Five starting points for software that uses AI

Five starting points for software that uses AI

This address originally promised nine ideas. The published piece covered five, and that is what this page keeps. Treat them as custom software ideas some organisations explore, not as products CoSource sells off-the-shelf.

CoSource builds around a real process you already run. We do not start from a trend list.

Book a free 30-minute discovery call

1. Personal stylist

An app that learns a user's taste, fit and the clothes they already own, then suggests what to wear. Machine learning is the usual way to get those suggestions closer over time.

Some fashion and ecommerce teams look at this when they already have product data and want a more personal recommendation path. It is only worth building if that data and that process exist.

2. Language tutor

An app that builds lessons around what the learner already knows, the pace they can keep, and the situations they need the language for. Natural language processing is the usual way to check answers and adjust the next lesson.

Some education teams look at this when a fixed curriculum is a poor fit. The product is the learning path, not a demo of the model.

3. Healthcare assistant

An app that records vitals, meals and medication, and reminds someone when a dose is due. It can also surface when it is time to speak to a clinician.

This is an idea some health organisations explore. It is not a CoSource product, and we do not claim clinical cover at all hours. Anything in this area has to start from the real care process, consent, and the rules that apply to health data.

4. Job finder

An app that reads advertised roles and a candidate's history, then shortlists matches. Some recruitment teams look at this when people already spend hours scanning boards by hand.

It is only useful if the sources are ones you are allowed to use, and if someone can still judge a fit the model will miss.

5. Eco shopping assistant

An app that compares products on how they are made, their carbon cost, and whether they can be recycled, then points to a better option. Some retailers look at this when they already hold supplier data and have customers who ask for it.

Without that data, the app is a guess. The work starts with the catalogue you actually have.

Build around the process

These five are starting points, not a catalogue. If a process is still in a spreadsheet, in email, or in two systems that do not talk, that is the brief. The model, if you need one, comes after.

Related: Custom software

Talk through the process, not the trend

Thirty minutes. Bring the job the software would do, and the data you already hold. We will tell you whether a custom build is the work.

Book a free 30-minute discovery call