Advokit

Advokit helps people navigate UK welfare benefits and services with clear, accessible guidance co-designed with people with aphasia.

Live at advokit.vercel.app.

Most benefits guidance is written at a reading level that excludes the people who need it most. Advokit presents the same information in short sections, plain language, and read-aloud audio, alongside first-hand stories from people who have been through the process themselves.

Features

• Benefit guides — step-by-step pages covering 23 benefits and services (PIP, Attendance Allowance, Universal Credit, Blue Badge, Carer's Allowance, Motability, travel concessions and more), each with a printable summary and downloadable help sheets.
• Benefit Buddy — a chatbot (OpenAI gpt-4o-mini) that answers questions about eligibility and the application process in plain English.
• Accessibility first — Atkinson Hyperlegible typeface, read-aloud on every page via the Web Speech API, progressive disclosure of detail, and video/audio stories with full transcripts.
• Comparison views — sortable table and list views across all covered benefits.

Tech stack

Framework	Next.js 16 (App Router) + React 19
Language	TypeScript
Styling	Tailwind CSS v4, PrimeReact
Data	data/home.json (benefit content), MongoDB (chat logs + feedback)
AI	OpenAI API
Hosting	Vercel
Analysis	Streamlit + pandas (analysis/)

Getting started

Requires Node.js 22+.

git clone https://github.com/HumphreyCurtis/Advokit.git
cd Advokit
npm install
cp .env.example .env.local   # then fill in the values
npm run dev

Open http://localhost:3000.

Environment variables

Variable	Required	Purpose
OPENAI_API_KEY	yes	Powers Benefit Buddy (app/api/chatbot/route.ts)
MONGODB_URI	yes	Chat log and feedback storage
MONGODB_DB	no	Database name, defaults to advokit

app/lib/mongodb.ts throws at import time if MONGODB_URI is missing, so all three need to be present before npm run dev or npm run build will work.

Project structure

app/
  api/chatbot/     Benefit Buddy API route
  benefits/[slug]/ Individual benefit guides
  benefits-table/  Comparison table
  benefits-list/   Comparison list
  benefit-buddy/   Chatbot UI
  stories/         Video stories and transcripts
  components/      Shared UI (ReadAloud, AudioPlayer, ProgressiveTips, ...)
  lib/             Benefit data accessors and MongoDB client
data/home.json     Source of truth for all benefit content
public/            Images, PDF help sheets, transcripts
analysis/          Streamlit app for reviewing chat logs
a-resources/       Brand assets (logos, palettes) — not served

Adding or editing a benefit

All benefit content lives in data/home.json, typed by app/types/index.ts. Add an entry with a unique slug and the guide page, list and table views all pick it up automatically — no component changes needed.

Chat log analysis

analysis/ is a separate Streamlit app for reviewing Benefit Buddy conversations. It reads .env.local from the repo root.

conda env create -f analysis/environment.yml
conda activate advokit-logger
streamlit run analysis/app.py

Scripts

Command	Description
npm run dev	Development server
npm run build	Production build
npm run start	Serve the production build
npm run lint	ESLint

Disclaimer

Advokit was co-designed with people and communities living with aphasia in London. The information here is general and may not fit every case — rules and forms change, and councils or assessors may do things differently. This is not legal advice. Please check official sources such as GOV.UK — Benefits or Citizens Advice, and speak to an adviser if you can.

License

MIT — see LICENSE.