GUIDED CONVERSATIONS

ABOUT

This AI-powered experience transforms how organisations listen to their people at scale. Learning professionals define a conversational questionnaire that Maestro follows — probing deeper, asking for clarification, and adapting naturally — while structured insight categories automatically surface themes, sentiment, and priorities from every response.

The result: the kind of nuanced understanding you'd normally only get from one-on-one conversations, captured across thousands of employees at once.

CHALLENGE

Traditional surveys fall short where it matters most. Structured questions miss sentiment. Rating scales miss nuance. Employees answer quickly and move on, and participation drops with every cycle. Meanwhile, learning leaders are expected to make high-stakes programme decisions with low-quality signal. They need to understand not just what employees think, but why, and existing tools simply weren't built for that.

SOLUTION

We reimagined organisational listening as a conversation, not a form. Powered by generative AI, Guided Conversations lets admins define questions and map them to structured insight categories: frequency, themes, sentiment, custom outputs. Maestro handles the rest: following the questionnaire, probing where answers are shallow, and aggregating responses into actionable, privacy-respecting insights. For learners, it feels like a natural check-in. For administrators, it produces the kind of intelligence that moves their L&D strategy forward.

PROJECT DETAILS

YEAR: 2026

INDUSTRY: Edtech

MY ROLE: Discovery, Product Strategy, AI Prototyping, UX/UI Design

TEAM: Lead Designer (me) | PM | Engineering Manager | 6 engineers (4 FE, 2 BE)

PLATFORM: Web (Responsive)

Core Experience Principles

  1. Conversational, Not Transactional Every interaction should feel like a genuine exchange, not a form disguised as a chat.

  2. Structured at the Back, Natural at the Front Admins define what they need to know. Learners just talk. The structure happens invisibly.

  3. Insights Without Exposure Individual responses stay private. Value is surfaced at the aggregate level, protecting employees while informing decisions.

  4. Authored with Intention Scenario creation is deliberate as admins define not just questions, but what Maestro should listen for and why.

  5. Reliable Enough to Act On Insights need to be reproducible and trustworthy and every output is designed to support a real decision.

  6. Adaptive by Default Maestro doesn't just follow a script but responds to what it hears, making each conversation feel specific to the person in it.

The configuration interface gives learning professionals a clear, structured flow for building a guided conversation from scratch. Admins define the conversation title and description and build out their question list. Tabbed navigation moves cleanly between Configure, Settings, Insight Categories, and Results, keeping authoring focused and the mental model simple. The outcome is a powerful authoring system that does not require technical expertise to produce a sophisticated, AI-driven listening experience.

Authoring the Experience

Each question is authored with intent. Alongside the question itself, admins write context that tells Maestro how to behave if an answer needs probing, ensuring the conversation stays focused on the right signal.

Questions are then assigned to insight categories directly in the drawer, so the connection between what is asked and what is extracted is explicit from the start. This keeps the authoring process deliberate rather than assumptive, and gives Maestro the guidance it needs to produce reliable, structured insights.

Authoring the Experience

Defining What Maestro Listens For

The Insight Categories tab is where authoring becomes strategy. Admins map questions to structured output types including Frequency, Themes, Sentiment, and Custom, telling Maestro not just what to ask, but what to extract. Each category is named, described, and linked to specific questions, giving the system the context it needs to surface signal rather than noise.

For theme-based categories, admins can seed starting themes to nudge Maestro toward expected patterns, and toggle on Allow Emergent Themes to let it surface what they did not anticipate. This balance of structure and openness is what makes the insights genuinely useful.

Structure and Openness in Balance

From Conversations to Decisions

The Results dashboard turns thousands of individual responses into a clear organisational picture. A summary leads with the headline findings, supported by theme frequency charts, per-question sentiment breakdowns, and a ranked list of priorities surfaced directly from what employees said.

Every result is aggregated and anonymised, so L&D leaders can share findings confidently without exposing individual responses. The output is designed to be presentation-ready, giving leaders a clear story, a prioritised recommendation, and data they can stand behind in a CHRO conversation.

Project Outcomes

With Guided Conversations, we expect to see:

  • 3× higher participation rates compared to traditional pulse surveys, driven by the conversational format reducing friction and fatigue.

  • 60% of insights surfaced containing emergent themes not anticipated by admins, demonstrating the value of adaptive listening over fixed questionnaires.

  • L&D teams reporting a meaningful reduction in time-to-insight, moving from lenghty survey analysis to real-time aggregated results.

  • Actionable organisation changes initiated within 2 weeks of a Guided Conversation cycle completing, compared to quarters-long survey-to-action gaps.

  • Employees rating the conversation experience as more engaging than traditional surveys.