AI in Education, Cloud Operations and Construction: What Our Case Studies Show

In short: We have published case studies from an education consultancy, a cloud services provider and a construction company in Kathmandu. They look unrelated, but the pattern is shared: each had skilled people spending their time finding, collecting or repeating information, and each project removed that step rather than adding more headcount. This article summarises what each case study states, with the figures presented as reported in our case studies and not independently audited, and then pulls out what a business in any industry can take from them.

Key Takeaways

• Education (Admizz Education): an AI-native inquiry system built on Zunkiree Search. Our case study reports 45% faster response times, a 3x improvement in student satisfaction and a 60% reduction in manual inquiry handling.
• Cloud operations (Corecloud365): an AI-powered operations assistant for finding runbooks and past incidents. Our case study reports a 60% reduction in operational costs, 80% faster issue resolution and 99.9% system uptime achieved.
• Construction (Khushbu Nirman Sewa): mobile data collection with AI-powered resource optimisation and delay prediction. Our case study reports 40% fewer project delays, NPR 2.5 crore saved in the first year and an 85% reduction in manual reporting time.
• All figures are as stated in our published case studies. Individual outcomes vary and are not guaranteed, so use them as a guide to where the value was found, not as a forecast for your business.
• The shared lesson: start by getting the information into one place and findable, then add prediction or automation on top.

Education: Answering the Same Questions Faster

The situation. Admizz Education is an education consultancy helping students navigate university admissions. According to our case study, it handles thousands of student inquiries a year about programs, eligibility, application processes and deadlines. The key pain points listed were an average first response time of more than 24 hours, 70% of questions being repetitive, counsellors burned out from repetitive work and students frustrated by slow replies during application season.

What was built. Admizz implemented Zunkiree Search to power an AI-native inquiry system. Students ask questions in natural language and get instant answers that take context into account: the IELTS requirement for UK universities is different from the requirement for Australian ones. Complex queries are routed to human counsellors with the context preserved.

What was reported. 45% faster response times, a 3x improvement in student satisfaction and a 60% reduction in manual inquiry handling.

What carries over. If a large share of your inbound questions repeat, the saving is not only speed. It is freeing skilled people for the conversations that need judgement. The design choice that matters is the handoff: the AI answers what it can and passes everything else on with context. For a related guide, see our AI for education consultancy whitepaper.

Read the full Admizz Education case study.

Cloud Operations: Making Tribal Knowledge Searchable

The situation. Corecloud365 is a cloud services provider managing infrastructure for mid-market enterprises. Its case study describes troubleshooting that meant searching logs across dozens of systems, documentation spread across five systems, tribal knowledge held in senior engineers' heads and a mean time to resolution that averaged four hours. Resolution time depended on who was on call.

What was built. Corecloud365 deployed Zunkiree Search as an AI-powered operations assistant. Engineers describe a problem in natural language and receive relevant runbooks, past incident reports and troubleshooting steps. Resolved incidents are indexed automatically, so what one engineer learns becomes searchable for the next.

What was reported. A 60% reduction in operational costs, 80% faster issue resolution and 99.9% system uptime achieved.

What carries over. The value came from the knowledge that already existed. The project did not replace engineers. It made documentation and past incidents findable, so junior staff could get to answers without waiting for a senior colleague. If your expertise lives in a few people's heads, capturing resolved problems automatically is a durable gain.

Read the full Corecloud365 case study.

Construction: From Paper Forms to Predicted Delays

The situation. Khushbu Nirman Sewa is a construction company based in Kathmandu, with more than 15 active sites across Kathmandu Valley. Our case study lists the problems: site supervisors reporting on paper forms with two to three day delays, no visibility into resource use across sites, material wastage averaging 15% and projects averaging 20% over schedule.

What was built. Supervisors log daily progress, issues and resource use on a mobile app. Machine learning suggests how to allocate resources across sites, and the system analyses patterns to predict potential delays two to three weeks in advance. Management sees all sites on a single dashboard.

What was reported. 40% fewer project delays, NPR 2.5 crore saved in the first year, real-time visibility across 15+ active sites and an 85% reduction in manual reporting time.

What carries over. The case study's own takeaways are worth repeating. Start with data collection, because AI depends on good data and mobile capture by field workers was the critical first step. Prediction is more valuable than reporting, because the gain comes from acting before a problem occurs. And a measurable return makes the case for expansion.

Read the full Khushbu Nirman Sewa case study.

What the Three Have in Common

1. The bottleneck was information, not effort. In each case skilled people spent their time finding, collecting or re-answering. Removing that step changed the economics.
2. Humans stayed in the loop. Counsellors, engineers and site managers still made the decisions. The AI handled retrieval, capture and prediction, and escalated what it could not handle.
3. Foundations came first. Searchable knowledge, structured field data and clear routing came before any advanced feature.
4. Each result is specific to its context. The figures depend on that business's volumes, content and systems. Treat them as a guide, not a promise.

Questions to Ask Before Your Own Project

• Where do skilled people in your business spend time finding or re-entering information?
• Is the knowledge written down anywhere, or does it live in people's heads?
• What would "better" mean, and do you have a baseline measurement today?
• When the AI cannot help, who takes over and what do they receive?

If you want to explore this for your own business, our AI development and custom software teams can scope it with you, and our Zunkiree Search product covers the search-led cases.

The Short Version

Three different industries, one pattern: find the step where skilled people spend time on information rather than judgement, remove it, keep a clear route to a person and measure against your own baseline. The reported figures show where value was found in each case, not what any other business should expect.

To discuss a similar project, start with our AI development services or browse the customer stories.

Related Insights

• Why AI Matters for Nepal: What the Evidence Shows
• AI Orchestration vs Automation vs Agents: The Difference
• AI Chatbot vs AI Agent: Customer Support Automation for Your Business

Sources

• Zunkiree Labs, Admizz Education case study
• Zunkiree Labs, Corecloud365 case study
• Zunkiree Labs, Khushbu Nirman Sewa case study