Experience building systems for daily operational use.

Before Leyline, co-founder and technical director Sam Kandi spent five years building systems inside a London healthcare communications business. His responsibility began with the operational problem and continued through delivery, adoption and maintenance.

The examples below come from one organisation and one main sector. In each case, responsibility continued after launch, through daily use, changing requirements and ongoing maintenance.

Selected work

Operational systems

Slow systems, repeated data entry and constant maintenance

Problem

Everyday work relied on separate files and systems that took too long to open and update. The same information had to be entered in several places.

People could not always be confident that they were looking at the latest data. Maintenance problems regularly interrupted the work.

Build

A single platform brought project management, sales pipeline, performance measures, tasks and timesheets together. Shared data replaced repeated entry across separate files.

The platform was designed and introduced with the people who used it every day.

Result

The platform became the division's central operational system and absorbed adjacent systems over five years. Maintenance requests fell from approximately 50 each month before implementation to three in total across the most recent 12-month period.

AI integration

AI was available, but it had not changed the work

Problem

People could use AI through a general chat interface, and it could help with isolated tasks. Even so, it felt separate from daily work and less useful than expected.

Build

The main constraint was access to context. The assistant could not use information held in the operational platform or email, so staff had to find and supply the relevant material for every task.

Controlled connections gave the assistant access to those systems. Reusable workflow instructions defined how it should complete recurring work.

Result

After the integration, the team used the same assistant to prepare for directors' meetings and update marketing activity. The improvement came from connecting AI to the organisation's information and workflows, rather than changing the underlying model.

For some recurring workflows, the change was stark: work that had required people to gather information and move it between systems was reduced to a single instruction to the assistant. Other workflows still required judgement or review, but far less preparation and system handling. This was the practical value of the integration: fewer steps between deciding that something needed to happen and getting it done.

Applied AI

A valuable recruitment channel created too much manual work

Problem: A new recruitment channel brought in more applications, but each CV still had to be reviewed and structured by hand. Processing the additional volume required too much HR time.
Build: An automated workflow used AI and document recognition to read incoming CVs and extract the information HR needed into a consistent structure.
Result: The workflow reduced the HR team's total CV-processing time by 50%, while the business continued using the higher-volume recruitment channel.