Main content

AI implementation requires restraint — little victories today support large wins later

Raju Vegesna Profile picture for user Raju Vegesna May 14, 2024
Summary:
When it comes to AI implementation, Raju Vegesna advocates for businesses to hurry up and wait. Read what two Zoho customers learned by not rushing to switch over the controls to generative AI.

Hurry Up and Slow Down © solarseven - Canva.com
(© solarseven - Canva.com)

The current generation of generative AI has been buzzed about as intuitive, powerful, and soon-to-be ubiquitous. It has also been hailed as a supreme waste of time — technology whose capabilities fall short of what was promised, with no sign that development will ever catch up to the hype cycle.

For the most accurate picture of where AI stands, companies have to listen beyond the current conversations among tech influencers and hear what other customers are saying. And, in this case, they're expressing skepticism about AI for AI's sake. But if AI can help them improve their workflows and decision-making, they will consume it. In some cases, they already are, but they may not even call it AI. It just helps them do their jobs better. A lot of vendor technology has woven AI into the fabric of its SaaS tapestry, running in the background and carrying out tasks away from prying eyes — whether it's complex predictive analysis or simple task automation.

Consider that AI has been developing on an expedited timeline. At first, rudimentary AI was capable of only a sliver of what we know now. Its functions could be thought of as ingredients in a recipe, and given a finite set of ingredients, there were only so many recipes users could come up with. They were probably simple, contextual-only recipes, at that. However, with even one more ingredient, the number of possible recipes increases exponentially — think about how many more sequences are possible when one more digit is added to a combination lock. 

Ultimately, companies cannot be successful with AI unless its findings and functions can be directly applied to Business Intelligence (BI). This is what companies need right now, and the promise of future AI functionality does little to solve the problems faced today. Instead, successful companies are experimenting with a few ingredients at a time before jumping right to a complicated recipe — practising a few appetizers to satiate their hunger in the short term before venturing to the main course.

Here's more on how a few companies are maximizing potential with minimal ingredients.

Beyond the industry standard

For Servifácil, one of the top gasoline retailers in Latin America, the promise of AI was that it would future-proof its business by handling many of its employees' repetitive, routine tasks and allow them to focus on what's most important. The company was also blind to how well its operations were doing, particularly its sales processes and how quickly and smoothly they were capturing new customers, and hoped AI would provide necessary data on which to guide organizational changes.

In searching for the right solution, the company had to weigh whether the AI potential of a particular vendor was worth sacrificing its usefulness in the present moment, meeting the business needs it has today. It also needed to ensure the software was easy to learn for anyone across the organization, knowing solutions are worthless if adoption is low. Still, others in the industry have been sticking to spreadsheets, and it was beyond time to at least upgrade to something centralized. It ultimately decided to go with a unified system that includes Zoho's CRM, Analytics, and Desk software solutions, with embedded AI functionality, hoping its employees would benefit from a bit more visibility across the board.

While it expected AI to produce results, Servifácil didn't expect the new software to be as successful as it was. Since implementation, the company has doubled the amount of gas it was selling to customers, month over month, and nearly tripled the amount of total contracts. Much of this can be attributed to the data produced by the unified system, which is automatically synthesized by AI and can be used to inform process changes immediately. Rather than wait around for AI to revolutionize the way it did business, Servifácil used the technology to meet its current business needs, and in the process exceeded them. AI played the supporting role, and the company was wise enough to let it excel there, first.

Keeping on task

Recently, Luxer One, a manufacturer of smart locker systems for residential buildings and commercial properties, knew AI would eventually play a central role in its business. But, it wasn't in a hurry. The technology was fairly new, and the hype was out of control. Rather than switch its processes over quickly, ceding total control to the AI, the company started small, allowing the AI to handle specific questions.

First, Luxer One integrated SalesIQ and ChatGPT with its support operations and used Zoho Analytics to determine the success of this change. Things went well — not only did Luxer One use the information to inform the development of new products, but it was able to show data points to prospective customers, demonstrating, with little ambiguity, how impactful its services would be.

The initiatives were successful, yet Luxer One remained measured. Rather than rush to expand its use of AI, the company dug deep into the results it already produced, stored and monitored from Zoho CRM. Its employees combed through chat transcripts, calculations, and corner case scenarios, ensuring the AI wasn't behaving suspiciously. This human-run fail-safe helped the company understand how its AI would operate and offered an opportunity to course-correct before the technology was rolled out to the larger organization.

Now, Luxer One has opened AI use to its developers, with one caveat — it can only be used for specific tasks or routines, rolled out one at a time. Within these guardrails, the company can guarantee AI is applied strategically rather than as a catch-all, eliminating errors and positioning itself to take advantage of the technology as it evolves. This approach also allows Luxer One to evaluate AI against its current processes before attempting to have it bite off more than it can chew.

Single source of truth

One day, AI will be capable of handling a sizable portion of a company's tasks without much oversight. We have a long way to go until we reach that point, though. The technology is too prone to mistakes at the moment, and trust is low — especially when companies learn that AI often runs in the background, collecting data for the sake of learning.

Companies need to start testing AI, as there's no doubt the technology will make an impact on the industry, but must do so by adopting the measured, incremental approach demonstrated by Servifácil, Luxer One, and many other midmarket enterprises. A single source of truth, ideally a centralized CRM, makes the process more effective and acts as a human bumper to keep the AI from veering off course. In building new recipes, you have to ensure each ingredient is ready to be added to the mix.

I look forward to continuing this conversation with customers in June — hope to see many folks at Zoholics in Austin.

Loading
A grey colored placeholder image