Is it brilliant or is it bunkum? How to build agile POCs with FME

New cloud platforms, AI services, APIs, and data capabilities crop up constantly. Each one arrives with big spurious claims about what it could do for your organisation.

So, you start a proof of concept (PoC) to find out whether it does what it says it does, or whether it’s complete bunkum.

The Three Things a Good POC Needs

An agile POC needs three main things:

• Speed. You need an answer quickly, before something else takes priority.
• Adaptability. Early assumptions are often wrong, so the prototype has to be easy to change as you learn.
• Connectivity. It has to connect to the data, systems, and services involved. Otherwise you’re demonstrating an idea, not proving it’ll work.

The Coding Bottleneck

A lot of organisations lean on code for their POCs. The trouble is, asking a developer to build custom connections and write code before you can even test the basic idea is slow going.

Every change means more development, which makes experimentation even slower and more expensive. Before you know it, the POC has snowballed into a substantial project before it’s proved any value at all.

That makes organisations reluctant to test ideas unless they’re already fairly confident they’ll work. This completely defeats the purpose of a POC.

Prototyping Without the Code

So how do you remove the coding bottleneck and speed up the learning? Well, you use a platform like FME.

FME lets you connect data, systems, cloud platforms, and APIs without building every integration from scratch.

Its visual workflows are quicker to assemble, inspect, and understand when compared to digging through custom code. This visual style also means you can easily add, remove, or replace individual components as your understanding develops.

If the POC works, you can productionise it in FME or use what you’ve learnt to build a coded solution. If it doesn’t work, you’ve found out before investing in a major development project.

An Agile POC with FME

An agile POC is about testing assumptions, learning from real results, and changing direction without wasting significant effort.

FME lowers the cost of experimentation, so you can explore new capabilities without making an immediate long-term commitment. When the next cloud platform, AI service, or data capability appears, you can find out whether it’s actually useful or absolute bunkum.