain competitiveness at a time of significant disruption to many industries.
MSBC is working with a client in the financial industry to introduce AI into the credit underwriting process. The project aims to increase underwriting capacity while reducing the cost of individual contracts. The challenge is to match the human ability to be flexible in negotiations, while maintaining commercial value.
The project involves upgrading the credit analysis model to cover a longer period of time, aligning it with industry best practice, and integrating insights from actuarial and pricing models to ensure consistency and accuracy. Thereafter, AI is used to collate and comment on credit scoring, and offer dynamic visualisations through a chatbot that delivers a unique experience based on each user’s needs.
Automating Individual Services and End-to-End Processes
Business leaders are AI-curious but only a handful of technology-led companies have made much progress in changing how their industries operate. Chief among those are IT businesses, where AI code generation, debugging and review are now standard. Results to date are mixed, with for example Microsoft reporting 30% of its code is AI generated. Meanwhile, 1.3 million developers worldwide are subscribers to its Github Copilot. At the same time, Opsera reports that developers accept around 30% of Copilot suggestions and only 17% remain in the final codebase after edits and reviews.
The limited progress on disrupting industries raises an important point about AI automation. Where existing services are already outsourced, for example to software-as-a-service providers of CRMs, accounting, or email handling, the vendor is expected to deploy AI. There is no need to change the current procurement process, or compete with these suppliers to automate specific and siloed processes. In contrast, when automating end-to-end processes within a business, understanding and controlling the overarching AI architecture and outcomes is essential.
How then, should businesses go about choosing an AI and automation partner, when the relationship will need to be much closer and iterative than with existing outsourced providers?
Understand Your Business Objectives
Before engaging with potential partners, it’s imperative to have a clear understanding of your business goals. Are you aiming to automate and simplify existing operational processes? Perhaps you want to improve your interactions with customers, or your focus is on developing new products and services. By writing out your primary objective, you can better assess which partners have experience in your area of interest.
We recommend stating your aims in a problem statement. This can embrace the SMART principles of being specific, measurable, achievable, relevant and time-bound. Avoid being too prescriptive about the technology to be used, or the particular features of an outcome. A problem statement addresses who is affected by the problem, what is the gap between reality and unmet need, when and where is the problem occurring, and why it is worth solving.
A clear problem statement will help advance your progress through the remaining stages of your AI project.
Evaluate a Partner’s Expertise
A partner’s technical prowess is essential, but their understanding of your specific industry nuances is just as important. For instance, a partner experienced in healthcare automation may not be suitable for a retail-focused AI project. An ideal partner will have relevant case studies demonstrating success in your sector, regulatory knowledge of industry-specific compliance requirements, and individuals with experience in your industry.
Thereafter, it is important to assess a potential partner’s technological capabilities. This can be challenging with AI, because it’s new technology and unfamiliar to even some of the most established IT professionals. If you are in this situation, focus instead on what the technology is able to deliver to you.
First up, AI should be scalable, meaning it grows with your business. It then needs to integrate with your existing technology, because an AI-side project running in a separate environment is hard to progress from proof of concept to roll-out. AI must also be secure and adhere to the highest data security standards. Ask for explanations and evidence of how an AI partner intends to manage these three factors.
MSBC teams up with Scan Computers to provide cloud services that allow AI to be developed in parallel with your existing environment. By matching the specifications of your current infrastructure, we ensure that your AI pilot project is capable of full rollout once testing is complete.
Not all automation is created equal. Some partners will be stronger in Robotic Process Automation (RPA), which automates repetitive tasks. Others will have experience with Intelligent Automation, combining AI with RPA for more complex decision-making processes. With GMI reporting that 40% of companies expected to integrate RPA into their processes by 2026, understanding the depth of a partner’s automation capabilities is crucial .
Consider Cultural and Operational Fit
A successful partnership extends beyond contracts to communication, flexibility and support. Some companies offer openness and complete transparency, which sounds great until you are inundated with updates and cannot see the wood for the trees. What is your desired cadence of communication, what format do you prefer and how much of the decision making are you prepared to entrust to your partner? Setting guidelines for communication early in an engagement goes a long way to building a successful relationship.
The next consideration is flexibility. Everyone says they offer agile development, but the devil is in the detail. How far in advance do you want to see sprint plans, will you be commenting on individual user stories and how frequently are you prepared to revise project details? It’s worth establishing whether time and materials, or managed services, would be preferable ways to progress. The most important thing is that your partner’s way of working suits the culture of your company.
How much support you will need is the third consideration. Partners should offer training and support after implementation, and can be involved with monitoring, maintaining and updating AI systems. The pace of change in the technology makes this a prerequisite for your project.
MSBC is partnering with an industrial client to analyse financial scenarios. The automation involves the preparing trial balances, performing reconciliations and drafting accounts. The AI involves creating synthetic data to run multiple scenario analysis and recreating the accounts in real-time under a range of assumptions about raw material costs, exchange and interest rates.
Analyse Cost and Value
While cost is a significant factor, it’s essential to weigh it against the value delivered. What is the target return on investment (ROI), the project efficiency gains and revenue increases? Do not be fooled by the promise of cheap installation, only to find that running costs such as cloud computing get out of hand. The total cost of ownership includes maintenance, upgrades and inference costs.
The best way to gain an expectation of what your partnership will deliver is to get feedback from your partner’s existing clients. Ask them about their levels of satisfaction with different elements of delivery, such as timekeeping, reporting and staying on budget. Check whether the actual outcome matched up with what was desired and gain an understanding of any unexpected challenges that arose. You might also ask to meet the engineers with whom you will be working, to ensure that you have rapport before making a long-term commitment.
There is a growing trend for both buyers and suppliers preferring value-based contracts. This is popular in AI automation because what works best evolves rapidly. The value that can be extracted from a project may double overnight with the release of a new model with enhanced capabilities.
Traditional relationships with outsourced suppliers place a lot of power in the procurement department. Emphasis tends to be on cutting costs and delivering against service level agreements. These tend to include penalties for underperformance. Offering the carrot of shared upside for increasing the value of a process may be more appropriate when seeking to transform the way a whole business works.
Conclusion
Choosing the right AI and automation partner is a multifaceted decision that requires thorough evaluation of their technical capabilities, industry expertise, and alignment with your business objectives. As the ultimate purpose is business transformation, a traditional relationship in which buyers beat down providers on price, will not produce the best incentives for progress. AI automation is a lengthy and continuous commitment and one best approached as a genuine partnership.