We Asked ChatGPT and Recroot to Review the Same Conversation. Here's What Happened.

Recently, we ran a simple experiment.

We took the exact same workplace conversation and asked both ChatGPT and Recroot to review it.

The conversation wasn't an interview.

It was a manager asking an employee why a project was delayed.

The employee explained the challenges, competing priorities, and the work completed. The response was professional and well-structured.

At first glance, both systems seemed helpful.

ChatGPT's feedback focused on the response itself:

• clearer structure

• shorter sentences

• better flow

• more concise wording

Useful feedback.

But we have designed Recroot to approach the conversation differently.

Instead of asking:

"Was this response written well?"

It asked:

• Did the employee take ownership?

• Did they answer the manager's real concern?

• Did they demonstrate accountability?

• Did they provide confidence that the issue is under control?

• What impression would this response create?

• What action is the listener likely to take after hearing this?

The reason is simple.

Recroot isn't designed as a general-purpose feedback tool.

It uses specialised AI evaluators, communication frameworks, and role-specific conversation models designed around real workplace outcomes.

For example:

• A Manager Conversation Evaluator looks for ownership, accountability, stakeholder management, leadership behaviours, and trust.

• An Interview Evaluator looks for evidence, competency demonstration, role fit, hiring signals, and authenticity.

• A Salary Negotiation Evaluator looks for confidence, value articulation, persuasion, and negotiation effectiveness.

• A Client Conversation Evaluator looks for credibility, relationship management, commercial awareness, and problem-solving.

Each conversation is assessed against what the other person is typically expecting to hear in that specific situation.

That's where the results became interesting.

The employee's response was clear and professional.

But Recroot identified that most of the answer focused on explaining why the project was delayed and very little on how the issue would be resolved.

From a communication perspective, the answer was good.

From a manager's perspective, it could still create concern.

Because managers often care less about the reason for the delay and more about:

• What are you doing about it?

• What is the impact?

• What is the recovery plan?

• How will this be prevented in the future?

That's not something generic communication feedback usually identifies.

But we discovered another difference.

Most AI tools review conversations after they happen.

Recroot lets you practise before they happen.

Instead of typing responses into a chat window, users can step into realistic workplace scenarios that simulate actual conversations.

• Video-based interactions

• Voice conversations

• Real-time questioning

• Multi-part discussions

• Meeting-style environments

• Role-specific scenarios

• Follow-up questions based on your responses

The goal isn't to create better answers.

The goal is to prepare people for the pressure of real conversations.

Because most people don't struggle when they're thinking about a response.

They struggle when they're sitting across from a manager, interviewer, client, or executive team and need to respond in real time.

And that's where communication skills are actually tested.

The more we explored this problem, the more we realised something:

People rarely fail because they don't know what to say.

They fail because pressure changes how they say it.

That's the gap we're trying to solve with Recroot.

Not just helping people communicate better.

Helping them understand how they're perceived, while giving them a safe place to practise the conversations that can shape careers.