Going the AI distance in 2025: Hold your own or risk falling behind
Nearly One-Third of GenAI Initiatives Stumble Post-Proof of Concept, According to Kellton's Perspective on Gartner's Forecast
Like a tidal wave, Artificial Intelligence (AI) is barreling towards businesses worldwide. But, as Gartner specialist Rita Sallam predicts, 30% of Generative AI (GAI) endeavors will hit the rocks post-Proof-of-Concept (PoC) by the end of 2025, dashing hopes and expenses. What's causing this wave of sinkings? Poor-quality information, insufficient risk protection, skyrocketing costs, or questionable commercial value.
Leaders yearn for AI returns, but manifest success remains elusive for many enterprises. As the AI universe expands, implementation costs grow, stretching both finances and patience.
A primary hurdle for many ventures lies in demonstrating the value of GenAI investment. GenAI promisingly transforms operational systems and opens new business avenues. However, such undertakings come with a hefty pricetag ranging from $5 million to $20 million.
Sadly, there's no silver bullet with GenAI, and costs aren't as predictable as traditional tech. Costs depend on the specific applications, infrastructure, and execution strategies. Whether a company wants to be an AI front-runner or focuses on productivity gains or augment current procedures, each strategy, and cost level, carries its risks, expenses, and strategic influence.
Gartner believes GenAI requires trust in indirect future financial investments rather than immediate returns on investment (ROI). However, many Chief Financial Officers are still holding back, fearing short-term risks compared to long-term rewards. This apprehension can impact the distribution of investments towards tactical versus strategic outcomes.
Thriving in the tumultuous GenAI seascape: Kellton's take on Gartner's forecast
Businesses are enticed by GenAI's disruptive power, revolutionizing sectors like never before. But, the high costs associated with GenAI projects can frighten leaner organizations. How can companies navigate GenAI costs without capsizing? Here's Kellton's perspective on sailing the GenAI seas economically:
- Harvest open-source and free models:
A smart way to reduce costs is to tap into the wealth of open-source solutions and free models available. Here's the lowdown on some options:
- Open-source APIs: Apple doesn't fall far from the tree, for instance with APIs like HuggingFace Transformers, TensorFlowHub, and PyTorchHub offering thousands of pre-trained models for text-based tasks.
- Microsoft Azure with SLA: Azure offers access to some of its AI models under Service Level Agreements (SLA), providing comfort for businesses relying on these services. Key offerings include Azure Cognitive Services and Azure Machine Learning.
- Meta's LLaMA 3.2: Facebook's parent company released LLaMA 3.3 as an open-source large language model, offering commercial use and cost-effectiveness.
2. Leverage efficient language models:
Large Language Models (LLMs) form the heart of many GenAI applications. Choosing efficient models can slash costs significantly:
- OpenAI GPT-4o-Mini: Think of this pint-sized version of OpenAI's popular model as the perfect appetizer. It packs similar capabilities at a fraction of the computational cost, offers faster response times, and requires less memory.
- Claude 3: Known for its efficiency, Anthropic's Claude 3 provides a balance between ethical AI performance and operational cost. It's superb for chat, large-scale deployments, has low token costs, and swift inference.
- Mistral: Mistral, another GenAI favorite, offers speedy inference, modest token costs, and high accuracy, making it an ideal choice for businesses needing affordability without sacrificing efficiency.
3. Embrace the cloud:
The cloud offers flexible and money-saving GenAI solutions:
- AWS and Azure: Heavy hitters like Amazon Web Services (AWS) and Microsoft Azure provide robust AI services, freeing businesses from upfront investments in hardware. Key services for AWS are SageMaker, Lambda, Comprehend, and Bedrock. Azure offers Machine Learning, Cognitive Services, and Databricks.
The cloud platform's flexibility in pricing and ability to scale up or down based on demand makes it a cost-effective choice. Enjoy effortless support for automated updates, maintenance, and compliance while reaching users worldwide.
Dodging the POC Game-Over: How to Avoid GenAI Project's Abandonment Trap
To ensure GenAI projects sail through PoC, consider this recipe for success:
1. Focus on Goals that Matter:
Victories elusive when PoC goals aren't aligned with strategic business objectives. To set sail towards success, ensure your PoC addresses pressing business problems.
2. Dock at the Stakeholder Harbor:
Stakeholder buy-in is key to a smooth journey from PoC to production. Involve business leaders, decision-makers, and technical teams to create a sense of ownership and forge unity.
3. Sail with Scalable Solutions:
Pick tools and platforms that scale gracefully from PoC to production to avoid redundant redesigns. Opt for solutions that integrate with existing systems, adapt to changing demands, and facilitate future growth.
4. Harness the Power of Success Metrics:
Set clear success metrics before embarking on your PoC voyage. Quantifiable KPIs, like increased efficiency, reduced costs, or higher revenue, ensures that you can demonstrate and justify PoC results.
5. Tread a Minimal Scope Course:
Strive for simplicity by focusing on essential functionalities during the PoC phase. Avoid detouring on unnecessary features and complications to keep your PoC focused on core objectives.
6. Ensure Knowledge Transfer:
Avoiding post-PoC abandonment requires seamless knowledge transfer. Document each step of the PoC and involve the PoC team in the transition process to minimize the learning curve for the next crew.
By following these guidelines, you can ensure that your GenAI projects stay afloat when it matters most, navigating the challenges of PoC and positioning the project for a smooth conversion into production.
Bottom Line
Smaller businesses no longer need to shy away from the power of GenAI. By leveraging open-source resources, choices efficient models, embracing the cloud infrastructure, and adopting additional strategies for cost savings, even moderately-sized businesses can unlock GenAI's potential without bankrupting the ship.
Remember, the magic lies in tailoring the GenAI strategy to your business needs and budget, starting small, demonstrating proof of value, and then expanding your ambitions. Keep in mind the ever-evolving AI landscape, and relentlessly reassess your approach to remain competitive and attentive to cost optimization.
- In the realm of product engineering, businesses can invest in artificial-intelligence technologies to gain a competitive edge, following the guidance provided by Gartner, which emphasizes the importance of indirect financial investments in generative AI.
- Financially savvy businesses can reduce costs associated with product engineering by utilizing open-source solutions and free models, such as HuggingFace Transformers, TensorFlowHub, and PyTorchHub, during the development phase.
- Technology leaders can choose efficient language models, like OpenAI GPT-4o-Mini, Claude 3, and Mistral, to decrease costs and maintain performance when it comes to product engineering projects using artificial intelligence.
