What if AI could be fast – without being furious to the planet? In the current AI arms race, bigger often means better… and a lot more power-hungry. But when every token costs energy, emissions, and compute, maybe it’s time to question the size-over-smarts approach. Enter the frugal model: smaller, contextual, and good enough to hold its own against the LLM heavyweights, without draining the grid. Let’s challenge the assumption that scale equals value and show how learner, fine-tuned models can achieve surprising results. Call it a new kind of AI performance: less drag race, more precision drift.

The hidden environmental cost of AI: One token at a time

As the AI gold rush accelerates, there鈥檚 a quiet environmental tax we鈥檙e all paying 鈥� one token at a time. A chat query may not feel like much, but beneath the surface lurk LLMs with a rather intimidating computational appetite. The hyperscalers鈥� top-of-the-line LLMs powering today鈥檚 smartest chatbots and emerging AI agents churn through electricity and water in quantities perhaps more befitting an aluminum smelter than a software service.

鈥淥ur GPUs are melting.鈥�

Sam Altman, CEO of OpenAI

Thus, changes in climate are not only driven by the usual suspects, like energy, transport, or agriculture, but increasingly by the digital technologies we embrace. Artificial intelligence, a technology seen by some as a tool to combat climate challenges, has its own environmental footprint. However, the potential of generative AI is undeniable. We shouldn鈥檛 slow down, but we must get smarter in the way we are using it.

Make every token count

That means making every token count: reducing waste, and designing systems that are not just intelligent, but efficient. Because LLMs don鈥檛 just consume data, they consume power. Recent studies show that generative AI, including both model training and user inference, , comparable to the energy use of a small country. By 2027, AI servers could draw as much as 134 TWh/year, roughly the energy needs of Sweden.

Agents on the rise

The real problem, however, may not only come from the models themselves 鈥� but the agents that will be built on top of them. Picture future companies run by a handful of humans and thousands of LLM-powered agents: optimizing code, outwitting other agents in stock trading, crafting legal frameworks, and winning the war for attention with AI-generated content. In this arms race, performance is currency, and currency is performance. The better your agents, the sharper your competitive edge and, today, the top-performing agents are driven by large, cloud-hosted LLMs. They鈥檙e impressive, certainly, but they鈥檙e also expensive financially as well as in terms of sustainability.

Watching Vin Diesel drift around

And the inefficiency adds up. As agents chain together models, expand context windows, and embed documents for every micro-task, we鈥檙e witnessing token inflation on a grand scale. Like watching Vin Diesel drift around a carpark to reach a spot 10 metres away, we鈥檙e using GPT, Gemini, and Claude top models to rephrase a sentence. So, we don鈥檛 need to throw a library at a question that can be answered by a reference to an article.

When querying LLMs, verbosity is expensive. Every token adds cost, complexity, and carbon. Each processed token draws computational power. But not every token needs to be spelled out. Many can be implied, if the model has context. In order to give your agent context and enable downsizing, there are three main strategies.

Train a model from scratch. You train a foundation model from the ground up, using your custom dataset. It will be tailored entirely to your domain but can be extremely expensive and requires massive amounts of time and data.

Finetune an existing model. Start with an existing LLM and retrain the upper level on your domain-specific data. Change the 鈥渨ay鈥� rather than the 鈥渨hat鈥� in LLM responses. A great use case is to finetune a model on generating cypher queries on text prompts (as we will see soon).

Augmented retrieval. Keep the model frozen but supply external knowledge at runtime. That鈥檚 where traditional RAG and its multidimensional cousin GraphRAG makes its entrance, giving the model a shared reference point without overloading it with detail. Rather than cramming background into every prompt, the model can now refer to entities and relationships already mapped in the graph.

A leap of faith: Will a small, context-aware model cut it?

There鈥檚 nothing new about this reasoning, but it takes a little leap of faith to switch from the smoke-and-flash of nitrous-fueled drag racers to the quiet grace of a machine engineered exactly for the task at hand.

This is more than just green tech evangelism: it鈥檚 sound business logic. Using fewer resources is basically good business, delivering reduced latency, lower cost, and independence of data centers and clouds.

The central tension remains: will a small, context-aware model really be enough when it is competing against the full firepower of GPT-whatever running on a nuclear-powered server farm? The battle between efficient precision and brute force brilliance is about to play out here below.

Ladies and gentlemen, start your engines

To find out, we designed a small task to generate cypher queries from natural text, and pitted a finetuned local model of roughly 4 GB against the much larger gpt-4o-mini. To make things more interesting, we also invited the Llama-3.3-70b-versatile from Meta.

The challenge. All three models were fed the same text 鈥� a question regarding information in the underlying graph database or a request to update it. They need to generate a valid cypher query which will be executed, and the responses are then compared.

The underlying database consists of a limited set of startups, technologies, and founders 鈥� linked together based on data from an ecosystem register.

An example query used for benchmarking: 鈥淲hich technologies are used across multiple startups?鈥�

Meet the contenders

Note that the tomasonjo/llama3- text2cypher-demo is finetuned to handle text-to-cypher. It is open source, based on the llama3 model, and can run fully locally on a laptop.

The cloud models are not particularly trained on cypher, but as it is part of the training material, they have a basic understanding of cypher.

Outcome: We ran a number of different queries. The results so far show that the local model and the gpt4o-mini perform on par, with the llama-3.3-70b shows slightly less performance.

Note: The local model didn鈥檛 just keep up 鈥� it outpaced gpt-4o-mini, slashing the generation time by more than 50 percent.

