GrantMate: Custom GPT with Onboarding
(In a hurry? Feel free to jump ahead to the TL;DR for a quick recap.)
I created GrantMate, a custom GPT designed to support overwhelmed grant writers facing tight deadlines, incomplete inputs, and limited SME access. Built as a strategic writing partner, it helps users clarify what they know, identify what’s missing, and generate momentum with focused, just-in-time drafting support.
Tools Used: ChatGPT, Articulate Rise, Snagit
Audience: New Grant Writers
Responsibilities: Instructional Design, Needs Analysis, Prompt Engineering, Testing & Iteration
The Problem
Grant writers frequently face intense pressure — tight not-negotiable deadlines, vague prompts, and little to no SME access. The early stages of writing are often daunting: solving logic-model puzzles with incomplete data, wrestling with silence from stakeholders, or staring at a blank page with hours ticking away. In these moments, even seasoned professionals can feel isolated and stuck.
The Solution
To address this, I built GrantMate, a purpose-built Custom GPT that acts as a strategic writing partner when the chips are down. It begins by prompting users to articulate what they know—and what they don’t—then offers draft language, placeholder data, and reframing suggestions that align with funder expectations. It’s designed not as a generic writing assistant but as a responsive collaborator that understands the lived reality of grant writing: ambiguity, complexity, and constant time pressure.
My Role
Part of why I began this project was to stretch myself, but also, I recently entered the world of grant writing and am trying to find ways to speed up my learning. Though I have a great deal of experience as a writer, I am new to the land of grants and the unique challenges it presents. So, I kicked things off by documenting real pain points: muddled bullets, missing budget details, last-minute edits. Then, I wrote an initial prompt combining clear rules, tone guidelines, and context.
My Design Process
While I had a good starting point, I definitely needed some iteration to improve upon my GPT. As grant deadlines can be unforgiving, I began with some “known” hypothetical applications and watched GrantMate sort through the best way to help—each iteration refining its rules, its empathy for the user, and its ability to ask clarifying questions, mimic SME tone, and scaffold partial information. I tested it, adjusting its logic flow until it felt like a true thinking partner.
I approached the development of GrantMate the way many grant writers approach their work: fast, focused, and grounded in real constraints. My goal was to create a tool that could mirror the pressures of the writing process itself—working with incomplete inputs, adapting quickly, and offering high-value feedback without delay.
To build the CustomGPT prompt, I began by identifying the common pain points I experience in my own professional grant writing: moments when I’m working solo, when SMEs are unavailable, or when I’m staring down a complex narrative section with only a few bullet points to work from. These scenarios shaped the GPT’s rules, tone, and priorities.
I acted as both the designer and primary test user, embedding GrantMate into my actual workflow across multiple live grant applications. This allowed me to test its utility in real-time: Could it help me get started faster? Was it asking the right clarifying questions? Did it give me usable draft language I could shape further? As I wrote, I iterated—refining how GrantMate handled missing information, how it responded to vague prompts, and how it scaffolded incomplete sections.
The result is a GPT that reflects the real-world rhythm of grant writing: scrappy, strategic, and focused on momentum. It doesn’t replace the writer—it supports them in the moments when they’re most likely to stall or get stuck. My goal was to create an AI tool that didn’t just generate content—it needed to reduce cognitive load, create momentum, and offer just-in-time writing support tailored to the high-pressure world of grant writing. I used rapid prototyping, user simulation, and instructional scaffolding to iterate the design in response to real-world scenarios.
To support adoption and clarity, I created a short Rise module to onboard new users to GrantMate. The module walks users through the tool’s intended purpose, core features, and best-use scenarios, helping grant writers understand how to interact with the GPT effectively. It focuses on quick wins, minimizing overwhelm, and aligning expectations for genre-specific, just-in-time writing support.
A Few Key Observations
Iteration Through Use
I used the GPT in real-time while drafting grant applications, testing it under the same conditions my audience would face: tight deadlines, incomplete SME input, and high ambiguity. This helped me identify what kinds of support actually moved the needle for overwhelmed writers. A major insight? The most powerful starting question was:
“What do we know so far, and what’s missing?”
This became the cornerstone of the tool’s logic—simple, grounding, and versatile.
Designing for Real Prompts and Real Programs
I added an optional input field for users to paste the actual grant prompt or funder guidelines they were working from. If users had this information, GrantMate would tailor its output more precisely; if not, it would gracefully explain why that input is helpful and continue to assist using general scaffolds.
