Arius Automation
Services>Web Applications and APIs
WEB APPLICATIONS AND APIS

Web applications built as products, not projects.

Full-stack SaaS applications, customer portals, marketplaces, and public APIs. From MVP through production scale. Built like software you intend to operate for years, not throw away after launch.

Representative customer-facing web application. Real products are designed and built around your specific users and business.

Book a product scoping call

Built in TypeScript and React. Deployed on Vercel, AWS, GCP, or your own infrastructure.

WHAT WE BUILD

Six shapes of web product we ship most often.

Web applications come in shapes. Below are the six we ship most often. Your project probably maps to one of them or a combination of two.

SaaS MVPs
Early-stage SaaS products. Auth, billing, the core product surface, the first three to five key features. Built so it can grow into the full product without rebuilding everything.
TYPICAL BUYER
Founders pre or post-seed who have validated the idea and need to ship the first version
Customer portals
Customer-facing dashboards for companies that already have the business but lack the software interface. Order history, account management, support, billing, settings.
TYPICAL BUYER
B2B companies with established customers who manage relationships through email and spreadsheets today
Two-sided marketplaces
Platforms with two user types that transact through your software. Buyer-seller, customer-provider, supply-demand. Includes the matching, transaction, and trust layers.
TYPICAL BUYER
Marketplace founders, vertical SaaS companies expanding into transactions
Public APIs and developer platforms
Public REST or GraphQL APIs that your customers or partners build against. Includes documentation, authentication, rate limiting, and developer experience tooling.
TYPICAL BUYER
Platform companies, B2B SaaS companies who realize their integration is becoming a product
Web product rebuilds
Replacing a legacy web product with a modern one. Often a Rails monolith from 2014 or an early-stage codebase that has accumulated debt. We rebuild without losing the data or breaking the business.
TYPICAL BUYER
CTOs at growth-stage companies, founders whose original codebase has outlived its useful life
Internal product teams as a service
Ongoing engagement where we operate as your embedded product team. Building features alongside your team, or building everything until you hire your own engineers.
TYPICAL BUYER
Founders who need a product team for six to twelve months, not a long-term hire commitment
THREE PROJECT SHAPES

Where you are determines what the engagement looks like.

Web product work comes in three shapes depending on what you are starting with. Below is how each one runs in practice.

01
MVP from scratch
Idea validated, no code yet.
You have validated the idea, talked to enough customers, and need to ship the first version. We build the core product, auth, billing, and the first three to five features. Designed to be extended, not thrown away.
TYPICAL DURATION
10 to 16 weeks
WHAT YOU GET
A working product in production
02
Rebuild or major upgrade
Existing product, time for a new one.
The original codebase has aged out. Rails 4, jQuery, a monolith that fights every feature request. We rebuild the product on modern foundations, migrate the data, and cut over without breaking the existing business.
TYPICAL DURATION
16 to 32 weeks
WHAT YOU GET
A modern codebase plus migration
03
Ongoing product development
Live product, embedded team.
You have a product in production and need engineering capacity to build features, fix issues, and ship improvements. We work as an embedded extension of your team, picking up tickets, attending standups, shipping in your sprints.
TYPICAL DURATION
Three to twelve months, often longer
WHAT YOU GET
Sustained shipping capacity
ARCHITECTURE DECISIONS

Six choices that shape every web app build.

Every web app project resolves six architectural decisions early. Each one has trade-offs that matter for years. Below is how we think about them so you can see the reasoning before we start building.

01
Multi-tenant or single-tenant?
OUR DEFAULT
Multi-tenant by default. Shared infrastructure, isolated data per customer, dramatically lower operating cost.
WHEN WE PICK DIFFERENTLY
Single-tenant when compliance (HIPAA strict, defense, certain financial regulations) demands it, or when customer scale and customization needs make multi-tenant infeasible.
02
Monolith or services?
OUR DEFAULT
Modular monolith. One deployable, multiple bounded contexts. Easier to operate, easier to evolve, simpler debugging.
WHEN WE PICK DIFFERENTLY
Services when one component has dramatically different scaling needs (real-time chat, long-running jobs), or when separate teams need to ship independently.
03
Server-rendered, SPA, or hybrid?
OUR DEFAULT
Hybrid via Next.js or Remix. Server-rendered for marketing and SEO surfaces, client-side for app interaction. Best of both worlds.
WHEN WE PICK DIFFERENTLY
Pure SPA when SEO does not matter (internal apps, behind-login products). Pure server-rendered when JS in the browser is a non-starter.
04
Which database?
OUR DEFAULT
Postgres for almost everything. Relational, mature, scales further than most teams need, generous feature set.
WHEN WE PICK DIFFERENTLY
MongoDB or DynamoDB for genuinely document-shaped data. Redis when caching or queues dominate. Read replicas, partitioning, or specialized data stores when scale demands them.
05
Auth: build or buy?
OUR DEFAULT
Buy for most use cases. Clerk, Auth0, WorkOS, or AWS Cognito depending on requirements. Auth is hard to build well and easy to get wrong.
WHEN WE PICK DIFFERENTLY
Build when requirements are unusual (custom flows, regulated industries, deep enterprise SSO needs) or when buying creates an unacceptable dependency.
06
Where to host?
OUR DEFAULT
Vercel for the frontend and Railway, Fly, or Render for the backend in most cases. Move to AWS or GCP when scale or compliance demands it.
WHEN WE PICK DIFFERENTLY
AWS, GCP, or Azure when enterprise security, multi-region, or specialized compute is needed from day one. Self-hosted when data residency requires it.
REAL PROJECTS

