Our Services

For a focused MVP we start at €15k. Multi-team enterprise projects start around €50k and scale based on scope, integrations and required SLAs. After a 15-minute exploratory call you'll get a concrete range based on your case, no commitment required.

For a focused MVP we typically land in 4-8 weeks. Enterprise multi-team projects run 3-6 months distributed across iterative releases, with weekly client demos.

You do. All code belongs to you and lives in your private Git repository. No vendor lock-in, no recurring license on the codebase, technical documentation included.

We provide operational documentation, training for your internal team and an optional maintenance contract. You can continue with us or hand off to your team with no hidden dependencies.

We specialize in Next.js 15+, React 19, TypeScript, Supabase, PostgreSQL. For backend and CLI we also use C#/.NET and Rust when it makes sense. For mobile we build native iOS and Android apps with React Native and Expo (see the dedicated Mobile App service).

GDPR-first by design: data encrypted at-rest and in-transit, Supabase Row Level Security, audit trail and automatic backups. For enterprise projects we run a dedicated security audit before go-live.

Yes. We integrate via REST/GraphQL APIs, webhooks, messaging queues or directly at the database layer when needed. We run a preliminary assessment to pick the most robust integration pattern.

No, when we use their enterprise APIs (no-training by default) or when we deploy open-source models locally via Ollama or vLLM. We document the data flow in writing before kickoff.

For a defined-scope POC (RAG chatbot over internal documents, single-tool agent) we start at €10k. For production implementations with multi-agent, integrations and MLOps we're at €25k and up, scaling with the number of tools and data sources. We provide a concrete quote after discovery.

RAG (Retrieval-Augmented Generation) lets the LLM answer based on your up-to-date documents instead of generic training. It's worth it when accuracy on internal knowledge is critical (support, legal, medical, technical domains).

Yes, we always architect with an abstraction layer that allows swapping OpenAI, Anthropic, Gemini or local Ollama models without rewriting business logic. Zero lock-in on the provider.

We define evaluation metrics together (relevance, factual accuracy, tone), build a golden test dataset and measure every release against the baseline. No 'it works well' without numbers.

Yes, via APIs, webhooks, n8n or custom MCP servers. Solid experience integrating with Supabase, HubSpot, Zendesk, transactional email and internal knowledge bases.

MCP (Model Context Protocol) is the open standard that allows Claude, Cursor and other AI tools to securely access internal tools and data. We build MCP servers that expose your internal APIs as tools callable by AI assistants.

SEO optimizes for classic search engines (Google, Bing). GEO (Generative Engine Optimization) optimizes for getting cited by ChatGPT, Perplexity, Claude and Google AI Overview — where users don't click links anymore but read the answer synthesized by the AI.

Starting at €1.5k/month for retainer with setup plus continuous optimization. Paid campaigns have separate media budget depending on your market (B2B tech typically CPL €30-150).

For high-competition queries 4-6 months for solid results, for long-tail and GEO we see signals within 8-12 weeks. We provide monthly reporting with real progress on keywords, impressions, CTR and conversions.

Yes, multi-channel: Google Ads, Meta, LinkedIn and TikTok. Full tracking setup (GTM, GA4, pixels, enhanced conversions), creative testing and weekly optimization to cut CPL and lift quality score.

Server-side attribution when possible (GA4 + backend events), UTM consistency, weekly reporting dashboard with CAC, LTV, payback period. Zero vanity dashboards, only revenue-tied metrics.

Provider setup (Resend, Brevo, Mailchimp), audience segmentation, welcome flow, multi-step nurturing, re-engagement, transactional email. Includes A/B testing on subject lines, send times and CTAs.

Yes, we architect with clean export and documented schema. Every tool choice is justified in writing, no hidden vendor lock-in.

For selected SaaS products we accept an equity stake + monthly retainer instead of the project-quote model. It's a partnership model: we win if the product wins. Evaluated case-by-case after discovery.

Typically 6-10 weeks for a focused MVP with auth, multi-tenancy, base Stripe billing and the first user workflow. Feature extensions and scaling land in subsequent iterative releases.

The client typically, with full operational documentation. Default stack: Vercel (frontend), Supabase (auth + DB + storage), Trigger.dev (background jobs), Stripe (billing). Self-hosting on AWS/GCP/on-premise is also an option.

Row Level Security on Supabase with per-tenant_id policies, per-tenant audit trail, per-tenant backups. Zero cross-tenant leak risk: policies are tested automatically in CI.

Stripe is the default (most mature for B2B SaaS), but we also integrate Paddle, Lemon Squeezy and local processors when needed. Full management of webhooks, dunning, proration, tax compliance (Stripe Tax).

Yes, ongoing maintenance and evolution retainer is part of the partnership model. Alternatively we transfer operational ownership to your in-house team with dedicated training.

Measured guided onboarding (time-to-first-value), smart transactional emails, usage dashboard and automatic alerts when a user is about to disengage. If there's no PMF, no feature compensates.

Both, from a single React Native/Expo codebase: the same code produces the app for the App Store and Google Play. You save time and cost versus two separate native builds, while keeping native performance and experience on every device.

For a focused mobile MVP we start at €18k. Full apps with backend, AI and integrations scale with scope. After a 15-minute call you get a concrete range for your case, no commitment.

Yes, it's our strength: conversational assistants, voice recognition (speech-to-text), ML/predictive models and personalized advice, on-device or via cloud. In Lottomatik.ai we integrated voice dictation and an AI engine directly into the app.

Yes. We manage builds (EAS), certificates, App Store and Google Play listings, the review process and OTA updates. You follow the progress, we get the app onto the stores.

Yes. Lottomatik.ai is an iOS+Android companion app built entirely by us, with an ML backend, voice AI, push notifications and authentication — from concept to the stores.

You do. All the code lands in your private repository, no vendor lock-in. You can continue with us or hand maintenance to your internal team, with documentation and training included.

To discuss strategies, alpha sources and existing infrastructure with the level of detail needed for honest scoping. It's a safeguard for both sides: you protect IP, we work on real information instead of generic assumptions.

You, 100%. Code, optimized parameters, signals, processed historical data — all yours, in your private repository, with no residual rights or royalties for us. Contractually documented.

Absolute rule: never use future data to compute past values. Point-in-time correct pipelines, walk-forward validation, rigorous out-of-sample tests, randomized Monte Carlo on gaps and slippage. Every assumption documented.

Experience across Interactive Brokers, Binance (spot + futures), Bybit, CQG, Rithmic, MetaTrader. REST/WebSocket and FIX connectivity for institutional brokers. Prime broker deploy evaluated case-by-case.

Per-strategy limits (max position, max exposure), kill-switch on intraday max drawdown, circuit breaker on latency degradation, automatic broker vs internal P&L reconciliation, multi-channel alerting (email, Telegram, webhook).

Multi-source: tick/bar from institutional brokers, Polygon, Binance archive, CCXT histories, premium vendors for equities. Documented cleaning pipeline (gap handling, split adjustment, timezone-aware, no survivorship bias).

Yes, with discipline: point-in-time feature engineering, walk-forward validation, probability calibration, rigorous backtest. ML is useful when it improves the baseline, not because it's ML — we decline projects where the client wants ML as marketing with no baseline.