The questions we often get.

A typical project takes between 6 and 20 weeks: startup MVPs 6-10 weeks, custom software 12-20 weeks, legacy modernization 16-32 weeks. The range depends on domain complexity, level of integration with existing systems, and how mature the requirements are at kickoff.

We deliver a realistic estimate within 5 business days of first contact, after a 60-90 minute discovery call. We work in weekly sprints with regular demos: if timelines shift mid-project, you know before it becomes a problem.

We work primarily remotely. For complex projects, regulated industries, or strategic kickoffs we're available on-site at the client. Operational headquarters in Fano, Marche.

Our experience: well-structured remote work (weekly sync calls + daily async communication) produces better results than full-time-in-office for almost any software project. But physical presence matters in key moments: kickoffs, milestone reviews, critical incident troubleshooting.

Weekly sprint cycles with 30-minute check-ins. Minor changes (requirement clarifications, UX tweaks) are absorbed without re-baselining. Significant scope changes are assessed together with an impact analysis on timeline and budget, leading to a new transparent agreement before proceeding.

We don't work on rigid waterfall contracts, but neither on the everything-is-always-negotiable model. The rule: small adjustments we absorb, significant scope creep is a new agreement. Never surprises on the invoice.

Typically 4-6 active projects at the same time, each with a dedicated team of 2-3 senior engineers and a single technical lead. Your project doesn't rotate between different people based on weekly availability.

Total weekly capacity is about 120 senior hours, distributed across a few well-tended projects instead of many thin slices. When a new project would risk degrading the active ones, we say so: better a month of waiting than a watered-down service.

A decision-making point of contact on your side who responds within 24-48 hours to scope questions, a weekly 30-60 minute call for demos, access to relevant systems (CRM, ERP, repositories) for integrations. In total: 2-4 hours per week of your contact's time.

If the contact can't sustain that load (e.g. an already-stressed founder, an IT manager juggling 10 vendors), we flag it in discovery. Better to delay kickoff by 2-3 weeks than ruin project momentum because answers take 10 days.

A small project (MVP, single integration, vertical AI agent) costs between 15,000 and 30,000 euros. A standard project (complete custom software, partial modernization) costs between 30,000 and 80,000 euros. A complex or multi-year project (enterprise system, extensive legacy migration) starts at 80,000 and often exceeds 250,000 euros.

Ranges are indicative for Italian SMEs. Variables that move the cost: integration with existing systems, compliance constraints (NIS2, GDPR, AgID), software criticality (an internal agent costs less than software that invoices customers). We deliver a precise estimate within 5 days of discovery.

Both. We prefer fixed price when scope is well-defined (we tell you the total turnkey cost and we absorb the risk). We work time-and-materials when the project is exploratory or evolution depends on continuous feedback (R&D, early-stage products).

In practice: 70% of projects end up fixed price, 30% time-and-materials with a weekly budget cap. The mode is decided together with the estimate, not imposed.

Included: design, development, testing, technical documentation, team training at kickoff, 30 days of post-delivery bug-fix warranty. Excluded: hosting and cloud infrastructure, third-party software licenses, post-warranty evolutionary maintenance (offered separately if desired).

Everything is written in the estimate, no surprise add-on costs. If during the project we identify an accessory cost (e.g. a required enterprise license), we flag it immediately and you decide whether to proceed.

Backend: TypeScript (Node.js), Python, Go depending on the use case. Frontend: React, Vue, Astro, Svelte. Database: PostgreSQL, MongoDB, Redis. Infrastructure: AWS, GCP, Vercel, Fly.io. We pick the best technology for the problem, not stack religion.

For legacy modernization we work with vintage stacks (AS/400 RPG, COBOL, .NET Framework, PHP) when interoperability is needed. The rule: the tool isn't the goal, it's the means.

Yes. We use Claude and GitHub Copilot as productivity tools: they accelerate boilerplate writing, refactoring, testing, documentation. Every generated line is reviewed and validated by a senior engineer before commit. AI is never on autopilot, and it's not used for security-critical code or custom cryptography.

The important distinction: AI makes us write code faster. It doesn't think for the client. Architecture, domain decisions, security trade-offs stay human.

OWASP Top 10 by default on every web project. Internal code audit pre-delivery, automated dependency scanning (Snyk, Dependabot). For regulated industries (healthcare, finance, public administration) we add security-by-design: threat modeling, principle of least privilege, encryption at-rest and in-transit, NIS2 and GDPR compliance by design.

Zero-day security doesn't exist. What we can guarantee is adherence to consolidated best practices, an audit trail of what was verified, and an incident response plan. For critical projects we offer optional external penetration testing.

Yes, before going into any project detail. We can work with your NDA template, or use our standard (bilateral 3-year NDA, jurisdiction Pesaro). Electronic signature via DocuSign or equivalent.

For projects with especially sensitive information (e.g. M&A tech due diligence) we can also sign enhanced NDAs with specific clauses. We've never started a project without an active NDA when the client requested one.

The client. All code written specifically for the project is the client's exclusive property. IP transfer is documented in contract and finalized at final payment settlement. The only exceptions: foundational open-source libraries (released under original MIT/Apache/BSD licenses) and generic Obsidian internal tools (not client-specific) which remain ours but are granted in perpetual use.

No vendor lock-in: delivered code is readable, documented, and can be maintained by any other competent team. If you decide to take the project elsewhere, we provide all the documentation needed for handoff without obstacles.

Weekly communication anticipates problems: delays are never a project-end surprise. Delays caused by our optimistic estimates or technical issues: absorbed, no extra cost to the client. Delays caused by scope creep or missing client input: transparent re-baseline with new timeline and costs agreed before proceeding.

The basic rule: the project has a time budget and a money budget, both tracked weekly. If either is at risk, you know within the week, not at missed delivery. Preventive transparency is the single most effective tool for not ending badly.

Both. Full delivery (most common case): we take charge of the project end-to-end from discovery to deploy. Staff augmentation: we integrate into your in-house team to bring specific expertise (e.g. an AI expert on a team that lacks one) for a defined period.

The mode is decided in discovery based on what you already have in-house. Staff aug works best if you have a tech lead orchestrating; full delivery works best if you prefer to delegate the entire technical responsibility. In both cases, knowledge transfer to the internal team is documented at the end of the project.

Yes. After the 30 days of warranty included in the price (bug-fix), we offer an optional evolutionary maintenance contract: continuous bug-fix, small product evolutions, proactive monitoring. Typical pricing: 10-15% of project cost on an annual basis, scalable based on request volume.

Most clients choose the evolutionary maintenance mode at least for the first year post-go-live; some then transition to an internal team (with documented handoff). No multi-year commitments: renewed annually with joint service-level evaluation.