Our benchmark shows that when armed with relevant context, local LLMs can match the performance of massive cloud-based GPT models 鈥� without the weight, the latency, or the energy bill. It鈥檚 a direct challenge to the idea that bigger always means better. With the right architecture, local isn鈥檛 just a fallback 鈥� it鈥檚 a strategic edge. The race for smarter AI is not about who burns the most fuel. It鈥檚 about who handles the corners best.

鈥淎ny intelligent fool can make things bigger, more complex, and more violent. It takes a touch of genius 鈥� and a lot of courage to move in the opposite direction.鈥�

E.F. Schumacher

Local and gpt-4-o-mini are on par.

鈥淟LMs don鈥檛 just consume data, they consume power.鈥�

Start innovating now

Take a leap of faith

and go with smaller models; with the use of knowledge graphs, results become more context-aware, efficient, and environmentally sustainable.

Build processes

for quality assurance and automatic benchmark probing with large models.

Continuously measure

and communicate reductions in energy usage and carbon emissions resulting from data operations, fostering sustainable AI practices from the start.

This article explores how Agentic AI can help redesign the operating model, shift decision-making, and unlock entirely new ways of creating value 鈥� all by thinking from the future, not just about it.

Most companies tweak. The bold ones transform. But a new mindset is emerging 鈥� one that goes full hacker-mode on the enterprise itself. No legacy constraints, no silos, no sacred cows. What if the business were rebuilt from scratch, as if it were being launched five or ten years from now, with Agentic AI not as a plug-in, but as a strategic co-founder? This isn鈥檛 about optimization 鈥� it鈥檚 about reimagining the purpose, structure, and intelligence of the organization from the ground up.

Somewhere between your last strategic offsite and your current AI working group, the business world changed. Again. Only this time, it wasn鈥檛 a new framework or another agile manifesto. This time, it鈥檚 a new brain.

Agentic AI doesn鈥檛 just automate what you already do鈥攊t questions why you do it in the first place. It proposes. It critiques. It simulates. It rewires how decisions are made, who makes them, and when they happen. And unlike your last re-org, it doesn鈥檛 wait six months to show any impact.

The problem is that most organizations are still trying to install AI into 20th-century workflows like it鈥檚 a plugin. 鈥淟et鈥檚 just optimize procurement鈥� becomes a board-approved project that takes 18 months, involves many large teams, and results in slightly faster procurement.

But what if you stopped optimizing? What if you hacked your own business?

Design from 2030, Not 2020

Let鈥檚 imagine, just for a moment, that you鈥檙e not burdened by legacy systems, organizational charts, compliance rituals, or 鈥渨e鈥檝e always done it this way鈥� syndrome. Imagine you鈥檙e starting from scratch. The year is 2030. Agentic AI is not just embedded in your processes鈥攊t鈥檚 embedded in your people. Or rather, your people are embedded in a new kind of system: one where intelligence is ambient, decisions are real-time, and the line between human and machine is not blurred鈥攊t鈥檚 collaborative.

Would you really recreate the same silos? Would you copy-paste last year鈥檚 operating model into the future and call it transformation?

In this future-forward model, decisions don鈥檛 just happen faster鈥攖hey happen better. Picture an AI agent that pulls real-time data from your supply chain, simulates three economic scenarios, aligns them with your top KPIs, and taps your planning team with a message like: 鈥淚f we shift production to Mexico for the next three weeks, we protect margins and avoid inventory bottlenecks. Proceed?鈥� It鈥檚 not fantasy. It鈥檚 already emerging in pilot programs. Retail, logistics, finance鈥攖hey鈥檙e not waiting for permission. They鈥檙e experimenting. Quietly, sometimes clumsily, but they鈥檙e moving.

What鈥檚 missing is boldness at scale. The audacity to say: 鈥淚f AI can hack my business, so can I.鈥�

Rebuilding your business from a blank slate sounds like the stuff of retreats and vision decks. But it鈥檚 more than that. It鈥檚 a challenge to strip things down to first principles. What decisions truly matter? Who needs to make them? What should be instantaneous, and what still needs reflection? Where do humans shine鈥攁nd where do they hold things up?

From Workflows to Intelligence Networks

The old mindset builds workflows. The new mindset builds networks. You don鈥檛 need another dashboard. You need a decision system. One that鈥檚 alive, adaptive, and鈥攜es鈥攕ometimes smarter than you.

When you start thinking this way, the org chart begins to look suspiciously like a museum exhibit. A relic from a time when communication flowed one way, and information took the scenic route. In an agentic enterprise, power is less about title and more about your ability to navigate systems, interpret signals, and co-create with machines. Managers stop managing tasks and start designing interactions. Strategy is no longer a quarterly slide deck鈥攊t鈥檚 a living process that evolves in real time, shaped by continuous inputs and autonomous agents that never sleep.

And leadership? Leadership becomes less about having the answers, and more about asking the right questions. Less about command and control, more about trust and iteration.

Yes, it sounds radical. So did the cloud. So did putting your ERP in someone else鈥檚 data center. So did letting your intern post on LinkedIn. But here we are. Still talking about 鈥減ilots鈥� and 鈥渦se cases,鈥� while the real opportunity is to reimagine the business entirely.

AI-native government

An example? In government, an AI-native model wouldn鈥檛 just digitize existing services 鈥� it would reimagine how policy is shaped and delivered. Agentic AI can simulate the real-time impact of legislation across demographics, recommend adjustments to improve equity or efficiency, and even co-design citizen services based on behavioral signals, not just historical data. It鈥檚 not bureaucracy with better bandwidth. It鈥檚 governance with built-in intelligence.

