2025 – The Year of Practical AI: Chain of thought is more powerful, productive, & present in our everyday lives

This Industry Viewpoint was authored by Rajiv Papneja, CTO of Prodapt

Expect many enterprises to realize ROI from their AI investments

This year will be about taking the AI impact from Proof to Value. Businesses will lay a sharp focus on measuring the impact of AI adoption. A new range of customer success KPIs will emerge. For example, ARR (Agentic AI Resolution Rate), First Contact Resolution (FCR) for AI, CSAT for AI interactions, and AHT for AI-assisted interaction and customer effort score (CES) for AI interactions. These benchmarks will evolve with the maturity and success of AI implementations, bringing forth the age of practical AI.

Second – 2025 will see a race to build foundational models quite unlike what we’ve seen in recent years. It has begun, already, with the emergence of DeepSeek on the global GenAI models landscape and the impact to the cost barriers in deploying large scale LLMs. Upstaged, US-based forerunners OpenAI, Meta and Google will invest heavily into LLMs that diversify Chain of Thought, deepen computing power, and democratize availability for everyone – from startups figuring their products to deep-pocketed enterprises toting billion-dollar capex plans. This reality check-infused competition could bring a great leveler in the market as AI capital is no longer moat: cheaper LLMs with smaller GPU requirements will flood the market. This will make AI more consumable and practical for enterprises. The low-cost LLMs will become a true differentiator for organizations keeping the focus on realism through performance versus cost comparisons and ethical considerations.

Here’s a quick view of the top 10 trends of 2025:

