AI audience platform
For an Indian marketing-technology company, Joistic rebuilt its audience platform from the ground up. Marketers describe the audience they want in plain English, AI turns that into a plan, the platform counts it live, and the audience is built for ad campaigns. It went from first commit to production in about five weeks.
Ad-tech company, India (under NDA)
About 5 weeks
223, 93.6% coverage
Dedicated AWS account, Mumbai
About the client
The client runs an audience platform for brands and agencies. We can't name them because the work is covered by an NDA. An earlier version had been built as a prototype by another agency and wasn't ready for real customers.
What needed fixing
Marketers know who they want to reach but not how to express it as filters and segments. The old prototype ran on demonstration data, with sample counts and demo logins instead of a working product.
The client needed a real product: separate workspaces for each brand, real audience counts, and infrastructure they could trust with customer data.
The system, piece by piece
Studio: describe an audience in plain English
A marketer types who they want to reach. AI turns the request into an audience plan, which is counted live against the audience store and can be refined into cohorts and saved.
Separate workspaces for each brand
Each brand has its own workspace and team, with roles for admins, members and data operators. Every query is limited to the signed-in brand's data.
Data uploads
Data operators upload large category files (up to 200 MB). A background worker validates them, protects personal identifiers by hashing them, and loads them into the audience store.
Builds and ad activation
Saved audiences are built into member lists for each brand's own connected Google Ads or Meta account, with brands taking turns in the processing queue.
Operations console
Platform admins see system health, uploads, jobs, AI usage and activations in one console.
Website lead capture
The marketing website's demo-request form feeds straight into the platform and emails the sales team.
Where AI is used
- Plain-English audience planning: the AI reads the marketer's request and produces a structured audience plan with criteria and signals.
- The platform can run on Gemini or Claude, so the client isn't tied to one AI provider.
- AI usage is tracked in the operations console so costs stay visible.
Built for customer data
- Personal identifiers are hashed before they're stored, and each brand only ever sees its own data.
- Production runs in its own AWS account with encrypted databases and secrets, audit logging, threat detection and alerts.
- Every change goes through 223 automated tests and browser tests before release, and infrastructure changes are reviewed as code.
What changed
- The platform went live in production in October 2026, about five weeks after the rebuild started.
- Real audience counts and separate brand workspaces replaced the demo prototype.
- Google Ads and Meta activation is built and switches on once the platforms finish their app reviews.
A prototype proves an idea. Customer data needs a product: separate workspaces, tests and infrastructure you can defend. That's the gap this rebuild closed.
- Sep 2026
Rebuild started from the client's product spec
- Late Sep 2026
Studio, uploads and brand workspaces on staging
- Oct 2026
Production live on a dedicated AWS account
TicketRoot
How Joistic built TicketRoot's event platform: ticketing, check-in, cashless wallets and AI photo matching, used at Comic Con and the Delhi Saras Mela.
Manufacturing & IoTPES Electrical
How Joistic built a field-staff attendance platform for PES Electrical, a smart-meter manufacturer, with QR check-ins, live location and fake-GPS detection.
Digital publishingKITABOO
How Joistic rebuilt kitaboo.com with a CMS, approval workflow and an AI connection, so KITABOO's marketing team creates and edits pages themselves.
Employee engagementCorporate wellness platform
How Joistic built a web and mobile step-challenge platform for a 10,000–20,000 person company, with wearable sync, leaderboards and team rivalry.
Beauty SaaS · CanadaLash Assist Pro
How Joistic built Lash Assist Pro, a live subscription app for lash artists in Canada, with Ivy, an AI assistant that explains retention and advises on each session.
Travel services · USATravel-document courier service
How Joistic built booking, shipping and AI passport-photo tools for a US travel-document courier, with automatic deletion of personal data after 31 days.
Sharper operations. Smarter teams.
Let's build yours.
Start with one workflow and see the results before you expand.
Not ready for a call? Send us a short brief and a founder will reply within one working day.