A less embarrassing future

So, if yet another AI pilot is on the table 鈥� same org, same roles, same handoffs 鈥� it might be time to pause. Because layering new tech on old thinking rarely leads to transformation. Instead, ask the only question that truly matters: What would we build, if we started today, from five years in the future? Not to optimize what’s already there, but to rethink what should be there. In that future-first mindset, Agentic AI becomes more than a tool 鈥� it鈥檚 a co-architect of the business itself. And the goal isn鈥檛 to make the past slightly more efficient. It鈥檚 to make the future slightly less embarrassing.

Start Innovating Now

Audit Your Decisions

Choose a function鈥攍ike finance, supply chain, or marketing鈥攁nd map the ten most frequent decisions made each week. Then ask: which of these could be delegated to an AI agent, and which require uniquely human judgment? The results will surprise you.

Pilot a Human + Agent Workflow

Redesign a single, low-risk business process with an AI agent embedded in the loop. Don鈥檛 aim for full automation鈥攋ust real-time collaboration. Try demand forecasting, contract review, or pricing adjustments. Measure speed, quality, and trust.

Build Your 鈥淎I from Scratch鈥� Blueprint

Assign a cross-functional team to answer: 鈥淚f we rebuilt this business in 2030, with no legacy and full Agentic AI capability, how would we run operations, make decisions, and structure teams?鈥� Document it. Then look for one idea you can implement now.

Interesting read? 乌鸦传媒鈥檚 Innovation publication, Data-powered Innovation Review – Wave 10 features more such captivating innovation articles with contributions from leading experts from 乌鸦传媒. Explore the transformative potential of generative AI, data platforms, and sustainability-driven tech. Find all previous Waves here. 聽Find all previous Waves here.

ETL used to be a puzzle only data engineers could solve 鈥� with enough time, code, and coffee. Now generative AI is quietly rewriting the rules.

Traditional pipelines鈥攕low, brittle, and endlessly maintained鈥攁re giving way to 脴 ETL: adaptive, intelligent flows that go from prompt to pipeline in seconds. No scripts, no tickets, no heroic debugging sessions. This article explores how data integration is becoming faster, smarter, and far more democratic 鈥� turning engineers into orchestrators and putting AI to work where it actually makes sense.

A shift from manual to machine-driven

ETL has always been about moving data from different sources, transforming it into a usable format, and loading it into a system where it can be analyzed. But with the explosion of data, increased regulatory pressure, and the move to hybrid and multi-cloud environments, this process has become much more complex.

Generative AI is changing the game. Instead of writing and maintaining endless scripts and workflows, organizations can now use AI models to automate ETL pipelines. These models understand the context of the data, learn from existing integration patterns, and generate optimized workflows on the fly.

This leads to significant benefits:

路 Speed: AI can generate and update ETL logic in minutes, not days.

路 Consistency: AI-driven pipelines are less prone to human error.

路 Adaptability: They automatically adjust to schema changes or new data sources.

Beyond automation: intelligent integration

AI isn鈥檛 just speeding things up鈥攊t鈥檚 making data integration smarter. By applying natural language understanding, organizations can describe what they want in plain English, and the AI creates the integration pipeline.

For example, a demand planner can say: 鈥淓xtract product inventory from Oracle, combine it with daily sales from Shopify, calculate stock turnover per SKU, and load it into Snowflake for reporting.鈥� With traditional ETL, manual SQL logic, batch jobs, and schema mapping must be created. AI will generate the pipeline on demand from the prompt.

This approach democratizes data integration. It removes the dependency on specialized engineers for every change and helps more people in the organization work with data directly.

How this fits in a modern data strategy

Generative AI for ETL is a natural fit in environments where data fabrics or data mesh architectures are being implemented. Modern data strategies are shifting from centralized control to decentralized ownership. Concepts like data mesh and data fabric are driving this shift, giving teams more flexibility to manage, consume, and share data across systems. In these models, every domain owns its data products, but the organization still needs consistency, compliance, and efficiency at scale.

Supporting decentralization without losing control

In a data mesh, teams manage their own pipelines. Traditional ETL tools can鈥檛 keep up with the constant change and complexity. AI-driven ETL supports this by giving each team a way to build and manage data flows independently鈥攚ithout starting from scratch or involving a central data engineering group every time.

Cross-cloud compatibility

Leading platforms are already moving in this direction:

路 Google Cloud: With services like BigQuery Dataform and Cloud Data Fusion, Google supports declarative and visual data pipeline development. Generative AI models from Google鈥檚 Vertex AI can integrate with these services to streamline data prep and transformation.

路 AWS: Amazon鈥檚 Glue Studio offers low-code/no-code pipeline development, and new AI integrations allow users to describe what they want in natural language. Combined with SageMaker and Bedrock, AWS is aiming to simplify the entire data lifecycle鈥攆rom ingestion to modeling.

路 Microsoft Azure: Azure Data Factory and Synapse Analytics are embedding AI directly into pipeline creation and monitoring. With Microsoft Copilot, users can ask for transformations, lineage, and integration logic using natural language.

路 Databricks: With its Lakehouse architecture, Databricks is adding AI to simplify pipeline generation in notebooks and workflows. Unity Catalog, when paired with LLMs, supports context-aware data discovery and security enforcement.

路 Snowflake: Their growing suite of AI features, including Snowpark and Cortex, allows SQL and Python users to automate parts of the data prep process. With Snowflake鈥檚 native LLM support, the platform is well-positioned to offer AI-driven transformations at scale.

路 Open-source & hybrid platforms: Tools like Apache Airflow, Dagster, and dbt are starting to explore AI plugins and extensions. These add automation and intelligence to open workflows, making it easier for developers to generate and maintain pipeline logic.

What’s next?