Four web products shipped to production.

Anonymized for client confidentiality. The shape of the work and the outcomes are real.

B2B SAAS

An MVP for a vertical SaaS product.

A founder with a strong domain background wanted to ship a focused SaaS product for a specific niche. Pre-product but post-customer-research. We built the MVP in fourteen weeks, covering auth, billing, the core workflow, and three integrations. Shipped to first paying customers without further engineering work beyond minor tuning.

STARTING POINT
Validated idea, customer interviews complete, no code or design.
WHAT WE BUILT
STACK
Next.jsPostgresClerkStripeVercel
KEY DECISIONS
Modular monolith for solo founder maintainability
Multi-tenant from day one
Buy auth and billing rather than build
OUTCOME
$8K MRR within 90 days of launch
Founder hired their first engineer four months later, took over development.
MARKETPLACE

A two-sided marketplace for a specialty service category.

A founder with industry connections wanted to launch a marketplace connecting two parties in a specific service category. We built the matching, transaction, and trust layers. Phase one shipped both sides of the marketplace with verified onboarding, transaction flow, dispute handling, and an admin console.

STARTING POINT
Concept, supply-side relationships, no software.
WHAT WE BUILT
STACK
Next.jsPostgresAuth0Stripe ConnectAWS
KEY DECISIONS
Stripe Connect for marketplace payments
Manual verification with admin tooling
Phased rollout by geographic region
OUTCOME
1,200 active providers, 8,500 transactions in year one
Marketplace operations are net positive within the first year of launch.
FINTECH

A customer portal for a financial services firm.

An established financial services firm with thousands of clients had been managing client relationships through email and a homegrown PHP portal from 2009. We rebuilt the customer portal from scratch with modern UX, secure document exchange, scheduled reporting, and integration with the firm's existing systems.

STARTING POINT
Legacy PHP portal, mature business operations, regulatory requirements.
WHAT WE BUILT
STACK
Next.jsNode.jsPostgresWorkOS SSOAWS
KEY DECISIONS
SOC 2-aware architecture from day one
Self-hosted for data residency
Phased migration alongside legacy system
OUTCOME
Customer-reported satisfaction up 47%
Legacy portal decommissioned eight months after new portal launch. Support tickets related to portal usability dropped 64%.
PLATFORM

A public REST API plus developer portal for a B2B SaaS.

A B2B SaaS company's customers kept asking 'is there an API?' The answer was technically yes (internal endpoints exposed by accident) but operationally no (no documentation, no auth, no rate limiting). We designed and built the public API as a real product.

STARTING POINT
Production SaaS with internal endpoints. Demand for public API.
WHAT WE BUILT
STACK
Node.jsPostgresRedisOpenAPI/SwaggerMintlify
KEY DECISIONS
REST with OpenAPI specification as source of truth
API keys with scopes plus OAuth for partner integrations
Versioning strategy committed to before launch
OUTCOME
11% increase in expansion revenue tied to API customers
API customers retain at a significantly higher rate than non-API customers. Two partner integrations launched in the six months following.
THE STACK

What we build with.

Defaults we reach for on most web app projects. We adapt when there is a real reason. The point is the product, not the stack.