Managing Cognitive Load
Because grant writing is often nonlinear and emotionally taxing, I structured the GPT to chunk tasks into manageable pieces. It frequently asks the user if they want to “stay here or switch focus,” mimicking natural cognitive breaks. It also reinforces progress by summarizing what the user has already accomplished—adding psychological safety and momentum.
Placeholder Language & Visual Clarity
I wanted users to know at a glance which parts of the output still needed revision or SME input. While GPT doesn’t support color coding natively, I established a consistent visual pattern:
All placeholder text is bracketed (e.g., [Insert impact metric here]).
This helped users quickly spot what still needed their attention, even under deadline pressure.
Guardrails for Genre and Use
In early testing, GrantMate was a little too helpful—it offered to write blog posts about tomatoes when asked. I restructured its prompt to make it genre-specific: GrantMate now politely declines off-topic requests and recommends other tools for blog writing or marketing content. This reinforced its identity as a trusted, purpose-built partner.
Rapid Prototyping with SAM
I used the Successive Approximation Model (SAM) to test and revise continuously. Each live test surfaced new insights about how overwhelmed users behave, what kinds of prompts reduce friction, and how much scaffolding is “just enough.” The goal was never to replace the writer—but to get them unstuck and moving forward faster.
Impact
GrantMate fills a critical gap in the grant writing process by offering just-in-time narrative support that feels strategic, not generic. As both designer and primary user, I integrated the GPT into my real-world workflow—testing it against live deadlines, incomplete project inputs, and pressure-filled grant cycles. The result? Noticeable gains in clarity, speed, and creative momentum.
By prompting me to articulate what I knew, identify what was missing, and generate draft language I could immediately refine, GrantMate helped reduce cognitive overload and accelerate my writing process. In several cases, it helped me break through writer’s block and begin drafting sections hours earlier than I would have otherwise.
Feedback from early testers—including fellow grant writers and nonprofit professionals—highlighted GrantMate’s conversational tone, intelligent scoping of vague inputs, and ability to transform messy bullet points into usable, funder-aligned drafts. Users described it as “like brainstorming with a second brain,” and noted its value as a thought partner when internal collaboration wasn’t immediately available.
Ultimately, the impact of this project goes beyond convenience—it provides grant professionals with a risk-free environment to think out loud, iterate quickly, and write with confidence, even in the most deadline-driven situations.
Results & Takeaways
Building GrantMate gave me a hands-on opportunity to blend instructional design, prompt engineering, and performance support into a real-world solution. Even in its early use, the GPT showed strong potential to reduce overwhelm, accelerate drafting, and help grant writers move from ambiguity to action—especially when working solo or under deadline pressure.
Throughout testing, I observed how the tool helped users:
- Get started faster with structure and momentum-building prompts
- Break through mental blocks by asking the right clarifying questions
- Maintain narrative coherence even with incomplete information
- Build confidence through affirming feedback and scaffolded steps
As part of my testing process, I ran the same prompts through both GrantMate and standard ChatGPT. The difference was clear: generic ChatGPT responses were often exhaustive, dense, and cognitively overwhelming, especially for users already under pressure. In contrast, GrantMate delivered focused, digestible, and action-oriented guidance, always grounded in the reality of limited time and information. This reinforced how important it is to design AI not just for accuracy—but for emotional and situational relevance.
One of my biggest takeaways as a designer was this:
The best AI support isn’t about doing the work for the user—it’s about giving them the right nudge at the right time to help them keep going.
This project also deepened my skills in:
- Prompt iteration and troubleshooting
- User-centered design for knowledge workers
- Designing AI interactions with clear constraints and ethical boundaries
Ultimately, GrantMate is more than a clever tool—it’s a case study in how AI can amplify human problem-solving when built with empathy, clarity, and real needs in mind.
TL;DR
As an instructional designer and new grant writer, I created GrantMate—a genre-specific, just-in-time Custom GPT that helps professionals navigate high-stakes grant applications when time is short and SME input is scarce. Unlike generic AI tools that overwhelm with verbosity, GrantMate scaffolds the writing process with targeted prompts, momentum-building feedback, and placeholder logic tailored to real funder expectations.
Built using the Successive Approximation Model (SAM), I tested and refined GrantMate through live use on real applications. Key insights shaped its design: a single clarifying question—“What do we know so far, and what’s missing?”—became its foundation. The result is a smart, kind writing partner that breaks down complexity, adapts to incomplete inputs, and empowers users to make tangible progress—even under deadline.
Jen Popa | Instructional Designer 