ZTABS builds multi-agent AI systems for Amsterdam enterprises — autonomous agent orchestration, agentic workflows, and complex task decomposition. From Zuidas financial compliance to Port of Rotterdam logistics, we engineer agent systems that handle Dutch enterprise complexity.
Multi-agent orchestration is reshaping how Amsterdam's enterprise landscape operates — moving beyond single-model AI inference toward coordinated systems of specialised agents that autonomously execute complex, multi-step business processes. Amsterdam's unique position as continental Europe's English-language business hub, the Netherlands' fintech capital, and a global logistics gateway makes it a natural centre for multi-agent AI adoption. The city's core industries — financial services (concentrated along the Zuidas business district), logistics and supply chain (anchored by the Port of Rotterdam and Schiphol Airport), agriculture and food technology (the Netherlands is the world's second-largest agricultural exporter), and professional services — all involve complex, multi-step workflows that single AI models cannot handle end-to-end. A multi-agent system decomposes complex objectives into subtasks, assigns each to a specialised agent, manages coordination and state, and handles failures through retry, rerouting, or human escalation. For Amsterdam enterprises, the value proposition centres on workflows that consume disproportionate analyst hours: regulatory compliance across multiple EU jurisdictions, supply chain orchestration spanning global networks, and professional services delivery requiring coordination of domain expertise. Dutch enterprises require multi-agent systems that operate within the Netherlands' regulatory framework: DNB (De Nederlandsche Bank) requirements for financial institutions, AFM (Autoriteit Financiële Markten) for market conduct, EU GDPR as implemented through the Dutch UAVG (Uitvoeringswet AVG), the EU AI Act (directly applicable), Dutch Wwft (Wet ter voorkoming van witwassen en financieren van terrorisme — the Dutch AML act), and sector-specific regulations for healthcare (NEN 7510), agriculture (NVWA requirements), and logistics (customs — EU UCC).
Amsterdam's multi-agent orchestration demand concentrates in the Netherlands' dominant industries. Financial services and fintech: the Zuidas district houses major Dutch banks (ING, ABN AMRO, Rabobank — all headquartered in Amsterdam or nearby), insurers (Aegon, NN Group, ASR Nederland), pension funds (ABP, PFZW — among the world's largest), and a thriving fintech ecosystem (Adyen, Mollie, Bunq, Backbase). Multi-agent systems for Dutch financial compliance: Wwft compliance orchestration — the Dutch AML act requires financial institutions to perform customer due diligence, transaction monitoring, and suspicious activity reporting to FIU-Nederland. A multi-agent Wwft system: a customer due diligence agent (KYC verification, UBO — ultimate beneficial owner — identification through KvK — Kamer van Koophandel — registry, sanctions screening against EU and Dutch sanctions lists), a transaction monitoring agent (analysing patterns against FIU-Nederland typologies — which differ from US FinCEN typologies in their focus on trade-based laundering and real estate transactions), an unusual transaction reporting agent (generating ongebruikelijke transactie meldingen in FIU-Nederland's prescribed format), and a PSD2/PSD3 compliance agent (monitoring open banking transactions for fraud patterns). Logistics and supply chain: the Netherlands' logistics infrastructure (Port of Rotterdam — Europe's largest, Schiphol Airport — one of Europe's busiest cargo hubs, plus extensive inland waterway and rail networks) processes millions of shipments annually. Multi-agent logistics orchestration: a customs agent (managing EU customs declarations under the Union Customs Code — determining tariff classifications, calculating duties, preparing import/export declarations through the Dutch Douane ATLAS system), a routing agent (optimising multimodal transport routes — sea, barge, rail, road — considering cost, time, emissions, and capacity), a documentation agent (generating bills of lading, packing lists, certificates of origin, and phytosanitary certificates for agricultural products), and a compliance agent (ensuring shipments comply with trade sanctions, export controls, and commodity-specific regulations). Agriculture and food technology: Dutch agrifood companies (Unilever, FrieslandCampina, Royal Cosun, plus thousands of agricultural enterprises) manage complex supply chains with strict quality and regulatory requirements. Multi-agent agrifood quality management: a supplier monitoring agent (tracking supplier certifications — GlobalGAP, BRC, IFS — and audit schedules), a quality testing agent (interpreting lab results against NVWA safety standards and EU Maximum Residue Levels), a traceability agent (maintaining lot-level traceability from farm to consumer — EU General Food Law requirements), and a recall management agent (coordinating product recall workflows when quality thresholds are breached). Professional services and legal: Amsterdam houses major international law firms (De Brauw Blackstone Westbroek, NautaDutilh, Loyens & Loeff) and the Netherlands' growing tech-law intersection (GDPR enforcement cases, EU AI Act compliance). Multi-agent legal orchestration: contract analysis agents specialised in Dutch law (BW — Burgerlijk Wetboek), EU regulatory frameworks, and international commercial terms.