We are moving toward a future where ETL as we know it may no longer exist. Instead, we鈥檒l see dynamic data integration powered by AI. The concept of 鈥淓TL pipelines鈥� will be replaced by intelligent agents that continuously ingest, transform, and validate data in real time, guided by policies and context, not hardcoded rules.

For organizations, this means that by embedding generative AI into ETL processes across platforms:

路 Time to value shortens: Data products go live faster, helping teams act quickly.

路 Complexity reduces: AI handles edge cases, schema drift, and exception handling in real time.

路 Data quality improves: Built-in rules and real-time validation become part of the generated logic

路 Business access increases: More users across domains can work with data confidently, without needing to be engineers.

This isn鈥檛 just a technological shift. It鈥檚 a change in how we approach data鈥攎oving from pipelines built manually to systems that can learn, generate, and adapt automatically. Remember that AI isn’t replacing data engineers鈥攊t’s changing their role. The most successful organizations are those that help their teams adapt to becoming orchestrators and quality managers rather than code writers.

Maybe 脴 ETL doesn鈥檛 mean 鈥渘o ETL鈥� 鈥� but it definitely means no more business-as-usual. As AI takes over the heavy lifting, data engineers get to step back from pipelines and step up to strategy. The script is changing, the prompt is the new interface 鈥� and the future of data integration might just be zero-code, zero-friction, and all impact.

Start innovating now

• Experiment with AI-Driven Data Pipelines: Start small with a well-defined ETL use case where manual processes create bottlenecks. Try implementing generative AI to automate a non-critical data flow and measure the time savings and accuracy improvements.

• Invest in Data Literacy and Documentation: Improve your metadata management and data documentation practices. High-quality documentation significantly enhances how well AI tools understand your data relationships and can generate appropriate transformations.

• Upskill Your Data Teams: Help your data engineers transition from code writers to pipeline architects and quality experts. Create opportunities for them to work alongside AI tools while developing new skills in oversight and optimization.

Interesting read? 乌鸦传媒鈥檚 Innovation publication, Data-powered Innovation Review – Wave 10 features more such captivating innovation articles with contributions from leading experts from 乌鸦传媒. Explore the transformative potential of generative AI, data platforms, and sustainability-driven tech. Find all previous Waves here. 聽Find all previous Waves here.

Across the enterprise, departments are placing increased demands on their organization鈥檚 data to enable multi-AI agents. It鈥檚 the IT operations (IT ops) department鈥檚 challenge to deliver the optimal environment for agentic AI to eventually bring business value.

Enterprises are grappling with a volatile, uncertain business climate 鈥� and to address this, they are increasingly turning to their data to draw actionable insights that enable competitive advantages through agents.

As networks grow more complex and the demands on them increase, IT ops departments need to develop better tools, including multi-AI agent systems to enhance the decision-making process by making recommendations aligned to set business goals.

Properly designed and implemented agentic AI solutions are game-changers 鈥� but IT ops must be prepared to take advantage of these powerful tools, which requires a well-crafted plan and a partner that can deliver more than just the technology.

The IT ops imperative

In conversations with IT professionals, my 乌鸦传媒 colleagues and I have identified a number of common challenges for IT operations at enterprises across all sectors. Simply stated, IT professionals are under pressure to boost service performance while reining in costs 鈥� including operating expenses and costs for infrastructure and cloud services. They鈥檙e also under pressure to better identify, provision, and deploy the solutions required to allow other departments to take advantage of emerging technologies such as agentic AI.

The organization鈥檚 own data is an important source of the information required to help IT professionals achieve these goals. Unfortunately, legacy business intelligence systems often fail to satisfy their needs. There are several reasons for this:

• Analytics systems rarely support strategic foresight and transformative innovation 鈥� instead providing business users with yet another dashboard.
• The results are often, at best, a topic for discussion at the next team meeting 鈥� not sufficient for a decision-maker to act upon immediately and with confidence.
• Systems typically fail to personalize their output to provide insights contextualized for the person viewing them 鈥� instead offering a generic, unsatisfying result.
• Systems often aggregate data within silos, which means their output still requires additional interpretation to be valuable.

In short, many legacy systems miss the big picture, miss actionable meaning, miss the persona 鈥� and miss the point.

Based on my experience, I recommend an organization address this through multi-AI agent systems.

With the introduction of Gen AI Strategic Intelligence System by 乌鸦传媒, this could be the very system that bridges the gap between the old way, and a value-driven future. This system converts the vast amounts of data generated by each client, across their enterprise, into actionable insights. It is agentic: it operates continuously and is capable of independent decision-making, planning, and execution without human supervision. This agentic AI solution examines its own work to identify ways to improve it rather than simply responding to prompts. It鈥檚 also able to collaborate with multiple AI agents with specialized roles, to engage in more complex problem-solving and deliver better results.

How would organizations potentially go about doing this?   

Define the technology and business KPIs

First, organizations must establish well-defined KPIs and associated roadmaps to take full advantage of agentic AI recommendations 鈥� KPIs that align technology with business objectives.

This starts by identifying the end goals 鈥� the core business objectives and associated KPIs relevant to IT operations. These represent the IT operation鈥檚 key activities that support other departments as they contribute to the organization鈥檚 value, and strengthening them is always a smart exercise. The good news is that even small improvements to any of these KPIs can deliver enormous benefits.

The roadmap should leverage pre-existing AI models to generate predictive insights. It should also ensure scalability, reliability, and manageability of all AI agents 鈥� not just within the realm of IT operations, but throughout the organization. And it should be designed to leverage domain-centric data products from disparate enterprise resource planning and IT systems.

