The Librarian
AI Product Design | Prompt Engineering | UX Strategy | Deployment | Community Design Role
designer + builder end-to-end product ownership
Live
in production deployed at remarkable-malasada-2bca55.netlify.app
2026
launch year currently iterating post-launch
01 / Overview
The Librarian is a Claude API-powered AI co-designer embedded in the Web Designer Anthology community. It holds the WDA design system, guides members through The Anthology Method, reviews work against UX and conversion standards, and answers design questions instantly — in context, at the moment of need.
But the purpose of The Librarian goes deeper than productivity. It was built to solve a problem that most tools don't even acknowledge: solo web designers work in isolation. They second-guess decisions alone. They search for answers alone. They evaluate their own work without a standard and without a thinking partner. The Librarian changes that — not by replacing the designer's judgment, but by standing beside it.
The Librarian isn't a chatbot. It's the co-designer that freelance web designers have never had — one that never clocks out, never needs context, and never makes them feel like they're working alone.
02 / The Problem
I didn't start by trying to build an AI tool. I started by noticing a pattern I couldn't ignore.
Inside the WDA community and in my own practice, the same friction kept surfacing. Designers weren't struggling with creativity. They were struggling with certainty. They'd get stuck in loops — second-guessing layout decisions, digging through documentation for answers they'd already looked up before, reworking sections that felt off but couldn't be diagnosed, spending hours trying to validate whether something was good or just acceptable.
I recognized it because I had been there too.
The deeper I looked, the clearer the pattern became. The bottleneck wasn't skill. It was the absence of a consistent decision-making system, an immediate feedback loop, and a way to access best practices in context — not in a course, not in a doc, but right now, mid-project, when the question actually matters.
The numbers behind the problem
Research into the freelance design experience quantifies what most designers already feel:
15–25 hours/week lostto decision loops, rework, documentation search
60–80% cognitive efficiencythe real output rate when working in isolation
$8–15K monthly revenue impactlost capacity, underpricing, and rework combined
That 60–80% efficiency figure is the one that matters most. Even when designers are technically working, persistent uncertainty — “Is this right? Am I missing something? Why isn't this converting?” — creates a constant cognitive tax that slows every decision and compounds over a full project.
Burnout typically follows within 3–6 months of sustained isolation. And over time, isolation erodes something more fundamental: professional confidence. Designers who started their careers saying “I'm a designer” slowly become “I make websites and hope they work.” That shift affects pricing, positioning, and ambition.
The real problem
Most tools help designers execute faster. None of them address the isolation, the uncertainty, or the cognitive load that makes solo design work exhausting. That's what The Librarian was built to solve.
03 / Design Process — Discover
I framed the problem less as “designers need help designing” and more as: solo web designers lack a structured, real-time thinking partner.
Not a course. Not a template library. Not static documentation. A partner that could hold the system, answer questions instantly, evaluate work against a standard, and guide a designer through every step of a project — from brief to delivery.
That reframing became the foundation for everything that followed. The question was no longer what content to create. It was how to make the system itself think with the designer.
04 / Design Process — Define
Once I stepped back from the symptoms, a clearer insight emerged: most of a designer's time isn't spent designing. It's spent deciding.
When I mapped where time actually went, I found that 18–30 hours per project were being lost to unstructured decision-making, repeated documentation lookup, and self-review loops with no clear standard. That led to a second, more important realization: the problem wasn't a lack of resources. It was a lack of integration. Designers had tutorials, templates, inspiration, and scattered knowledge of UX best practices — but none of it lived in one place, and none of it showed up at the moment of need.
That's where the opportunity opened up. Instead of creating more content, I asked: what if the system itself could think with the designer?
Not just surface information — but apply a design system, guide decisions step by step, and critique work in real time. That's when the concept of The Librarian emerged.
05 / Design Process — Develop
I approached this less like designing a feature and more like designing a behavior. The question wasn't “what should the AI say?” It was “how should the AI think?”
Defining the role
I defined The Librarian with four core responsibilities:
Hold the WDA design system — layout patterns, hierarchy, conversion principles, UX standards
Guide designers through The Anthology Method — step-by-step process support from brief to delivery
Answer in context — not generic advice, but responses tied to the specific page, goal, and user
Evaluate work — actively review against a clear standard, not just assist
Building the system prompt architecture
I built the core intelligence of The Librarian through layered system prompts — this is where the real design work happened. The system prompts encode:
The Anthology Method framework — Discover, Develop, Deliver — so The Librarian can guide a designer through each phase
Page-level frameworks for home pages, about pages, service pages, sales pages, and portfolio presentations
UX and conversion standards, including section hierarchy, CTA placement, and information architecture
Visual identity and brand design principles
Voice and tone constraints — clear, direct, and mentor-like, never performative
Instead of a single prompt, I built a modular architecture that allows The Librarian to shift between roles — guide, reviewer, explainer — and maintain context across a full design workflow.
