No. 2: Combat cost overruns for AI implementation services
The best implementation service for each organization will depend on budget and individual needs. Services can be small and niche vendors focused on a particular industry vertical or technical domain or they can be large consulting firms.
“By using larger service providers, organizations can get end-to-end, strategic consulting in contrast to smaller providers who are focused on depth and flexibility,”says Alexander.
Although niche vendors offer industry or technology specific expertise, large service providers offer end-to-end capabilities. A managed service provider delivers services, such as network, application, infrastructure and security, via ongoing and regular support, which a niche vendor cannot. Although this option may seem more expensive, after all of the costs of self-providing and self-managing, the price tag may make more sense.
No. 3: Include forward-looking costs and potential compliance costs
Think beyond software and implementation costs. The majority of organizations that scale implementation spend more than half of their budgets on adoption-driving activities, such as workflow redesign, communication and training.
“The most common additional cost exposures are maintaining compliance with other applications,” says Alexander. “Once the choice of AI technology is finalized, it is important to determine what existing applications the AI software may need to access. For example, if the intent is to introduce automation to perform tasks that involve data transmission through an ERP system, then additional software licensing may be required.”
Determine the future costs of the AI project in the pre-implementation phase to determine the actual budget. Staffing, security, privacy requirements, public cloud licensing and skill development are some of the common hidden costs that may go overlooked.