Finally, the roadmap must identify initiatives to ensure the quality and reliability of the organization鈥檚 data by pursuing best-in-class data strategies. These include:

• Deploying the right platform to build secure, reliable, and scalable solutions
• Implementing an enterprise-wide governance framework
• Establishing the guardrails that protect data privacy, define how generative AI can be used, and shield brand reputation

Choose a partner that delivers more than tech

Second, the organization must engage the right strategic partner. While innovative agentic AI systems are essential, that partner must also be able to support the IT team with business transformation expertise and industry-specific knowledge.

乌鸦传媒 leverages its technology expertise, its partnerships with all major platform providers, and its experience across multiple industrial sectors to design, deliver, and support agentic AI strategies and solutions that are secure, reliable, and tailored to the unique needs of its clients.

乌鸦传媒鈥檚 solution draws upon the client鈥檚 data ecosystem to perform root cause analysis of KPI changes, and then generates prescriptive recommendations and next-best actions 鈥� tailored to each persona within the IT department. The result is goal-oriented insights aligned with business objectives, ready to help IT empower the organization through actionable roadmaps for sustainable growth and competitive advantage.

*Meaningful, measurable benefits

乌鸦传媒 estimates that with the right implementation and support, the potential benefits include augmenting the IT workforce through autonomous processing, touchless data crunching, improved data and systems integrations, continuous monitoring of controls and compliance, and real-time access to reports and insights.

The potential for IT operations to translate these internal gains into meaningful advantages for other departments across the enterprise means that leveraging agentic AI for its own strategic insights cannot be ignored.

^*Results based on industry benchmarks and observed outcomes from similar initiatives with clients. Individual results will vary.

The Gen AI Strategic Intelligence System by 乌鸦传媒 works across all industrial sectors, and integrates seamlessly with various corporate domains.聽Download our PoV聽here聽to learn more or contact our below expert if you would like to discuss this further.

Google Cloud and 乌鸦传媒 brought together more than 1,800 innovators from around the globe for the Google Cloud Agentic AI Hackathon 

With 93 percent of business leaders  believing that scaling AI agents in the next 12 months will provide a competitive edge, according to the 乌鸦传媒 Research Institute, it was paramount for 乌鸦传媒 and Google Cloud to come together to help customers harness the promise of multi-agents systems and provide a pragmatic way to deploy them at scale  

In this context, by conducting our proven Google Cloud Hackathon for three years in a row, our motivation was higher than ever to develop a suite of repeatable agents capable of solving real-world business challenges.   

The event used Google Cloud鈥檚 latest agentic AI tools available in Vertex AI to deploy multi-agent systems with AgentSpace, Agent Development Kit (ADK), and Agent Engine to build solutions for 23 real-life use cases submitted by 乌鸦传媒 clients, following the newly released Agent-to-Agent (A2A) standard. 

Over three weeks, hackers explored how intelligent agents can reason, act autonomously, and adapt to complex environments and how they can be applied to real-world business challenges. 

This was a proving ground for how companies can move from experimentation to execution with agentic AI, while fostering a culture of learning and human-AI collaboration.   

A strategic collaboration for innovation 

The hackathon represented our partnership in action. It leveraged Google Cloud鈥檚 advanced AI technologies and 乌鸦传媒鈥檚 deep industry expertise to co-create solutions that are not only technically robust but also business ready. 

Our original goal was to develop more than 200 agents on Google鈥檚 latest agentic AI platforms. The response exceeded expectations:  256 teams from 39 countries, created more than 320 unique agents during the event. Forty-nine mentors and judges were mobilized to select standout, customer-ready innovations.  

Breakthrough solutions with real-world potential  

The use cases included advanced solutions from a wide range of industries and were judged on innovation, feasibility, and alignment with business needs. 

Here are the nine standout projects from the final round. 

• Aerospace: An agentic AI-powered multi-agent system orchestrates the end-to-end requirement validation process by extracting key data from PDFs and IBM DOORS. It then refines and curates requirement statements, analyzing gaps and inconsistencies, and generates comprehensive, actionable validation reports.
• Consumer products and retail: AI agents optimize procurement, reduce waste, and manage inventory to meet sustainability goals. 
• Automotive and manufacturing: An AI system that automates supply chain and manufacturing to cut delays and costs with proactive decision-making. 
• Consumer products and retail: A retail analytics system examines sales data, advising on new promotion strategies and putting together new marketing material. 
• Banking and insurance: A contact center tool that fetches customers data, suggests live actions, and recommends next steps. 
• Public service: An assistant that simplifies public service access with easy sign-up and step-by-step help. 
• Telecommunications: The AI system detects service issues, recommends fixes, and sends alerts for faster support.
• Banking and insurance: An IT assistant that automates ServiceNow tasks like password resets and triaging to reduce help desk load. 
• Public service: This health tool personalizes check-ins and surveys to monitor patient well-being and trigger alerts. 

Several projects are now advancing to MVP piloting and client co-innovation tracks. The list of agents will also be made available to 乌鸦传媒 clients through our Group AI Agents gallery as part of our RAISE (Reliable AI Solution Engineering) agent accelerator. 

Learning through doing 

More than 96 percent of participants completed the Google Agentic AI learning path. The hackathon became a live learning lab, combining structured enablement with hands-on experimentation. 

Bootcamps, workshops, and mentorship from 乌鸦传媒 and Google Cloud SMEs ensured that every team had the support they needed, setting 乌鸦传媒 as one of the largest Google Agentic enabled partners ready to meet the pressing market demand. 