The design insight
The system prompt is the UX. Every response The Librarian gives is shaped by the design decisions I made in the prompt architecture — what it knows, how it responds, where it pushes back, and when it defers to the designer's judgment.
Iteration — getting to the right specificity
The hardest part wasn't getting responses. It was getting the right level of specificity. Early versions were too generic, too verbose, or disconnected from real design constraints. I refined through repeated testing — feeding real client scenarios, actual page structures, and asking The Librarian to critique work rather than generate ideas.
Each iteration focused on one question: would this actually replace 10–15 minutes of thinking for a real designer? If the answer was no, I adjusted. The breakthrough came when I stopped treating it like an assistant and started treating it like a structured decision engine.
06 / Design Process — Deliver
I deployed The Librarian via Netlify and distributed it to WDA members through a direct URL — a decision that prioritized access over friction. Rather than building a complex in-platform integration, I optimized for the moment of need: a designer mid-project, on any device, needing an answer now.
Observed impact — my own practice first
Before opening it to the community, I used The Librarian extensively in my own design process. The shift I noticed wasn't just speed — it was cognitive. Decisions that used to require stepping away, searching documentation, or sitting with uncertainty for an hour became grounded and resolved in minutes.
The most meaningful change: I stopped feeling worn out at the end of a project. The cognitive load that usually accumulates over a complex build — the decision fatigue, the background anxiety of “did I miss something?” — was significantly reduced. I could produce premium work, know it met the standard, and still have creative energy left when I was done.
40–60% reduction in decision overheadper project
12–18 hours saved per projectfor a typical 40-hour website build
25–45% faster project completionwithout sacrificing quality
Community behavior shift
After launch, the type of questions in the community changed. Before The Librarian, members asked open-ended questions: “What should I do here?” After, they asked comparative questions: “Which of these two directions is stronger?” That shift signals something important. They weren't replacing their thinking — they were upgrading it.
Known iteration: image uploads
A current technical limitation is the Express payload limit affecting image uploads in the chat interface. This is a known issue on the roadmap — the fix involves increasing the payload size configuration on the Netlify deployment. Shipping V1 with a known limitation and a clear path to resolution is the right call over delaying launch.
07 / Outcomes & Impact
The Librarian's impact operates on three levels: time, revenue, and cognition. The time and revenue gains are significant. The cognitive shift is the one that actually changes how designers experience their work.
For the individual designer
Projects that typically take 40–50 hours now take 22–30 hours
1–2 additional projects per month become achievable — translating to $3,000–$6,000 in recovered monthly revenue
Decision fatigue reduced by ~30–50% per project
Fewer revision loops and restart cycles — one of the most significant hidden time costs in solo design work
Designers finish projects feeling confident in their work, not just relieved it's done
For the WDA community
Members have a thinking partner available at any point in their process — not just when the community is active
The design system lives in The Librarian, making standards accessible without digging through documentation
The Anthology Method becomes a guided experience rather than a framework members have to hold in their heads
Most tools help designers work faster. The Librarian helps them decide faster, execute with confidence, and finish without burning out. That's a fundamentally different value proposition.
08 / Reflection
Building The Librarian changed how I think about design tools. Most tools help you execute faster. This one helps you decide better. And for a solo designer, that's the difference between finishing a project and trusting the outcome of it.
The most surprising lesson was how much of the design work happened in the system prompt architecture, not in the interface. I spent far more time designing the AI's behavior — its role, its boundaries, its voice, its decision-making framework — than I spent on the visual experience. That was a meaningful shift in what “product design” means when the product is AI-native.
If I were starting over, I'd run more structured usability sessions with community members before finalizing the prompt architecture. The iteration I did was real and grounded, but more structured observation earlier would have surfaced the cold-start UX challenges sooner.
V2 priorities: resolve the image upload payload limit, build conversation history so The Librarian retains context across sessions, and develop a structured feedback mechanism so member interactions can inform ongoing prompt refinement.
Comments