HeyEnso
A web workspace where specialized agents investigate markets, competitors, creators, visibility, and campaign ideas around a shared project context.
Give the marketing work a team
HeyEnso grew out of a frustration with how fragmented marketing research can become. A project’s identity might be spread across a website, app store listing, social profile, search results, analytics, campaign notes, and a dozen open tabs. I wanted the experience to feel more like working with a small team of specialists than opening one more dashboard.
The user starts with a project and gives the system enough context to understand its audience, category, platforms, official links, and goals. Enso acts as the marketing manager, then coordinates specialist agents: Ravi looks at competitors and market signals, Lena examines website and app visibility, Mira explores creators and partnerships, and Leo develops campaign directions.
Context before cleverness
Each session is grounded in the project rather than treated as an isolated prompt. The agents delegate focused work, gather evidence, and return structured findings with source links, recommendations, and saved activity. A persistent workspace makes the research useful after the first conversation, while the platform scope changes what each specialist should investigate.
The result is intentionally practical: a clearer next decision, a shortlist worth pursuing, a creative direction, or a better understanding of what people are saying about the project. External research calls use credits, so the product also has to explain cost, handle low-credit states, and stop safely when the work cannot continue.
Building the system
The interface uses Next.js and TypeScript. Firebase Authentication, Cloud Functions, Firestore, and Storage hold project context, sessions, activity runs, saved research, and generated media. Genkit and Gemini power the generation layer, while Exa and Apify support web and public-profile research. Stripe handles the paid plans and credit packs.
The lesson from the build
Reliable AI work is mostly about boundaries. Context, schemas, source validation, memory, tool contracts, loading states, and failure paths all shape whether an agent feels useful. HeyEnso was my first serious attempt at making those details feel approachable instead of exposing the machinery behind them.