At re:think about 2025, Gartner’s Danielle Casey delivered a transparent roadmap for product and expertise leaders navigating the generative AI curve: not all use instances are created equal—and never all will succeed.
Drawing from lots of of case research throughout industries, the session broke down the place GenAI is already delivering worth, the place it’s simply starting to indicate promise, and the place adoption might by no means scale as a consequence of complexity, danger, or lack of ROI.
For product leaders, the takeaway was easy: In case you’re not being deliberate about use-case technique, you’re already falling behind.
The place Generative AI Works At the moment
Nearly all of present enterprise deployments fall into a good band of feasibility:
- Low complexity
- Average worth
- Fast to implement
Suppose content material technology, summarization, retrieval, and surface-level buyer interactions.
Gartner spotlighted:
- A Fortune 50 automaker utilizing GenAI to generate marketing campaign visuals at scale
- A healthcare supplier transferring from primary be aware summarization to discharge prediction and danger modeling
- A world journey firm constructing a GenAI-based reserving agent that elevated legitimate bookings tenfold
The lesson? Begin with easy use instances—however plan for scale.
The place GenAI Is Headed: Three Applied sciences to Watch
Gartner recognized three forces accelerating the subsequent wave of enterprise AI:
1. Area-Specialised Language Fashions (DSLMs)
Neglect general-purpose LLMs. DSLMs are:
- Skilled on business, perform, or task-specific information
- Extra correct, extra environment friendly, and quicker to deploy
- Higher suited to vertical workflows and privacy-sensitive environments
Instance: A doc LLM designed to know advanced monetary paperwork by studying each the textual content and the doc format. It outperforms normal AI fashions in duties like contract evaluation and compliance, serving to groups work quicker and extra precisely.
DSLMs allow smaller, cost-effective fashions tailor-made for real-world enterprise logic over normal data.
2. Multimodal Interfaces
Gartner initiatives that by 2030, almost each enterprise system will help multimodal interplay. That features:
- Textual content
- Voice
- Charts
- Tables
- Maps and visible information
One instance: a Canadian wealth administration agency utilizing GenAI to course of and generate studies throughout textual content, tables, and charts—slicing report time by 80%. It expands automation potential by as much as 50%, unlocking duties that weren’t beforehand AI-compatible.
3. Agentic AI
That is the place automation turns into clever.
Gartner defines agentic AI by six traits—goal-setting, planning, autonomy, collaboration, reasoning, and adaptableness. It’s a shift from “responding to inputs” to executing towards outcomes.
Instance: an Australian water utility utilizing three autonomous brokers—managing water ranges, optimizing power utilization, and scheduling pump upkeep—all working with interdependent targets.
The place GenAI Would possibly Not Work (But)
Gartner known as out obstacles which might be slowing or stalling adoption:
Market:
Interoperability suffers when AI brokers don’t communicate the identical language. With out widespread protocols, collaboration between specialised and normal techniques is tough.
Enterprise:
Organizations nonetheless wrestle to tie GenAI to measurable outcomes. Many pilot packages look spectacular, however fall wanting proving sustained worth or ROI.
Know-how:
Not each job suits a GenAI-first strategy. To be used instances requiring ultra-high accuracy (e.g., prediction, simulation, digital twins), hybrid fashions—rules-based, classical ML, neuro-symbolic AI—are nonetheless important.
What Enterprises Ought to Do Subsequent
Gartner supplied three actions to deal with now:
1. Audit your present GenAI use instances.
Look past quantity. Are they delivering ROI—or simply outputs?
2. Prioritize belief and management.
Undertake platforms that steadiness automation with governance, observability, and mannequin flexibility.
3. Spend money on the enablers of scale:
- Area-specialized fashions
- Multimodal UX
- Agentic architectures that develop with you
Kore.ai’s Take
The message is obvious: success in AI received’t come from remoted use instances—it’s going to come from how intelligently and deliberately organizations construct.
At Kore.ai, we’re aligned with Gartner’s imaginative and prescient and proud to help enterprise groups in deploying techniques that aren’t simply generative, however orchestrated, agentic, and prepared for real-world complexity.
In case you missed the keynote, now’s your probability to catch up.
Watch Gartner’s full session on-demand