Sigli

CASE STUDY – 1

Scaling AI-Driven Innovation for One of the UK’s Leading Property Platforms

Overview

This case study presents how an AI and data engineering team helped one of the UK’s leading property technology platforms modernize and scale its machine learning infrastructure, enabling faster feature development, improved analytics, and stronger competitive positioning in the estate agent SaaS market.

The Challenge

The client operates one of the UK’s most advanced data platforms for estate agents, serving major real estate brands such as Knight Frank and Carter Jonas.

Despite their strong market position, the company faced increasing pressure to innovate rapidly in an industry where AI-driven capabilities were becoming a key competitive advantage.

The existing machine learning infrastructure created several operational bottlenecks:

• slow and inefficient ML models
• limited scalability of data processing workflows
• difficulty introducing new AI-powered features quickly
• complex dependencies between multiple data sources and models

The company needed to modernize its machine learning ecosystem while maintaining high standards of data quality, platform reliability, and security.

An additional challenge came from strict confidentiality requirements around proprietary datasets, which prevented the use of standard cloud-based ML infrastructure.

The Solution

To address these challenges, a dedicated AI and data engineering team was embedded directly into the client’s workflow.

The implementation followed a structured three-phase approach:

• research and analysis of existing ML architecture and pipelines
• development and integration of new data processing workflows
• validation, optimization, and long-term operational support

Several dozen new data pipelines were developed using technologies including SQL, DBT, Apache Airflow, and Python-based ML frameworks.

The solution focused on:

• optimizing existing ML models
• building scalable data infrastructure
• improving data processing efficiency
• enabling rapid deployment of new AI-powered product features

Due to confidentiality requirements, all systems were deployed on the client’s on-premise infrastructure rather than in the cloud.

The project was designed not only to solve immediate performance issues but also to establish sustainable workflows and best practices enabling future innovation.

Barriers & Challenges

One of the key challenges was the lack of comprehensive documentation for the existing codebase and ML architecture.

This required significant reverse engineering before optimization work could begin safely.

Additional challenges included:

• processing extremely large and complex datasets from multiple proprietary sources
• managing inconsistent data quality and schema variations
• operating entirely on on-premise infrastructure with limited computational flexibility
• supporting migration from legacy systems to modern tooling
• maintaining stability across highly interconnected ML models and pipelines

Careful sequencing, extensive testing, and close collaboration with the internal team were essential to ensuring a smooth transition.

Results

The project delivered substantial improvements across infrastructure performance, scalability, and product innovation.

Key outcomes included:

• development of several dozen optimized data pipelines
• significant acceleration of ML processing workflows
• implementation of advanced AI-driven analytics and prediction models
• successful launch of new product features, including:
  • real-time property tracking
  • market trend analysis
  • AI-powered prospect identification

The engagement also established scalable workflows that enabled the client’s internal teams to continue developing and deploying ML-powered features more efficiently in the future.

As a result, the company strengthened its ability to compete in a rapidly evolving AI-driven real estate technology market.

CASE STUDY – 2

How Allkind Group De-Risked AI Integration Across Three EdTech Brands

Overview

This case study presents how an education technology group serving over one million users successfully implemented AI and data infrastructure across three independent brands, improving engagement, scalability, and operational efficiency while minimizing the risks associated with large-scale AI adoption.

The Challenge

Allkind Group operates three EdTech brands focused on educational support and accessibility solutions:

• Lexima
• Sensotec
• Schoolsupport

The organization faced growing pressure to integrate AI capabilities in order to remain competitive in the rapidly evolving education technology market.

However, implementing AI across multiple independent brands introduced significant risks, including:

• investing in AI where simpler automation would be more effective
• fragmented systems and disconnected data
• poor adoption among teachers and educators
• duplicated operational processes
• high implementation costs with uncertain business value

The challenge was not only technological but also strategic — determining where AI could create genuine value and where traditional automation or improved infrastructure would be more appropriate.

The Solution

To reduce implementation risk, the project began with a comprehensive discovery and analysis phase focused on validating AI suitability before development started.

Rather than applying AI indiscriminately, the solution prioritized practical use cases with measurable business value.

The implementation focused on three strategic pillars:

• integration of systems and data across all three brands
• selective deployment of AI-powered functionality
• development aligned with accessibility and educational requirements

A multidisciplinary team of AI engineers, software developers, QA specialists, and solution architects designed and implemented:

• integrated analytics infrastructure
• personalized learning recommendations
• predictive student progress analysis
• sentiment analysis tools
• workflow automation for teachers and educators

The system was built using Azure cloud infrastructure, TypeScript, Node.js, Python, PyTorch, and React-based interfaces optimized for accessibility.

The architecture enabled all three brands to maintain operational independence while benefiting from shared data insights and analytics.

Barriers & Challenges

One of the main challenges was balancing business expectations with realistic AI implementation strategies.

Initial stakeholder expectations involved introducing AI across nearly all processes, but early analysis revealed that many operational needs were better solved through infrastructure improvements and workflow automation.

Additional challenges included:

• integrating three separate brands with different systems and operational models
• ensuring accessibility standards for users with disabilities
• preventing algorithmic bias in educational AI systems
• managing organizational adoption across more than 100 employees
• meeting GDPR and international educational compliance requirements

The project also required careful change management to ensure that educators and staff actively adopted the new tools and workflows.

Results

The implementation delivered measurable improvements across user engagement, operational performance, and platform scalability.

Key outcomes included:

• 15% increase in user engagement
• 20% growth in active users
• 40% increase in positive testimonials
• 5x improvement in core infrastructure performance

The organization successfully integrated AI-driven personalization and analytics while avoiding unnecessary complexity and costly architectural restructuring.

The project also enabled the company to build scalable foundations for future AI initiatives while maintaining strong accessibility and educational standards.

As a result, Allkind Group strengthened its market position and improved the experience of both educators and students across its platforms.