Unlocking AI at scale  

AI can reimagine business but organizations need to scale to really unlock the full potential and uncover real business benefits. Hackathons are one pathway to AI readiness, by providing training and the opportunity to experiment with the technology as well as build working agents. Implementing agentic AI requires a high level of AI readiness, while creating the right human-AI chemistry to ensure lasting adoption.  

The Resonance AI Framework by 乌鸦传媒 helps leaders envision AI鈥檚 potential, embed it into the foundation of their operations, and enable human-AI chemistry. It is designed to allow effective interaction between people and intelligent systems, and creates the trust, understanding, and collaboration needed for human and AI agents to build reliability over time, ensuring hybrid teams thrive. This democratization of AI empowers businesses to embed AI into the fabric of everyday operations.聽聽

A culture of collaboration and intrapreneurship 

The hackathon fostered experimentation and cross-functional collaboration. Teams were assembled from different business lines, geographies, and technical backgrounds. This diversity of thought was a key driver of success. 

The hackathon was structured in four phases: 

1. Onboarding, use case definition, and account selection 
2. Client onboarding and business and technical scoping 
3. Training and hackathon program execution 
4. In-production workshops. 

This framework enabled creativity to flourish within clear guardrails and ensured that promising ideas could transition into actionable prototypes. 

Watch the highlights

See the energy, creativity, and impact firsthand

This video features highlights from the live sessions, interviews with participants, and demos of the winning solutions.  

Customizing hackathons: What this means for the enterprise

A hackathon is not only a fantastic innovation hub, it鈥檚 also an opportunity to engage our employees from around the world. It provides insights into real-life use cases as well as upskilling knowledge and building culture, and shows how to stay ahead of the competition with new ideas. 

With our breadth of experience, 乌鸦传媒 can work with clients looking to explore their own internal hackathon, helping define and prioritize agentic use cases specific to their needs, and upskill employees with challenge-based learning to accelerate skill development and adoption of emerging technologies, empowering teams to experiment and collaborate fosters long-term transformation. 

We can also help explore the possibilities and partner with Google Cloud for strategic collaboration to accelerate business outcomes. 

Looking ahead 

This competition brought together developers, designers, business analysts, and others to deliver multiple points of value: 

• Upskilled 乌鸦传媒 talent with hands-on learning and certifications on the latest Google Cloud technologies 
• Expanded our AI agent gallery available to clients 
• Supported AI at scale as part of our Resonance Framework. 

Our goal is now to empower clients to accelerate transformation through intelligent, autonomous systems grounded in human-AI collaboration. A hackathon is just one of the many tools we have available to enable AI-powered enterprises. 

To explore how agentic AI and Google Cloud鈥檚 generative capabilities can accelerate innovation in your organization, reach out to our Google Cloud experts. Whether you’re looking to pilot a solution, scale a use case, organize a hackathon, or build a roadmap for transformation, we鈥檙e here to help you take the next step, wherever you are in the journey.  

Gen Garage is redefining innovation by harnessing AI to build real-world solutions in areas such as disaster management, sustainable farming, and environmental risk mitigation

By fostering talent transformation and embracing cutting-edge technologies, we empower tomorrow鈥檚 professionals to shape the future through hands-on impact. Step into Gen Garage, 乌鸦传媒鈥檚 乌鸦传媒 & Data innovation hub 鈥� where visionary minds, guided by expert mentors, craft transformative AI-driven solutions at the intersection of talent and technology. Fueled by generative AI, machine learning, and automation, Gen Garage accelerates operational excellence and delivers innovations that drive efficiency, inclusivity, and sustainability. As businesses lean into data-powered insights, our solutions stay aligned with evolving needs 鈥� helping organizations stay future-ready, while making a difference today.

DisasterX: AI-Powered Faster and Smarter Response for Disaster Management

DisasterX is a cutting-edge AI-powered application designed to revolutionize disaster response by enabling real-time, adaptive decision-making in high-pressure environments. Leveraging agentic AI, DisasterX overcomes the challenge of delayed responses by autonomously analysing vast amounts of real-time data from multiple sources. This allows for optimized rescue efforts, efficient resource allocation, and enhanced recovery strategies. By continuously learning from past disaster scenarios, DisasterX improves prediction accuracy, minimizes human error, and accelerates response times鈥攗ltimately saving lives and reducing economic and environmental impact. Its autonomous capabilities ensure scalability and resilience, seamlessly adapting to both local and large-scale emergencies. As the rise of agentic AI reshapes automation, DisasterX stands at the forefront of intelligent, proactive disaster management, delivering greater efficiency and reliability in crisis situations.

Picture 1 : An AI agent (Image credit 鈥� Pixabay)

KisanGPT: AI-Driven 乌鸦传媒 for Smarter, Sustainable Farming

KisanGPT is an AI-powered platform designed to revolutionize farming by providing real-time insights on crop health, weather forecasts, and sustainable agricultural practices. Using advanced language models, satellite data, and weather analytics, it offers personalized recommendations to help farmers optimize yields, conserve resources, and tackle climate challenges. The platform supports multilingual access and speech recognition, ensuring inclusivity for farmers across diverse regions. By integrating market trends, government policy updates, and best farming practices, KisanGPT enhances decision-making, boosts profitability, and promotes eco-friendly agriculture. This AI-driven solution not only improves efficiency but also fosters a more resilient and sustainable farming ecosystem.

The Green Horizon: AI-Powered Vegetation Hazard Management

The Green Horizon is an AI-powered solution that detects and manages vegetation hazards near power lines, preventing wildfires, outages, and safety risks. Using satellite imagery, machine learning, and weather forecasting, it provides real-time monitoring, predictive insights, and proactive risk mitigation. With an intuitive chatbot and geospatial analytics, it empowers organizations to optimize resources, reduce costs, and ensure safer, more sustainable infrastructure. By automating hazard detection and integrating user feedback, it enhances decision-making for utility companies and environmental agencies. This innovative approach not only improves operational efficiency but also supports long-term sustainability and infrastructure resilience.

