We created Donde to give research a home: a place to compare perspectives, understand the differences, and return to the evidence behind a decision.
Explore the work01The opportunity
A useful video, a saved post, a recommendation that contradicts the last one. Gathering sources is easy. Understanding how they fit together takes work.
We began with travel research, then developed twelve research spaces for decisions such as buying a product, moving somewhere or evaluating a technology release. Our hypothesis: research is more useful when agreement, differences and personal context stay visible.
How could we make research easier to use without hiding the reasoning behind it?
Choose a research space and add context.
Add supported sources you choose.
Explore patterns and meaningful differences.
Return to what each source actually said.
02From zero to one
We shaped the idea, created its identity, designed the experience and built the system that makes it work.
Research with context. The product choices, visual language and architecture developed together. Our research combined product, domain and technical exploration to shape the decisions below.
We framed a source-based decision workspace, defined its capacity model and narrowed the first useful research journeys.
We reviewed source formats, attribution needs and failure cases, then built a category and evidence model for twelve decision spaces.
We created the Donde wordmark, curious characters, lavender palette and an approachable visual language for dense research.
We designed progressive disclosure from project to finding to evidence, including waiting, partial results and recovery states.
We built an Expo and React Native application with shared components, routing and responsive layouts for web, iOS and Android.
We connected Supabase identity and data, source APIs, Gemini analysis, credit accounting and billing, alongside the product website.
A complete brand identity
We gave a dense research product a curious, approachable identity. The brand welcomes people into the work, while the evidence interface stays calm and readable.
We created a connected identity system: logo, character, color, typography, iconography and interface patterns, then carried it into the website and distribution materials.
03Product judgment
What we chose, why it mattered and what it cost.
A polished summary can make conflicting advice look more certain than it is. We kept synthesis distinct from source positions, with links and timestamps where available. Source counts represent perspectives, not a truth score.
Showing nuance takes more space and attention. It gives people the information to judge which advice fits their situation.
A trip, a purchase and a software upgrade need different questions. Defined research spaces supply relevant categories and context while sharing the same project, finding and evidence model.
Boundaries make the starting point clearer. The reusable system still lets us change a space without rebuilding the entire product.
Source processing is asynchronous. We connected progress per source, partial results, retries and recovery for abandoned jobs. Successful evidence and its credit settlement commit together so interface state and billing agree.
Recovery adds backend complexity. It protects completed work and gives people a useful way to continue after a failure.
The implemented flow, simplified
Track each source independently.
Keep ready results available.
Recover without restarting the project.
Settle the credit with usable output.
Product interaction and engineering are designed together.
04Platforms & architecture
We used Expo and React Native so one product codebase serves web, iOS and Android, including tablet layouts.
Research logic, state, design tokens and core components are shared. Expo Router connects the experience; responsive layouts and platform-specific authentication and payment paths handle the differences.
Wide screens keep findings and evidence side by side. On a phone, the same task becomes a focused sequence.
Actual web and native Android interfaces with illustrative research data. Web captures: October 2026. Native captures: September 2026. Select any screenshot to view it in full.
The user selects supported sources and frames a question.
Server-side jobs retrieve and analyze sources, storing attributed evidence.
Shared clients present progress, findings and evidence within account boundaries.
The working stack
| Layer | Tools | Purpose |
|---|---|---|
| Experience | Expo · React Native · TypeScript · Expo Router | Shared web and native product logic with platform-aware layouts. |
| Data & identity | Supabase Auth · Postgres · Realtime · Edge Functions | Accounts, private projects, live progress, processing jobs and credit accounting. |
| AI & source APIs | Google Gemini Interactions API · TranscriptFetch · Supadata | Source caption retrieval and evidence-based analysis, with fallback and recovery paths. |
| Commerce & delivery | Stripe · Apple / Google billing · EAS · Netlify | Platform-specific payments, native builds and web-app delivery. A separate branded marketing website supports discovery. |
Security is product work
Donde documents SOC 2 aligned security work. This is an engineering baseline, not an independent SOC 2 attestation or a claim that every control has been audited.
SOC 2 is an independent examination of an organization’s controls against trust services criteria. Aligned controls and readiness work do not equal an auditor’s SOC 2 report, and “certified” would misstate our status. About SOC 2
Owner-scoped data access and authenticated server functions protect private research.
Credit settlement belongs to backend transactions, not a client-side counter.
Security readiness documentation addresses secrets, dependencies, incident response and recovery.
A managed platform helps with infrastructure; AS1 still owns application access, data policies, secrets and integration decisions. A provider’s SOC 2 report does not cover our application automatically. Supabase’s shared responsibility model
05Brand, marketing & distribution
We carried the curious Donde identity into clear messaging about the job: compare sources, understand differences and keep evidence beside a decision.
We connected identity, messaging, product education and acquisition. Conversion means helping the right visitor reach the next useful action, then measuring whether that action leads to value.
Visit the Donde websiteProduct and use-case pages, guide structure, descriptive metadata, canonical URLs, structured data and a sitemap connect the website to specific research needs.
The website explains the experience before asking for a commitment. Web entry and store links lead toward a first research project; free source credits introduce the capacity model.
We connected browser access with App Store and Google Play discovery, product screenshots, descriptions and platform-specific download routes. Web and native billing follow their respective platforms.
From discovery to first value
Discover a page for a decision that matters.
See the workspace and the evidence model.
Start on web or download the native app.
Create a project and inspect a useful finding.
We would measure search-to-visit, product-entry clicks, first successful research and return use, then paid capacity demand. These are validation goals, not reported conversion gains.
06Delivery & learning
We delivered the product system described here. Product-market fit has not been established.
These case studies show our ability to connect brand, product, design and engineering. The next stage is to establish adoption, paid demand and customer outcomes.
A common foundation for distinct decisions.
Web, iOS and Android through Expo.
Identity, product, interface and engineering.
What the build taught us
Trust lives in the whole system: readable evidence, visible uncertainty, recoverable processing and fair credit handling support the same promise.
The next validation
We need to test whether people inspect evidence, return to projects and use briefs in real decisions. Generating more findings alone does not establish value or product-market fit.
The right team for the stage
We can take a product from idea to a working release, including its brand, frontend, managed backend and integrations.
The team model should evolve with the risk and demands of the business.
Before market fit is proven, scope and direction change quickly. Where appropriate, Supabase gives us a practical managed foundation. We can own application data models, authentication, backend functions and integrations while iterating on the product.
We recommend a lead responsible for data architecture, backend reliability, infrastructure and cybersecurity, working with AS1 on brand, product and design. Supabase may remain the right platform; scale calls for stronger engineering ownership, not an automatic rewrite.
Health-tech, fin-tech and other sensitive products: involve specialist engineering and security leadership from discovery. Data sensitivity, regulatory duties and operational risk can require that collaboration before market fit or rapid growth.
Your idea could be next
Bring us the problem, the people it serves and the stage you’re at. We’ll help shape a focused path from idea to product.
Discuss your project