Back to case studies
Case study/PReasy/featured

PReasy – AI for PR and Public Affairs

AI-assisted document workflow for PR and Public Affairs teams.

The problem

PR and Public Affairs work was spread across drafts, email threads, shared drives, and ad-hoc review chains. Nothing connected the document to the campaign it belonged to.

The solution

A workspace where PR documents, campaigns, and stakeholder context live together. AI assists with drafting, edits, and structure — but the document, not the prompt, is the artifact.

Behind the build

The trick was making AI-assisted editing feel like a tool, not a takeover. Every AI action is reversible and visible in the document history. The product is live at preasy.ai.

Project notes

PReasy is a workspace for Public Relations and Public Affairs teams. The product exists because PR work — briefs, statements, talking points, stakeholder maps — is mostly typing, reviewing, and chasing approvals across tools.

PReasy puts documents, campaigns, and AI assistance in one place.

Decisions worth noting: • Documents are first-class. AI suggestions land in the document with an explicit accept/reject step — they do not silently overwrite. • Campaigns wrap documents. A document always knows which campaign and which audience it belongs to. • Workspace history is the audit trail. Every edit, every AI suggestion, every approval is a row. • Hosted on Google Cloud, served through Vercel, with React Query handling the document/campaign cache.

The product is live and used by PR teams. I do not publish customer numbers.

More case studies

Other products built end-to-end.

Commerce / 01Live

Headless B2C Ecommerce Platform

Built end-to-end as sole backend and platform developer on MedusaJS v2. Storefront APIs, admin workflows, Razorpay payment integration, notification delivery (Novu, FCM, ZeptoMail, Twilio SMS), Jio DLT compliance, Redis caching, and AWS S3 asset storage — all live in production.

8 integrations
third-party
Idempotent
payment webhooks
Live
in production
Capability description · NDA · no screenshots
system flow
architecture · not a demo
01
Storefront API
02
Payments
03
Notifications
04
Infrastructure
Stack
MedusaJS v2 · Next.js · PostgreSQL · Razorpay · Redis
Commerce / 02In Progress

Multi-Vendor B2B Marketplace

Built on MercurJS and MedusaJS v2 with custom module development throughout. Configurable pricing engine with vendor-specific rules, order splitting with vendor assignment, finance and invoicing layer, and bulk order import pipeline — in active development.

Custom
pricing engine
Multi-vendor
order splitting
In Progress
active build
Capability description · NDA · no screenshots
system flow
architecture · not a demo
01
Vendor Onboarding
02
Pricing Engine
03
Order Splitting
04
Finance
Stack
MercurJS · MedusaJS v2 · Next.js · PostgreSQL
Case / 03

FPMSMulti-tenant SaaS with Stripe Connect and tenant-isolated workspaces.

A multi-tenant SaaS where each freelancer-client pair runs in its own scoped workspace. Stripe Connect carries marketplace payments end-to-end, and the database is the source of truth for what was paid for what.

Next.jsTypeScriptMongoDBPrisma ORM+3
Read case study
Case / 04

BizDirLocation-aware directory MVP, shipped in two months.

An MVP with map-based filtering, category and rating facets, and SEO-friendly business pages. Built and launched in roughly two months with 500+ business records seeded and indexed.

Next.jsTypeScriptNode.jsExpress.js+2
Read case study
    PReasy – AI for PR and Public Affairs | Kelvin Gevariya | Kelvin Gevariya