The problem
Social media managers and marketers produce a steady stream of content: captions, images, short videos and voiceovers. Generic AI tools can create each of these, but they do not know the brand's voice, so every output needs rewriting before it can be posted.
Work also ends up scattered across several tools, with prompts, drafts and past results kept in different places. For a marketer handling content daily, that is a lot of switching and a lot of manual effort per piece.
The goal was one workspace where content of different types is generated in line with the brand, where earlier work can be found again, and where a person logs in and gets on with it.
What we built
We built a React and TypeScript single-page app integrated with Google's Gemini API.
- Generators for text, images, video and voiceover, so the common content types live in one place
- A brand-aware assistant, so output is shaped by the brand instead of arriving generic
- Four modules: Dashboard, Generators, Brand Assistant and History, each with a clear job
- Full authentication, so each user has their own access to the app
- Dark and light mode, for people who spend long hours in the tool
- State persistence in the browser, so the app keeps its state between visits
TypeScript was used so the code stays predictable as features are added. An app like this is a web application rather than a website, so it falls under our React web development work.
The result
Marketers now have a single workspace for generating brand-aligned content, with less manual effort per piece.
The two outcomes we aimed at were less manual content effort and output that matches the brand, and both were the design targets of the Brand Assistant and the Generators modules.
Frequently asked questions
Can you build an AI tool on the Gemini API for my business?
Yes. We have built a content platform on the Gemini API using React and TypeScript. The first step is a discovery call to agree what the tool should generate and who will use it.
Will the AI actually follow our brand voice?
The brand-aware assistant is designed so output reflects the brand. AI output should still be reviewed by a person before it is published, and we say so plainly when scoping.
How long does it take to build a platform like this?
A full app or platform usually takes 4-8 weeks, depending on how many generators and modules you need. A smaller fix or add-on takes 1-2 weeks.
How much does it cost?
Pricing is project-based. We hold a free discovery call, learn what you need, and then give a fixed quote. We do not publish price lists because the scope decides the cost.