Project overview | 3 min read
Sage: Using AI to bridge communication gaps between neurodivergent and neurotypical folks
A conversational AI tool designed to help neurodivergent users navigate tone, intent, and ambiguity in text-based chats, without speaking for them.
Date
March 2025
Project type
Personal project, AI exploration
Role
Product designer
Team
Me, advised by Elizabeth Lin
Problem
Tone and intent are often lost in digital communication for neurodivergent folks
Many neurodivergent individuals process language and social cues differently. Something as small as a “k” can feel dismissive, and without visual or vocal context, misinterpretations are easy and common. This can lead to anxiety, social burnout, and communication breakdowns.
Solution
I prototyped an AI assistant to guide them in decoding and responding
Sage is a AI-powered chat assistant concept that helps users decode tone and intent, and craft more effective replies—without taking control of the conversation. Rather than auto-generating responses, Sage acts as a guide, offering insights into how a message may be interpreted and suggesting ways to respond thoughtfully. It empowers users to navigate ambiguity with confidence, while retaining full agency over their communication.
Research & inspiration
An opportunity to bring my passion project to life
This project began with a simple question: Why does digital communication feel so exhausting for some, but not others?
As someone with ADHD—and friends on various points of the neurodivergent spectrum—I kept hearing the same thing: “It feels like I’m speaking a different language.”
Posts like these, paired with personal conversations, highlighted a shared experience: neurodivergent users often struggle to interpret tone or be understood over text. This gap in digital communication sparked the idea for Sage.
Initial concepts
Prototyping lightweight interactions to test the product concept
To test the concept, I started by prototyping lightweight interactions around three key features:
Tone Indicators: Surface how a message might be interpreted (e.g., “gen” for genuine, “sar” for sarcastic) so users can adjust before hitting send.
Response Adjustments: Context-aware suggestions that help clarify ambiguity, not just autocorrect grammar.
Reply Reminders: Gentle nudges when a user forgets to respond to a specific part of a multi-threaded message—because we’ve all missed the “btw can I bring my dog?” text.
These concepts focus on augmenting the chat experience without automating or overriding it. Sage helps you re-read your own words through someone else’s eyes before they do.
Design approach
Evoking a sense of psychological safety through carefully crafted visual design
The interface was intentionally designed to evoke calmness and psychological safety—think therapist’s office meets group chat. Playful but grounded. Thoughtful without being clinical. It needed to feel like a safe space, especially for users navigating emotionally taxing moments.
Prototyping for real moments
[Micro]-interaction experiments with Sage
To bring Sage to life, I explored prototypes around moments when communication feels most uncertain. These concepts focus on augmenting user awareness, not automating conversation—always keeping control in the user’s hands.
Sage introduces itself after time away
When users return to a chat after a gap, Sage gently pops in to remind them it’s there for support. This reduces anxiety around forgotten context and makes it easier to re-engage in conversations without feeling lost or overwhelmed.
Sage flags unclear tone before you hit send
As users type, Sage detects potential tone issues—like sarcasm or unintended bluntness—and quietly surfaces insights. This gives users a chance to rephrase before their message is misread, making digital conversations feel less risky.
Sage steps in before you send
Right before a message is sent, Sage flags if something might come across the wrong way and offers quick suggestions. This creates a safe moment to pause, reflect, and fine-tune communication—especially helpful in emotionally charged or ambiguous situations.
Why does this matter?
An opportunity to make human connection stronger and inclusive
With growing awareness around neurodiversity, inclusive design is becoming a differentiator. Sage represents a future where AI isn’t just optimizing engagement metrics—it’s fostering meaningful connection.
Opportunities for monetization could include:
Privacy-safe enterprise tools that help teams navigate emotionally charged conversations (think AI-assisted HR conflict mediation or feedback reviews).
Licensing the tone analysis tool to messaging platforms (e.g., Slack, Discord)
Premium features for therapists, coaches, or workplace DEI teams to support inclusive communication training.
Origin story
An exploration born from personal struggles and those around me
Sage was born from a course on designing interfaces for connection, taught by Elizabeth Lin. As a designer with ADHD, I’ve experienced firsthand the “wait… was that message rude or am I overthinking it?” spiral. I wanted to design something that would help people like me feel seen, supported, and understood in digital conversations.
Next steps
So what's next?
While designing Sage, I kept running into deeper questions, many of which I’ve saved for future iterations and some of which are out of my current budget to explore due to AI token costs. However, these ideas reflect not only technical possibilities but also the human and ethical nuances of building an AI that mediates communication.
Lessons learned
Designing for AI as a problem-solving lens
AI as a design material
I learned to see AI not just as automation, but as a tool for shaping connection: clarifying tone, easing anxiety, and supporting without replacing human intent.
Humanizing technology with interaction design
Through color, type, and tone, I explored how to make Sage feel calm and empathetic, translating abstract feelings like trust and safety into concrete design.
Big-picture growth
This project pushed me to think systemically about emerging tech, balance user needs with ethics, and design concepts that are imaginative yet grounded in real-world impact.
Next up
Continue on to the next project about a fire prevention product for home cooks!