Kiandra Insights

Delivering AI projects with purpose: real lessons from the trenches

Aarti Nagpal - Software Development Team Lead
by
Aarti Nagpal
Software Development Team Lead
|
July 7, 2025
Aarti Nagpal
Software Development Team Lead
July 7, 2025
Illustration of a person holding a glowing tablet displaying "AI", surrounded by abstract data streams, charts and lines symbolising artificial intelligence and information flow.

AI isn’t some futuristic toy we’re tinkering with on the side. It’s already woven into the way we get work done at Kiandra. Whether it's helping sift through mountains of invoices or modernising stubborn legacy code, we’re using AI to tackle the headaches that come with real-world software delivery.  

We’ve experienced that AI delivery isn’t neat or predictable. It’s messy and iterative and full of trade-offs and lessons and failures. The best moments are watching someone use a system we built and say, “This just made my job easier.” To us, that’s success.

AI in the real world? Here's what that looks like

When you're leading a delivery team, complexity is part of the job description and AI helps us cut through it.  

We’ve used AI to make a tangible difference for clients, including the refactoring of older codebases in a fraction of the time it would’ve taken us the traditional way. We’ve trained models to spot performance anomalies in customer-facing apps, before they spiral into outages and lost user trust.

Tools like GitHub Copilot and Cursor have earned their place in our workflow. They won’t replace an expert developer, but they do help us work smarter with cleaner code, fewer bugs, and more time to think deeply about design, UX, and maintainability. At Kiandra, AI isn’t about replacing people, it’s about giving them back their time to focus on what really matters.

Why governance is the backbone of everything

Governance isn’t a checklist you tick off at the end. It’s the groundwork.

Before we even open a code editor, we’re asking these questions in our discovery workshops:  

  • Do we have the right data?
  • Can we use it?
  • Is it complete and fit for purpose?
  • Is it ethical and fair?
  • Can we maintain privacy and security?  

These aren’t just compliance questions, they’re essentially design questions and part of our architectural decisions. They shape the entire direction of a project. We lean on ethical frameworks to help us think beyond just building functional systems, to building trustworthy ones.

Measuring impact beyond the accuracy score

Accuracy is important, but alone and siloed, it’s shallow. What matters is how models behave under real pressure, in real environments.

Take one of our projects where we built a model to classify invoice attachments from emails. We weren’t just gunning for a flashy “high accuracy” stat. We needed to know what happened when it made a mistake. A false negative might let a risky invoice slip through. A false positive? It could delay legitimate payments and frustrate everyone involved.

That’s why we dug deep into metrics like:

  • Precision – Did the model flag the right risks?
  • Recall – Did it catch all the ones that mattered?
  • F1 Score – How well did it strike a balance?

These numbers gave us a grounded view of model performance and helped set realistic expectations with stakeholders.

If you'd like to learn about how AI can help solve real delivery challenges for your business, let's talk. We're always happy to show you what practical, secure, people-focused AI looks like in action.

Share article
LinkedIn.com

More insights

Aerial view of a commercial airplane taxiing on an airport runway, casting a sharp shadow. The aircraft is centred with visible engine nacelles and wing structure, surrounded by intersecting taxiway lines and concrete markings.

Common challenges in the travel industry and how AI can solve them 

Cassandra Wallace
7/7/2025

Travel businesses need systems that do the work with faster quotes, smarter pricing, and better traveller experiences. We’ve mapped the common travel challenges in the industry, and exactly how AI can solve them.

Read more
Aerial view of a cargo port at night showing a large container ship being loaded or unloaded by cranes. The scene is illuminated with industrial lighting and features overlay graphics of connected white dots and curved lines, representing a digital or AI-powered logistics network.

Rethinking logistics with AI: real challenges with real solutions

Cassandra Wallace
2/7/2025

In transport and logistics, most operators don’t need to be sold on the benefits of AI, they just need a clear path to making it work. This blog post outlines the most common challenges we see across logistics businesses, and how AI, when implemented correctly, can solve them and improve performance.

Read more
An illustrated person holding a tablet, looking up at a large, abstract AI brain filled with interconnected circuits and nodes, symbolising the integration of technology and human thinking.

AI and the human factor: what I’ve learned leading a team through change

Aarti Nagpal
26/6/2025

As a Software Delivery Team Lead, I’ve experienced firsthand how our strength lies in pairing deep technical expertise with a culture that genuinely values people. We’re not here to chase hype. We’re here to build things that matter, with teams that are empowered, curious, and supported.

Read more

Let’s discuss your next project

Whether you’re curious about custom software or have a specific problem to solve – we’re here to answer your questions. Fill in the following form, and we’ll be in touch soon.

Email

Would you like to receive an occasional email showcasing the latest insights, articles and news from our team of software experts?

Thanks for reaching out! One of our software experts will be in
touch soon to help you with your enquiry
Oops! Something went wrong while submitting the form.

This website uses cookies to improve your experience. By browsing our website you consent to the use of cookies as detailed in our Privacy Policy