AI should help people find their evidence, not invent their story
Why My Next Step puts a person’s actual experience before a polished application.
Getting startedAbout 3 min read
A polished answer can leave someone behind
An AI assistant can write a persuasive paragraph from very little information. In employability work, that fluency can be a problem as well as a benefit. The paragraph might sound confident while containing claims the person cannot explain.
My Next Step begins somewhere more grounded: what has the person actually done? An ordinary experience may contain evidence of communication, reliability or organising. The job is to help someone examine the experience, not to give them a better fictional biography.
Keep the example connected to the claim
The application keeps stories as saved material and connects skills to that evidence. Its current documentation treats skill progress as a projection from saved stories, rather than an independently mutable score. That makes the relationship between an experience and a claimed skill more visible.
The distinction matters when someone uses the material towards an opportunity. They should be able to return to the example, recognise it and decide what they want to say. A sentence that sounds impressive but has lost that connection is a weaker foundation for confidence.
Words someone can recognise
As an illustration, “I helped a new colleague close the till” describes an action someone could explain. Turning it into “I led operational transformation” adds a different claim. More impressive language is not necessarily a clearer account.
There is room to improve grammar and organise the example. I want the person to recognise the result, correct it and choose whether to use it. The final wording should still lead back to something they actually did.
Give AI a supporting role
AI can help ask a question, identify something worth examining or make a person’s own wording clearer. Those are useful forms of assistance. They do not require handing over the decision about what is true.
The person needs time to confirm what happened and what they want to use. If a suggestion adds a responsibility, result or feeling they never supplied, it needs checking before it enters the record. However plausible the sentence sounds, it is still their experience to explain.
A demonstration is not an outcome
The interview-practice screen on this site uses fictional development examples. It shows a choice I wanted to make visible: which saved experience would you like to practise with? It does not show a real participant’s progress or establish that live conversation is ready for everyday use.
I need to be as careful with claims about the tool as with the stories inside it. A screenshot can show the interface; source code can show an implemented behaviour. Whether someone grew in confidence or found work would need a different kind of evidence.
Those outcomes would need their own careful evaluation. Until then, I can explain the purpose, show the design and name the current development boundary without replacing uncertainty with a success story.
What stays when the tool is put away?
I keep wondering what someone will take away after using the tool. Will they recognise something in their own experience? When the screen is closed, could they explain it in their own words?
Help with language can matter, especially when someone knows what happened but struggles to put it into words. I want that help to preserve their meaning and leave them able to decide whether the sentence feels true.
That is what I hope to make room for: a person recognising something real in their own experience, then deciding what to do with it. Whether the tool helps that happen is something I still need to learn from the people using it.