1. Industry-contextualized Services as Software (IcSaS): Agentic AI will lead to a massive surge in the softwarization of services. For example, GenAI-powered bots are already transforming contact center operations by intelligently solving customer problems with high levels of decision-making agency. A billing dispute that required the involvement of several human agents with deep experience in Telco billing procedures will be solved in under 4-5 clicks by GenAI-powered agents we deploy in customer environments today. These are services, delivered as software. However, we believe enterprises need technology partners with deep industry expertise – particularly training domain-specific LLMs to deliver industry-contextualized Services as Software (IcSaS). For example, technology providers will build AI Agents on top of Best of Breed software to power near-autonomous incident resolution, and to measure performance and set industry benchmarks, they will coin indices like Agentic AI Resolution Rate (ARR), etc. 
2. Crafted Intelligence for complex enterprise solutions: As the initial wave of enthusiasm around Generative AI matures, enterprises are identifying the sweet spots for sustainable value creation. The future lies in intelligently combining Generative AI, Statistical AI, and classical automation to address intricate enterprise challenges. Tailored “CRAFT” solutions will bridge operational efficiency with customer-centric innovation, ensuring measurable ROI.
3. Artificial General Intelligence – The approaching frontier: This year, forerunners like OpenAI, Google DeepMind and others building advanced AI systems will release products that combine Generative & Agentic AI powers to create products that will act as AI assistants to entire startups – think fractional CMO + Engineering Lead + Product Marketing rolled into one. These advancements, with enough ethical & responsibility checks and balances – will greatly widen AI’s role in the workplace. 
4. Multimodal AI in risk-aware use cases: Multi-modal AI, integrating voice, video, and text capabilities, will continue to evolve with a “safety-first” approach. Regulatory scrutiny around explainability, biases, and privacy—particularly in user profiling—will push operators to deploy these technologies in low-risk domains. For instance, voice and video analytics will enhance operational efficiencies in areas like customer care workflows, while high-risk scenarios will remain under careful assessment.
5. AI Infrastructure goes strategic: As operators shift from Proof-of-Concept (PoC) projects to full-scale production, the focus on AI infrastructure is becoming increasingly strategic. Telcos, for instance, are moving away from ad-hoc budgeting toward unified investments in platforms that handle both AI and network functions, enabling long-term cost efficiencies and innovation. For example, SoftBank, in collaboration with NVIDIA, has pioneered the world’s first AI-RAN (Radio Access Network), seamlessly integrating AI and 5G workloads on the same infrastructure. This approach not only enhances the operational efficiency of base stations but also transforms them into dual-purpose platforms capable of supporting AI-driven services, unlocking new streams of value for telecom operators.
6. Unified Platform vision to power AI agents: This year, enterprises will move decisively toward building a unified, enterprise data & reasoning platform covering every function of their business – IT/Networks, Operations, CX, HR, Finance & more. Partnering with AI-first providers, they will deploy AI agent fleets trained on industry-specific data to solve customers’ problems autonomously, drive highly retrievable internal knowledge bases to democratize information, automate remediation for common issues, and bring predictive abilities to place operations on auto-pilot. Several technology product organizations, including ServiceNow & Salesforce, have begun serious investments and made releases in this direction. With deeper agentic AI adoption, KPIs measuring how they deliver will come into play too. For example, software products tailored for AI-powered customer interactions will measure customer satisfaction and first-contact resolution count of their digital labor force, bringing a competitive differentiation among AI agents. 
7. SDLC Modernization: Products stand out in how fast they move from design & build to test & ship – and with what maturity. AI will catalyze this journey, from automating business requirement gathering & design document creation to GenAI copilots for code generation, test script automation and so on. Technology providers have begun democratizing AI-powered accelerators to streamline and automate the SDLC with Human-in-the-loop capabilities, bringing up to 40% Time to Market reduction for new software product launches. Soon, enterprises will stand apart through successful AI integrations across the SDLC – and piloting AI agents that can execute highly complex actions.\
8. Zero Touch’ Operations (Networks, Field Ops, HR Ops & more): The Connectedness industry – technology enterprises, social network platforms, digital payments companies, telcos, network equipment providers, hyperscalers, etc – will make their networks and operations more autonomous by adopting AI accelerators. From order fulfilment to provisioning & activation to content moderation & user monitoring, hyperautomation of routine tasks will achieve ‘Zero Touch’ quality to these functions. Predicting network events accurately and initiating preventive repair with humans as supervisors has been a long-standing dream of telcos. This year, with AI-powered network automation, this will move closer to reality.
9. Enterprises ring in a sustainable future: Enterprises will embrace corporate sustainability as a core strategy to reduce carbon footprint and deliver eco-friendly services. By transitioning to renewable energy, optimizing network efficiency, and adopting green technologies like energy-efficient data centers and low-power 5G infrastructure, they will minimize environmental impact. Innovations such as LEO satellite networks and digital services reduce resource-intensive operations, while circular economy practices like device recycling and refurbishing promote sustainability. These efforts not only align with global climate goals but also enhance brand reputation, demonstrating a commitment to a greener, more sustainable future.
10. Cybersecurity – the Foremost Need of the Hour: In 2025, as AI-driven technologies and rapid digital transformation redefine industries, cybersecurity has become imperative for telcos and techcos. Innovation at high speed and the proliferation of connected devices and data have amplified security risks. Organizations are prioritizing advanced threat detection, AI-powered incident response, and zero-trust architecture to safeguard their ecosystems. End-to-end encryption, secure edge computing, and compliance with evolving regulations are key to ensure resilience in an increasingly complex threat landscape.

Cautious, foresighted adoption is key

We are clearly amid an AI revolution. Unfettered powers have always resulted in less-than-desired outcomes – be it human or otherwise. Think about a cyber-attack completely thought up and executed by AI agents. The perils are many and evolving.

Enterprises should monitor their AI approach on intent, guardrails and execution to ensure adoption haste does erode privacy, entrench bias and override control. Enterprises should self-impose ethical AI rules and voluntarily publish performance against these benchmarks to drive credibility into their AI approaches. Considering that GenAI workloads consume a lot of compute & storage resources. enterprises need to account for the burden they’re placing on energy grids.

As the AI-native generation, we should look beyond the advantages and comprehend the full nature of AI’s influence on life. We should strike a balance with thoughtful policymaking that will futuristically combine the innate strengths of AI – speed & depth – with the uniquely human quality of persistence.

###

About the Author: Rajiv Papneja is the CTO of Prodapt, an AI-first strategic technology partner providing consulting, business reengineering, and managed services for the largest telecom and tech enterprises building networks and digital experiences of tomorrow.

 

If you haven't already, please take our Reader Survey! Just 3 questions to help us better understand who is reading Telecom Ramblings so we can serve you better!

Categories: Artificial Intelligence · Industry Viewpoint · Software