I design words, flows, and trust systems for complex products & AI
I help people understand what happens next across onboarding, checkout, error states, and AI-powered interactions. Beyond writing, I help teams establish frameworks and playbooks to adopt AI safely.
Good product language is evidence of good product thinking. It shows the user what changed, what matters, and what they can safely do next.
In AI products, that means designing for calibrated trust: clear limits, visible reasoning, graceful failure, and user control instead of vague promises of magic.
I help companies navigate the transition to AI, showing teams how to combine generative tools with strict content guidelines and human-in-the-loop editing to preserve user trust.
Skirts & Sparkles: Localizing Dutch Nightlife for Expats
Amsterdam's music scene is deep but culturally and linguistically opaque for international newcomers. I designed a vibe-based discovery flow and escrow-protected ticket resale platform, integrating a plain-English jargon decoder, transit warnings, and anti-scalp ticket reassurance systems.
AI Enablement & Team Adoption
How I Help Companies ScaleAI Content Playbooks
Standardizing prompt libraries, human-in-the-loop workflows, and QA templates for teams integrating generative LLMs into their product stack.
Calibrated Trust Rubrics
Designing guidelines, confidence metrics, and clear fallback pathways for AI applications to manage user trust during inevitable edge-case errors.
Enablement Workshops
Facilitating training for product, design, and content departments to scale AI writing workflows safely, consistently, and ethically.
Featured Work
03 Proof PointsSelected Portfolio
03 Featured · 09 ArchivedPawPath: Redesigning the Pet Health Dashboard
Pet owners were managing three to five apps to track their animals' health. PawPath had the right data and completely the wrong structure. I rebuilt the information architecture from scratch around how owners actually think: by pet, not by category.
Nobody Reads Error Messages. So I Made Them Worth Reading.
Audited 120+ error messages across a B2B platform. Most blamed the user or said nothing useful. I rebuilt the entire error taxonomy with a three-part framework, translating the standards into system instructions for Copilot.
Scaling Copy Audits with AI and Team Enablement
Trained a 12-person content team on a hybrid human-in-the-loop AI prompt playbook, completing a 200+ string audit in three days and establishing permanent 60% operational efficiency.
Rewriting a Checkout Flow That Was Bleeding Revenue
A major e-commerce client was losing 40% of users at payment. I rewrote every string in the four-step checkout, turning legal-speak into confident next steps.
The First Five Minutes: Rewriting Onboarding for an EdTech Platform
New users were signing up but not completing their first lesson. I cut the copy by 60%, front-loaded value, and replaced walls of text with progressive disclosure.
Teaching Users to Trust an AI They Can't See
Redesigned the first-run experience for an AI-powered feature, replacing vague "magic" language with honest copy that set real expectations and gave users real control.
When the Bot Doesn't Know: Designing Graceful Failure
Wrote the full conversation flow and fallback logic for an AI assistant's edge cases. The moments where most chatbots lose trust are the ones I focused on hardest.
Bloom: A Kinder Postpartum Check-In
Bloom's wellness check-in was designed by engineers. I redesigned the entire experience around emotional permission, validating the feeling before trying to measure it.
Nuzzle: Matching by Compatibility, Not Just Cuteness
Nuzzle wanted something smarter, a system that surfaces lifestyle compatibility. I designed the conversational matching flow from research to prototype.
The First Six Weeks: What New Mothers Actually Need at 3am
I ran a 6-week diary study with 12 first-time mothers to find out what they actually needed, and discovered the entire product strategy was answering the wrong question.
Why Pet Parents Churn: It Wasn't the Features
Twenty in-depth interviews and a behavioral log analysis later, I'd found the real culprit for 68% 30-day churn: a single onboarding moment that made users feel like data clerks.
Trusting AI Vet Advice: The Confidence Paradox
A 450-person survey and 15 think-aloud sessions revealed that calibrated uncertainty built more trust in veterinary AI advice than overconfident answers.
An assistant should earn trust before it asks for attention.
This prototype shows the tone, boundaries, and fallback behavior I design for AI-supported product experiences.
Find the decision moment
I map where users pause, second-guess, or abandon the flow, then write for that exact moment instead of polishing copy around it.
Design for calibrated trust
I make system limits, confidence, next steps, and user control visible, especially when AI is involved or the stakes are high.
Leave a system behind
I turn one-off fixes into frameworks, taxonomies, and reusable patterns so teams can keep making better content decisions after launch.
Annika Bhargava
UX Writer and AI Enablement Strategist with 8+ years shaping how products talk to people. I've spent my career at the intersection of language and technology, writing the words that help people trust, understand, and actually use the products they interact with every day.
I believe the best UX writing is invisible. You don't notice it. You just know what to do next. In the age of generative AI, that clarity is the difference between adoption and churn.
I partner with product teams to build clear conversation structures, error message frameworks, and AI trust guidelines. I also help organizations scale their AI copy processes, building customized content playbooks and training internal teams to run efficient, high-quality human-in-the-loop workflows.
I've worked across consulting, edtech, and enterprise software—from establishing content standards at Microsoft to driving conversion at Capgemini and guiding AI workflows at the British Council.