Market Trends / Key Opportunities and Developments:

Gen Garage strategically aligns its initiatives with prevailing market trends to address pressing societal and business needs.

The increasing investment in AI for disaster management presents a significant market opportunity for DisasterX to deliver innovative and data-driven solutions. With the rise of smart cities and the widespread adoption of IoT sensors in disaster-prone areas, vast amounts of real-time data can be leveraged for predictive analytics and rapid response. Gen Garage is at the forefront of this transformation, utilizing AI to enhance disaster preparedness and resilience. As climate change intensifies the frequency of natural disasters, the demand for intelligent, automated response systems continues to grow, positioning DisasterX as a key player in optimizing disaster mitigation and emergency management strategies.

KisanGPT taps into the growing demand for AI-driven agricultural solutions. By leveraging real-time analytics and precision farming techniques, Gen Garage maximizes market opportunities, helping farmers and agribusinesses adopt smarter, data-driven strategies. With advancements in AI and increasing support for sustainable farming practices, the platform positions itself as a game-changer in modern agriculture, driving innovation and long-term growth in the sector.

The Green Horizon initiative taps into the growing need for AI-driven environmental risk management. By integrating geospatial intelligence and predictive analytics, Gen Garage maximizes market opportunities, enabling utility companies and agencies to adopt smarter, data-driven strategies for sustainability and infrastructure resilience.

Gen Garage is where innovation gets hands-on 鈥� and where emerging talent learns by doing. By combining mentorship with real-world problem-solving, we鈥檙e helping young professionals grow into AI changemakers while delivering solutions that matter. From climate-smart farming to disaster response, the Garage proves that AI for good isn鈥檛 just a concept 鈥� it鈥檚 a daily practice. The challenges may be big, but with the right mix of curiosity, code, and collaboration, we鈥檙e building something that lasts. Stay tuned in the next edition of the Data-powered Innovation Review for more recent cases!

Start innovating now 鈥�

Empower Future Talent

Get involved in innovation projects that enhance AI skills and leadership capabilities, preparing young professionals for real-world challenges.

Leverage AI for Social Impact

Adopt AI solutions to drive sustainability, inclusivity, and efficiency across industries, from disaster management to smart farming.

Stay Ahead of Market Trends

Engage with cutting-edge technology and AI-driven insights to maintain a competitive edge in an evolving digital landscape.

Interesting read? 乌鸦传媒鈥檚 Innovation publication, Data-powered Innovation Review – Wave 10 features more such captivating innovation articles with contributions from leading experts from 乌鸦传媒. Explore the transformative potential of generative AI, data platforms, and sustainability-driven tech. Find all previous Waves here. 聽Find all previous Waves here.

Climate change isn’t a distant threat鈥攊t’s a reality to deal with now.

Businesses need to rethink how they operate, especially when it comes to supply chains, which are crucial for global trade. Just like in the movie Interstellar, where survival depended on data, AI, and adaptability, today’s supply chains need to be flexible and smart to handle disruptions and climate challenges. AI-powered insights and actions are like the movie鈥檚 robot TARS: helping predict risks, optimize logistics, and reduce waste. Data ensures that every decision is as precise as a gravity equation. AI enhances precision in supply chains by analyzing vast data in real time, predicting risks, and optimizing logistics. It鈥檚 the key to transforming supply chains into smarter, greener, and more resilient systems that balance profitability with ecological responsibility.

Supply chains aren鈥檛 just stretched 鈥� they鈥檙e under siege. Disruption is no longer the exception; it鈥檚 the norm. That鈥檚 why resilience 鈥� the ability to anticipate, adapt, and recover fast 鈥� has shifted from nice-to-have to non-negotiable. A recent report from Institute delivers the reality check: 80% of organizations faced supply chain disruptions last year, most more than once. That鈥檚 an uptick despite better planning 鈥� proof that we鈥檙e still reacting more than we鈥檙e preparing. Meanwhile, sustainability pressures are mounting. With supply chains responsible for over 60% of global carbon emissions, according to the World Economic Forum, they鈥檙e no longer just operational engines 鈥� they鈥檙e climate liabilities too.

Let鈥檚 face it鈥攚hat we鈥檙e doing right now isn鈥檛 cutting it. The cracks in our supply chains are showing, and incremental fixes won鈥檛 be enough. It鈥檚 time for bold moves. If we want supply chains that can truly withstand shocks and stay ahead of the curve, we need to lean into smarter, faster, more adaptive solutions. That鈥檚 where AI steps in鈥攏ot just as a tool, but as a game-changer. With its ability to forecast disruptions, optimize operations, and accelerate response times, AI is shaping the supply chains of the future. To stay ahead, companies must embrace green supply chain management (GSCM), where sustainability is built into every step. AI supercharges this shift, turning GSCM into a smart, data-driven engine. From cutting carbon to driving circular economies, AI enables supply chains that are not just efficient, but truly green.

Resilience, not yet autonomous: Supply chains still heavily rely on people

Supply chains are navigating a perfect storm: geopolitical instability, extreme weather, shifting consumer expectations 鈥� and growing uncertainty in global trade. Disruptions are no longer outliers; they鈥檙e part of the operating environment. While many organizations are embedding risk management into supply chain strategy, execution is still stuck in manual mode. Too much effort goes into collecting, cleaning, and stitching together data 鈥� leaving little room for insight, foresight, or speed. AI and machine learning are still underused, and critical response actions often rely on human intervention alone. The result? Slow reactions, mounting workloads, and talent focused on firefighting instead of forward-thinking.