FRONTEND
Frontend framework
React-based, server-rendered when SEO matters.
Next.jsReactTypeScriptTailwind CSSshadcn/ui
BACKEND
Backend services
TypeScript by default, Python or Go when the use case calls for it.
Node.jsTypeScripttRPCExpressPython (FastAPI)Go
DATABASE
Database and data layer
Postgres for almost everything. Specialized stores when there is a reason.
PostgresDrizzlePrismaRedisS3
AUTH AND BILLING
Identity, auth, and billing
Buy these unless you have a strong reason to build.
ClerkAuth0WorkOSStripeLemon Squeezy
HOSTING AND OPS
Hosting and operations
Vercel and managed services when they fit. Cloud platforms when they do not.
VercelRailwayAWSGCPCloudflareSentryPostHog
API DESIGN AND DEVELOPER EXPERIENCE

When the API is part of the product, it needs to be designed as one.

Most APIs we encounter were built as side effects of internal services. Customer-facing APIs need more thought: contract stability, developer experience, documentation, security. Below is what we build into every public API.

  1. 01
    API design and contract
    REST or GraphQL chosen for the use case. Resource modeling, naming conventions, error formats, status codes. The contract should make obvious sense to a developer who reads it cold.
    WHAT WE BUILD
    OpenAPI specification (REST) or schema (GraphQL) committed to before any code is written. Used to generate clients, documentation, and validation.
  2. 02
    Authentication and authorization
    API keys for server-to-server, OAuth for user-context, scoped permissions for fine-grained access. Built so that customers and partners can authenticate without you needing to support every request.
    WHAT WE BUILD
    API key management UI, OAuth flow with consent, scope-based permissions, audit log of every authenticated request.
  3. 03
    Rate limiting and quotas
    Per-customer rate limits, fair usage policies, tier-based quotas tied to billing. Prevents abuse, makes the API economics predictable, and lets you upsell when customers hit limits.
    WHAT WE BUILD
    Token bucket rate limiting per API key, header-based rate limit communication, customer-facing usage dashboards.
  4. 04
    Webhooks and event delivery
    Customers want to know when things change. Reliable webhook delivery with retry, signing, and replay protection. Otherwise customers poll your API every thirty seconds.
    WHAT WE BUILD
    Webhook delivery service with exponential backoff, HMAC signatures, dead-letter queue, customer-side replay support.
  5. 05
    Documentation and developer experience
    API docs that are accurate, current, and helpful. Generated from the spec, not maintained separately. Plus quickstart guides, code samples in popular languages, and a sandbox.
    WHAT WE BUILD
    Auto-generated docs (Mintlify, Redocly, or custom), interactive API explorer, copy-paste curl examples in every language.
  6. 06
    Versioning and backward compatibility
    APIs are contracts with your customers. Breaking changes destroy trust. A versioning strategy committed to before launch beats one figured out under pressure two years in.
    WHAT WE BUILD
    URL or header-based versioning, deprecation timelines built into responses, customer notifications for upcoming breaking changes.
HOW WE BUILD

Four phases over ten to thirty-two weeks.

Web app timelines vary wildly by scope. An MVP can ship in ten to sixteen weeks. A full rebuild runs sixteen to thirty-two. The phases stay roughly the same, the durations expand or compress.

01
1-2 WEEKS
Discovery and product strategy
Understand the user, the business model, the technical constraints, and the success metrics. For MVPs this includes scoping what is and is not in version one. For rebuilds, this includes data migration planning.
FOCUS
Product strategy, scope, architecture sketch.
02
2-4 WEEKS
Design and architecture
Visual design and product design happen alongside architecture decisions. Wireframes, high-fidelity designs, system architecture, data model, API contract if applicable. Reviewed before code is written.
FOCUS
Design system, technical architecture, contracts.
03
6-22 WEEKS
Build with weekly demos
Development in two-week sprints with weekly demos. Production deployments from day one (or as early as possible) so the deployment process gets exercised continuously. Real users invited in as soon as there is something to use.
FOCUS
Continuous shipping with regular user feedback.
04
1-4 WEEKS
Launch and handoff
Launch to general availability. Train your team (if you have one). Hand off documentation, infrastructure access, and runbooks. Optional retainer for ongoing work or transition to your in-house team.
FOCUS
Launch, training, ongoing support setup.

For ongoing product development engagements, the build phase becomes continuous and we operate as an embedded part of your team.

FAQ

Frequently asked questions.

What founders and product leaders ask before committing to a web app build. Click to expand any answer.

READY TO SHIP A REAL PRODUCT

Bring us the product idea or the codebase you have outgrown. We will build it like real software.

Forty-five minutes. Walk us through what you are trying to build, who it is for, and where you are starting from. We will tell you whether the project fits this service, what a build looks like, and what timeline and cost to expect. Sometimes the right answer is to use off-the-shelf for the first six months, and we say so.

Book a product scoping call

No pressure. Just value.

Arius Assistant

Hi, I'm Ari 👋

I can help you automate tasks and answer questions about your business.