Proof of Progress: Real stories of real people putting AI skills to work

A marketing leader used AI alongside diagramming and governance tools to speed up writing policies and procedures - work that had been held back by limited time and specialist skill.

“Process development work, which in my Head of Marketing role could take weeks … has transformed into tasks that can be completed in single sessions of focus - giving incredible productivity gains.”

Impact: Already happening. Process work that once took weeks now fits into a single focused session, with better onboarding of the team onto updated processes.

In a regulatory affairs team, manual data entry and reconciliation were slowing daily work and reporting. After workshops to address colleagues’ concerns, our learner designed an AI chatbot to handle routine queries, with clear limits on what it can do.

“The tool automation is still in progress and set to automate 30% of our manual work, saving each team member about 20 hours per month and significantly improving the accuracy and timeliness of our reports.”

Impact: In progress. AI-powered tool, built with the team’s concerns designed in from the start, on track to hand back around 20 hours a month per person.

Working with the safeguarding lead, one person on a Corndel programme used AI to turn meeting notes into a clear map of the team’s current process, pinpointing where automation could help.

“The expected outcome is that administration time will be reduced and that our team will have more time to focus on more quality-led outcomes.”

Impact: Already happening. A safeguarding process mapped end to end, with automation pinpointed so the team can spend less time on admin and more on people.

After building and prioritising a backlog of AI ideas with the team, a brokerage lead presented 20 use cases to technology partners and senior leaders, and is now taking the top three forward.

“One of those is on the subject of Market Data Cost Saving and Optimisation, which could save Brokerage millions and the broader bank tens of millions per year.”

Impact: Projected. Twenty AI use cases identified and prioritised with partners; the lead case targets roughly a 20% cut in annual market-data cost.

This person used Copilot to work through several large health-and-safety datasets - hazard observations, near misses, incidents, audits - to surface recurring themes.

"The challenge was that while all the information was there, it was difficult and time consuming to spot reoccurring themes or opportunities for improvement."

Impact: Already happening. What previously took days of manual review became a short, evidence-based process that led directly to agreed actions and clearer operational priorities.

Hundreds of hours of live radio go largely unused because journalists don’t have time to mine it for stories. They designed an AI workflow to transcribe and surface story leads at scale, then tested it live under real deadline pressure.

“I set out to identify where AI could deliver practical value in a live newsroom environment, rather than theoretical use cases.”

Impact: Already happening. In a single pilot, the workflow processed around 20 hours of live phone-in content and generated 19 AI-assisted story outputs, of which 6 were published on the live online page on the same day - every one verified by a journalist first.

A compliance specialist redesigned a regulated breach-management process that had relied on email and manual steps. AI now runs advisory checks before a person reviews - it can flag gaps, but can’t approve, reject or close anything.

“I applied my learning on responsible AI adoption to introduce AI in a way that supports, rather than replaces, human decision-making.”

Impact: Projected. Once live, the redesigned process is anticipated to cut review and rework effort by 30–40% - an estimated 15–25 hours a month per reviewer at peak.

Drafting regulatory documents in oncology was slow and inconsistent, with versions emailed back and forth. Our learner designed a workflow where AI produces first drafts only, with people reviewing and approving every step.

“The Corndel AI programme gave me the confidence and knowledge to start by asking the right questions before proposing AI.”

Impact: Already happening. A human-in-the-loop drafting workflow, tested on 2–3 real projects, with a target of 30–40% faster approvals within six months - and clinical decisions kept firmly out of scope.

Finding a relevant past decision used to mean hours of manual searching. Our learner piloted a searchable database of previous cases, plus AI support for drafting public summaries - all checked against editorial standards.

“Initial use has shown it can significantly reduce the time taken to locate comparable cases - from potentially hours of manual searching to a matter of minutes.”

Impact: Already happening. A searchable database of past decisions that surfaces precedent in minutes, supporting more consistent outcomes, with every summary still reviewed against editorial standards.

A Risk Manager used Copilot's research agent to check documentation and surface privacy and compliance risks earlier.

"Switching to this AI workflow improved my risk reports, helped me find privacy issues sooner, and made audits easier. It now saves me over four hours a week."

Impact: Already happening. Cut routine compliance checks from around two hours to 30 minutes, with earlier issue-spotting and stronger reporting.

Seeing colleagues expected to ‘just use’ AI, our learner ran hands-on Copilot sessions for an estates and compliance team, with practical examples and guidance on using the outputs responsibly.

“This led to increased adoption of AI tools, with positive feedback showing improved confidence and practical use in daily tasks.”

Impact: Already happening. After the sessions, 93% of attendees found it very useful, 57% now use Copilot regularly (a further 29% occasionally), and 71% asked for more training.

Noticing colleagues were expected to use AI without really understanding it, our learner built a workshop on the basics of prompting and bias.

“I created a workshop for colleagues explaining the fundamentals of prompting and bias in AI. This was so successful that other departments are now also getting me to do these sessions with their team.”

Impact: Already happening. Colleagues report starting to work with the organisation’s own AI tools and feeling more confident doing it, and demand has spread department to department.

An administration manager in a professional services firm noticed their colleagues were spending too much time on repetitive emails - so set up reusable email templates and, crucially, taught the rest of the team to use them.

"My team was spending a lot of time sending repetitive emails, which was inefficient, especially with 20 people involved. I set up 'Quick Parts' in Outlook to automate common email responses … and passed the knowledge to our team leaders, who then shared it with the rest of the team."

Impact: Already happening. Adopted across a 20-person team, saving up to an hour a day and improving consistency.

One team applied AI to process analysis and integration design, then turned it into a way of working others wanted to copy.

"The success of this initiative [using Copilot] positioned our team as a benchmark for innovation, leading to requests for similar approaches in other departments."

Impact: Already happening. The team cut manual effort by roughly 35–40%, saving an estimated 15–20 hours per sprint and signing off designs around 25% faster - and now reuses the approach across other projects.