Feature
There are two main drivers for generative AI (GenAI) adoption right now, and both of them are fear-based: fear of missing out and fear of being disrupted. Fear in any circumstance can cloud our judgment and decision making, and in this case, businesses must remember that the R&D they’re investing in and delivering on behalf of customers should still result in a profit. The challenge is that GenAI is still nascent in the sense that its unit economics are not fully understood yet. Not only are they not fully understood — they’re also still rapidly evolving and improving.
While organizations are rushing to implement the technology, I’ve found that a handful of barriers to adoption arise repeatedly for enterprises regardless of size or industry. The good news is that there are ways to approach, overcome and work around these barriers to start reaping the benefits of the technology.
Challenge 1: Cost
The hardest barrier to overcome is the cost associated with building or training your own model. This can be time-intensive and labor-intensive and come with a million-dollar (plus) price tag when accounting for the skill sets and compute infrastructure involved.
Solution
It is possible to get started without a huge upfront investment in training your own model. Start out by experimenting with prompt engineering, which feeds your company’s data into an existing large language model (LLM) at the time of inference. For example, inputting individual customer data into a model that already knows what a typical customer retention email looks like can create a personalized retention email that offers an individualized discount.
This costs nothing but time and experimentation to ensure accuracy and precision, as there is no supplemental compute or engineering cost. Optimization is important here to ensure your business is getting what it needs from the model and not too much or too little.
You can even go beyond prompt engineering and explore techniques like retrieval-augmented generation (RAG), or even modest fine-tuning techniques, like low-rank adaptation, or continued pre-training.
Challenge 2: Change
GenAI technology is changing literally every day. How can organizations know if they’re making the right technology investments in such a rapidly evolving environment? Will the right investment today be the right investment tomorrow?
Solution
The answer here is don’t go all in on a particular technology. Hedge your bets. I can guarantee that the best model today won’t be the best at some point in the future. The most important ability an organization can have is to adapt and pivot quickly as industry standards develop. You can build surrounding APIs and infrastructure that support any model approach. The organizations that will survive the churn of models and technology over the next two years are organizations that will adopt a “model vending machine” approach.
Challenge 3: Perception
There is still a considerable distance between the perception of where we are with GenAI technology and its reality. However, the buzz around GenAI has many organizations believing it can transform their business overnight.
Solution
To realistically gauge the success of any GenAI initiative, start by focusing on how it can be used to improve the efficiency of existing revenue streams and incrementally increase margins. I liken the advent of GenAI to that of the typewriter or copier in that it allows us to move much faster through significant efficiency gains. Foundation models or large language models (LLMs) are, at their core, a calculator for words. Businesses derive value from these models when they apply their own data and their own products to these. Don’t just add a chatbot to your site and expect that to drive revenue. You have to think about how to integrate these capabilities into your core products.
Challenge 4: Balance
Finding the balance between customization and scalability of a GenAI solution can be tricky. Any solution must be able to both work for the distinct needs and goals of a business and grow and scale as the technology and the organization evolve.
Solution
When you buy an AI solution, it has limited customization opportunities. When you build an AI solution, you can tailor it to your business. Custom AI services tend to offer more options, in terms of customization and scalability in comparison to off-the-shelf solutions. However, as IT leaders decide the right course for their business, it’s important to weigh the pros and cons of different service providers and solutions to ensure their organization is getting what they need in both the long and short term. I’d strongly advise against making long-term commitments to any single model provider right now. That’s where a platform with multiple foundation models can make it easier to flex between different model providers as their capabilities change over time (e.g. Anthropic, Cohere, Meta, etc.).
Challenge 5: Interface
The most common mistake organizations can make with GenAI is not considering the interface through which they’re providing their solution. I’ve said this before, but I truly believe that it can’t be overstated: slapping a chatbot on the bottom of your website is not going to deliver the desired outcome nor the business value of the technology.
Solution
Always, always, always start with the business value you’re trying to achieve with GenAI and only then work backward to create the technical deliverables that will deliver on that business goal. Custom apps and solutions often provide the most useful experience for users and as such, deliver the most business value.
Webinar
Oct
22
Beyond Storage: Smarter Content, Bigger Impact with DAM + AI
Discover how the DAM + AI duo makes content smarter, stronger and more accessible.
Register
Webinar
Nov
6
Fix the Content Bottleneck: Build a Better WebOps Strategy
Content stalled? Dev overloaded? You’re not the only one. Learn how streamlined WebOps bridges the publishing gap.
Register
Webinar
On demand
Agentic AI Playbook: Real-World Customer Service Use Cases You Can Deploy Now
Boost self-service by 30% and slash call volume by 63% with agentic AI.
Watch Now
Webinar
On demand
CMS Briefing: A Live Look at What’s Next in AI-Driven Platforms
Learn how leading organizations are using AI‑driven tools to publish faster, personalize smarter and stay secure.
Watch Now
Webinar
On demand
Ready or Not: How Data-First Organizations Are Unlocking Agentforce Potential
Learn how to cut through the noise, activate Agentforce and build a Salesforce AI strategy that actually delivers.
Watch Now
Webinar
On demand
AI in Customer Service: Faster Resolutions, Happier Customers
Don’t let rising demand burn out your team. See how to build a smarter, more resilient support org.
Watch Now
In Closing
The excitement around GenAI is real and warranted. As with any new technology, the challenge for organizations lies in channeling that excitement into projects and solutions that drive real business value in order to show a positive ROI on these investments. To that end, I’d give the same advice to any company: define the outcomes you want GenAI to drive and continuously measure your efforts to see if you are, in fact, achieving those objectives. There are lots of components to optimize with GenAI, so keep reviewing your business case and underlying technologies to ensure enough marginal value is being realized to justify the investment.
Most importantly, don’t get stuck at this point in time. The answers will change as the tech continues to develop. Without a straight-line use case to measure ROI, the most important thing to do is to keep experimenting and keep learning. Take smaller bets, measure the results and optimize as you go.