What鈥檚 missing? Technology that doesn鈥檛 just capture and store data, but actively turns it into prescriptive insights and clear, actionable recommendations. Unfortunately, most tools in the market today still fall short of that promise. Instead, businesses are left stitching together manual processes and siloed teams to make sense of a rapidly changing environment. To build truly resilient supply chains, we need to shift from reactive, human-heavy models to intelligent, tech-augmented systems. The future isn鈥檛 about replacing people鈥攊t鈥檚 about empowering them with tools that amplify their decision-making, speed up response times, and free them to focus on what matters most.

Greening the chain: How AI and data are changing the game

Data and AI are at the core of this transformation, delivering unmatched insights, predictive accuracy, and optimization potential. By leveraging real-time data and predictive analytics, AI can identify potential risks鈥攕uch as supplier delays, extreme weather, or geopolitical issues鈥攂efore they impact operations. This early warning capability allows businesses to proactively mitigate threats through alternative sourcing, dynamic rerouting, or inventory adjustments. AI also enables scenario modeling, helping organizations test various disruption scenarios and build contingency plans with data-backed confidence. As a result, companies can maintain continuity, reduce downtime, and ensure customer satisfaction, even in the face of unexpected challenges. In today鈥檚 volatile global environment, AI is no longer a luxury but a critical enabler of resilient and future-ready supply chains.

AI-enhanced supply chain resilience framework

The AI-enhanced supply chain resilience framework strengthens supply chain agility and robustness by harnessing advanced AI technologies. It integrates real-time data from IoT devices into a centralized system for comprehensive analysis. Through predictive analytics and machine learning, the framework forecasts demand and detects potential risks鈥攍ike supplier disruptions or market shifts鈥攅nabling proactive risk mitigation and smarter decisions in areas like inventory and logistics.

AI-driven communication tools improve collaboration with suppliers and stakeholders, ensuring seamless, transparent information flow. Continuous monitoring and adaptive feedback loops allow the supply chain to respond swiftly to changing conditions, driving ongoing improvement and innovation. By adopting this framework, businesses gain end-to-end visibility, reduce vulnerabilities, and ensure operational continuity鈥攗ltimately building a more resilient and high-performing supply chain.

Leveraging AI enables businesses to streamline operations, improve efficiency, cut costs, and elevate customer experiences. One powerful application is demand forecasting, where AI analyzes historical data to accurately predict customer needs. This leads to smarter inventory management鈥攎inimizing overstock and stockouts while optimizing capital use. Another key use case is route optimization. AI-driven tools evaluate factors like weather, traffic, and transport costs to determine the most efficient delivery paths. This reduces time and expenses while ensuring faster, more reliable service that meets growing customer expectations.

How organizations can harness it effectively:

According to the , 55% of Forbes Global 2000 OEMs are projected to have revamped their service supply chains with AI and by 2026, 60% of Asia based 2000 companies will use generative artificial intelligence (GenAI) tools to support core supply chain processes as well as dynamic supply chain design and will leverage AI to reduce operating costs by 5%.鈥疶his鈥痵ignifies a widespread adoption of AI to improve efficiency and gain a competitive advantage in supply chain management. Further, Generative AI can be harnessed to monitor global events and proactively identify emerging risks. It can automatically generate risk assessments, simulate scenarios, and suggest strategic mitigation plans鈥攅mpowering supply chain teams to manage risks more effectively. Its conversational interface enhances user experience and accelerates response times. Over time, this evolves into a system-guided, data-driven approach, drawing from a rich library of scenarios and mitigation strategies to deliver contextual, timely responses to risk events.

Considering all of the facts

The fusion of data and AI isn鈥檛 just a tech upgrade 鈥� it鈥檚 a strategic shift for building supply chains that can bend without breaking. Organizations that embed intelligence into their operations now won鈥檛 just survive the next disruption 鈥� they鈥檒l lead the transition to greener, faster, more adaptive ecosystems. By 2025, global supply chains will be reengineered out of necessity and powered by innovation. AI won鈥檛 just help companies 鈥� it will help nations stay resilient, competitive, and climate-conscious. It will redefine how we make, move, and manage everything. And like TARS in Interstellar, the most effective systems won鈥檛 just follow instructions 鈥� they鈥檒l anticipate, adapt, and act as true copilots. What supply chains need now isn鈥檛 just visibility. It鈥檚 vision.

Start innovating now 鈥�

Give your supply chain an AI-enabled sixth sense

• Plug your supply chain into real-time feeds鈥攆rom IoT sensors to storm trackers鈥攁nd let AI act like your all-seeing oracle. Spot trouble (like delayed shipments or political curveballs) before it hits the fan

Make generative AI your strategic co-pilot

• Leverage Generative AI to generate real-time risk assessments, simulate disruption scenarios, and recommend mitigation strategies, all in a conversational interface

Build a digital twin鈥攜our virtual supply chain lab

• Think of it as a flight simulator for your supply chain. A digital twin lets you mirror operations in a virtual space to test 鈥渨hat-if鈥� scenarios鈥攆rom port delays to carbon constraints鈥攚ithout breaking a sweat in real life.

Interesting read? 乌鸦传媒鈥檚 Innovation publication, Data-powered Innovation Review – Wave 10 features more such captivating innovation articles with contributions from leading experts from 乌鸦传媒. Explore the transformative potential of generative AI, data platforms, and sustainability-driven tech. Find all previous Waves here. 聽Find all previous Waves here.