What clients say about our Artificial Intelligence work

We asked permission to share these quotes and numbers. Every figure below comes from a real project with a named company, and we are happy to arrange a reference call if you want to hear it first-hand.

Client testimonials

Unedited feedback from project sponsors and technical leads.

★★★★★
We had tried building a churn model internally twice and abandoned it both times. Pro AI Perspectives finished a working version in seven weeks. The model now flags at-risk accounts 30 days before renewal, and our retention team acts on those alerts daily. We kept 14 % more accounts in the first six months.
GT avatar
Grace ThorntonHead of customer success, Bridgeline Insurance
★★★★★
The document intelligence pipeline processes roughly 400 invoices a day for us. Before that, two full-time staff spent their mornings keying in supplier details. Those people now handle exception cases and vendor negotiations instead. The accuracy sits at 96 % on clean scans and about 89 % on poor-quality photos, which is better than we expected.
RH avatar
Ravi HussainOperations manager, Northway Facilities
★★★★☆
I appreciated the honesty during the discovery call. Daniel told us one of our three proposed use cases probably would not work because we did not have enough labelled data, and he was right. The two models we did build are in production and running well. I took one star off only because the final documentation arrived a week late.
LM avatar
Laura McBrideCTO, Pennant Retail Group
★★★★★
Our fleet of 120 vans used to run about 22 % empty miles. After Pro AI Perspectives deployed the demand-forecasting model, that figure dropped to 14 % within the first quarter. The fuel savings alone paid for the project in five months. Helen and the team also trained our dispatchers to read the model output, which made adoption much smoother.
JW avatar
James WhitfieldDirector, Ashford Couriers
★★★★★
We engaged Pro AI Perspectives for a data audit before committing to any model work. The report identified three critical gaps in our CRM data and recommended fixes that took our team about six weeks to implement. When we came back for the predictive model, the training data was clean and the project finished ahead of schedule.
NA avatar
Nadia AkhtarData lead, Clearview Professional Services

Case studies

Deeper looks at two projects, with permission from the clients involved.

Logistics

Demand forecasting for Ashford Couriers

Ashford Couriers operates 120 vans across the Midlands and South East. Their dispatchers planned routes manually each morning using gut feel and last week's volumes. Empty-van miles sat at 22 %, costing the company roughly £18,000 per month in wasted fuel and driver time.

We ingested 14 months of delivery data, weather records and local event calendars. A gradient-boosted regression model now predicts next-day parcel volumes per depot with 91 % accuracy, measured as mean absolute percentage error below 9 %. Dispatchers receive a forecast dashboard at 6 a.m. and adjust routes accordingly.

22 % → 14 %Empty-van miles
5 monthsPayback period
Delivery van on an English country road
Insurance

Customer churn prediction for Bridgeline Insurance

Bridgeline Insurance had a 28 % annual churn rate on commercial policies. Their retention team contacted every account nearing renewal, but the calls were untargeted and the team could only reach about 40 % of the book each quarter.

We built a classification model using policy history, claims data and interaction logs. The model scores every account weekly and flags those with a churn probability above 65 %. The retention team now focuses on the highest-risk accounts, reaching them 30 days before renewal instead of the previous 10-day window.

In the first six months after deployment, Bridgeline retained 14 % more accounts compared to the same period the previous year. The model runs on their existing Azure infrastructure, and we retrain it monthly with fresh claims and interaction data.

+14 %Accounts retained
30 daysEarly warning window
91 %Model precision
Team reviewing a churn prediction dashboard in a meeting room

By the numbers

Aggregate figures across all client engagements since 2019.

38Models in production
97 %On-time delivery rate
12Industries served
4.8 / 5Average client rating