From Discovery to Maintenance: How Startup House Runs a Project Step by Step

The process in one view

Each stage ends with a concrete deliverable, and each stage is optional: if you already have research, you can start at Scope Building. If you have a scope, you can start at development. Here is what happens inside each step.

Step 0: from first call to project brief

The process starts with you sharing your idea or business challenge. We run a discovery workshop and prepare a detailed project brief, then a Software Requirements Specification (SRS) defining scope, features, and budget. From there we create a roadmap, assign specialists, and development can begin within weeks, with full transparency, demos, and regular updates throughout.

Step 1: Product Discovery (4 to 8 weeks)

Product Discovery is a structured research process that turns assumptions into evidence before a single line of production code is written. It answers the question: should we build this?

The six phases:

1. Research. Stakeholder interviews, market analysis, and competitor review, with AI-assisted analysis expanding coverage without extending the timeline.
2. Problem definition. Data synthesis, problem framing, and user personas, prioritized against business goals.
3. Ideation. Co-created solutions with your team, ranked by user value and business feasibility.
4. Prototyping. A low-fidelity prototype focused on user flow, with AI tools accelerating wireframing and iteration.
5. User testing. The prototype tested with real users, findings documented and prioritized.
6. Handover. A decision-ready package: MVP specification, cost estimate, and tech roadmap.

What you receive: a validated prototype, a prioritized MVP feature list, a development cost estimate, user personas, and tech recommendations. As Jonathan Goodman, founder of NegotiateNow, described it: "During Product Discovery, SH helped me to understand the lean start-up way of product development with initial research, discovery workshops and many hours of discussion and iteration."

Step 2: Scope Building (2 to 5 weeks)

Where Discovery answers "should we build this?", Scope Building answers "what exactly are we building, and what will it cost?" It is the bridge between knowing what to build and building it.

The process: structured workshops for vision alignment, feature prioritization by user value, business feasibility, and development cost, a technical feasibility review by senior engineers against your existing stack, high-fidelity interactive wireframes, targeted usability testing (where fixes cost hours, not sprints), and a final MVP definition.

What you receive: a finalized MVP feature specification, a development cost estimate, tested wireframes, a technical feasibility assessment, and development-ready documentation. Every prioritization decision is documented with its rationale, so stakeholders know what was deferred and why. The package stands on its own: if you build with another partner or in-house, nothing is missing.

Step 3: development (MVP: even 1 week to first user)

Development runs in four phases:

1. Scoping and research. Market validation, with AI mapping the competitive landscape and stress-testing core assumptions.
2. Product design. Wireframes and a clickable prototype, with AI generating multiple design variations in parallel, so you choose from several concepts instead of one.
3. User testing. Feedback gathered and synthesized, fixes applied to finalize designs.
4. Development and maintenance. The build itself, followed by feature refinement based on user feedback.

Work runs in agile sprints with demos and regular updates. Pivoting mid-build is planned for, not penalized: AI-accelerated workflows keep the cost of changing direction low until your data says otherwise. In practice, an MVP can go from kickoff to first user in even 1 week: we build on proven components and write custom code only where it matters, while enterprise scope extends the timeline.

Real examples from our case studies: Simon Care Management's healthcare MVP went through 7 user testing sessions with patients and caregivers. NegotiateNow's marketplace was built on a no-code stack to reach the market quickly. Doogie, our internal AI showcase, went from kickoff to working application in 2 months.

Step 4: Quality Engineering, embedded, not bolted on

QA at Startup House is Quality Engineering: testing embedded in the development process from the first sprint, not a gate at the end. Setting up a QE process for a product already in development typically takes 2 to 3 weeks.

The five steps: a quality audit and test strategy, test case design mapped to product requirements, manual and automated testing (manual for complex journeys and edge cases, automated for regression and performance), performance and security testing under real-world load, and release sign-off with CI/CD integration.

Automation uses Cypress or Playwright, converted gradually from validated manual test cases, with AI-assisted test writing to scale the practice. Documented test plans and sign-off reports live in your repository, so when auditors ask for QA evidence for your ISO 27001, SOC 2, or HIPAA requirements, it is already there.

Step 5: maintenance and evolution

Launch is the midpoint, not the finish line. Ongoing work covers quality assurance, cloud services with monitoring, and continued feature development. Most clients continue after the MVP: incorporating user feedback, adding features, and scaling. Cloud infrastructure runs with CI/CD pipelines and support tiers up to 24/7.

The long-term pattern shows in our client list: Siemens Financial Services since 2016, the cybersecurity platform since 2020, and Siemens Healthineers for more than 6 years. We build for the phase after launch, because that is where we usually still are, years later.

Where AI fits across the whole process

AI tooling runs through every stage: research analysis in Discovery, parallel design variations in prototyping, code generation and test writing in development, and feedback synthesis in testing. Governance is constant: human review on every commit, security scanning on every push, and AI usage documented in pull requests.

About Startup House

Startup House is a 50-person software development company based in Warsaw, Poland, established in 2016, ISO 27001 compliant, with 100+ products shipped for clients on 5 continents.

Want to see what this process looks like for your product? Book a 30-min call.

FAQ

How long does a full project take at Startup House?

Product Discovery runs 4 to 8 weeks, Scope Building 2 to 5 weeks, and an MVP can go from kickoff to first user in even 1 week. AI products like KnowHub can deploy in as little as 2 weeks once data is prepared.

Do I have to start with Product Discovery?

No. If you have well-documented user research and a defined value proposition, you can start directly at Scope Building. If you have a validated scope, you can start at development. Each stage's deliverables stand on their own.

What do I receive after Product Discovery?

A validated low-fidelity prototype, a prioritized MVP feature list, a development cost estimate, user persona documentation, and tech recommendations, everything needed to begin development with clarity.

How is testing handled during development?

Quality Engineering is embedded from the first sprint: QE engineers work inside the sprint cycle, automated tests (Cypress or Playwright) run on every build, and every release goes through documented sign-off. Initial QE setup takes 2 to 3 weeks.

What happens after launch?

Most clients continue into maintenance and evolution: user feedback incorporation, new features, QA, and cloud monitoring with support tiers up to 24/7. Startup House partnerships commonly run for years, like Siemens Financial Services since 2016.