Our multi-agent orchestration methodology for Amsterdam enterprises follows an architecture-first approach designed for the EU regulatory landscape and the Netherlands' multilingual, internationally-oriented business environment. Discovery and architecture (weeks 1-3): detailed workflow mapping including all regulatory touchpoints, system integrations, and jurisdictional requirements. Agent architecture blueprint: which agents exist, their individual charters, model selections, tool configurations, and typed communication schemas. Core principles: single-responsibility agents, typed inter-agent communication via structured messages, and centralised state management by the orchestrator. Technology stack: LangGraph for complex orchestrations, custom TypeScript for simpler pipelines. LLM selection: Claude Haiku or GPT-4.1 mini for classification ($0.25-$0.80/M tokens), Claude Sonnet 4.6 for core reasoning ($3/M tokens), Claude Opus 4.7 for quality review ($5/M tokens). Dutch-language capability: modern LLMs handle Dutch well, but we validate performance on domain-specific Dutch terminology — financial (Wwft, UBO, KvK), legal (BW, Rv), logistics (douane, accijnzen), and agricultural (keuringsdienst, residunormen) vocabulary. Tool integration through MCP servers: KvK (Chamber of Commerce) API for company data and UBO verification, Douane ATLAS for customs declarations, FIU-Nederland reporting interface, DNB regulatory filing systems, and enterprise systems (SAP, Oracle, custom platforms). EU AI Act compliance is central to our Amsterdam deployments: risk classification assessment (determining whether the multi-agent system qualifies as high-risk AI under the EU AI Act — most financial services and healthcare applications do), conformity assessment documentation, human oversight mechanisms (human-in-the-loop gates at decision points), and transparency requirements (users informed when interacting with AI). GDPR/UAVG compliance: data minimisation, purpose limitation, lawful basis documentation, DPIA (Data Protection Impact Assessment) for high-risk processing, and EU-hosted infrastructure (AWS eu-west-1 Ireland or eu-central-1 Frankfurt, Azure West Europe Netherlands). Development (weeks 3-8): iterative agent development with comprehensive testing. Deployment: shadow, assisted, and autonomous modes with stakeholder sign-off at each transition.
Logistics / International Trade
A Dutch logistics company (headquartered in Amsterdam, operating 12 warehouses across the Netherlands, Belgium, and Germany, processing 180,000 shipments monthly, EUR 340M revenue) managed international trade compliance through a combination of manual processes and legacy customs software. The problems: (1) Customs declaration preparation consumed 65% of the trade compliance team's time. Each import/export declaration required: tariff classification (determining the correct HS/CN code from 15,000+ commodity codes), duty calculation (applying the correct duty rate, preferential trade agreements, anti-dumping duties, and VAT), document preparation (commercial invoices, packing lists, certificates of origin, EUR.1 movement certificates for preferential trade), and submission through the Douane ATLAS system. The compliance team of 14 specialists processed an average of 4,200 customs declarations monthly — averaging 35 minutes per declaration. (2) Classification errors were costly. Incorrect tariff classification resulted in either overpaid duties (estimated EUR 380K annually in excess duty payments discovered during post-clearance audits) or underpaid duties (creating liability for back-duties, interest, and potential fines — the Douane had assessed EUR 210K in back-duties in the previous 2 years). (3) Trade agreement utilisation was suboptimal. The EU maintains 40+ preferential trade agreements offering reduced or zero duty rates. The compliance team estimated they claimed preferential treatment for only 60-65% of eligible shipments — missing EUR 500K+ in annual duty savings because verifying origin requirements for each trade agreement was too time-consuming. (4) Sanctions and export control screening was manual and inconsistent. Each shipment required screening against EU sanctions lists, dual-use goods regulations (EU Regulation 2021/821), and country-specific export restrictions. Manual screening: 8-12 minutes per shipment, with occasional missed matches due to name variation and transliteration inconsistencies.
We built a 5-agent trade compliance orchestration system over 10 weeks. (1) Classification Agent: automatically classifying goods using product descriptions, technical specifications, and historical classification data. The agent: parsing product descriptions from commercial invoices and shipping documents, matching against the EU Combined Nomenclature (CN) — 15,000+ commodity codes, applying General Interpretive Rules (GIR) for ambiguous classifications, and providing a confidence score with the top 3 most likely classifications and reasoning for each. Using Claude Sonnet 4.6 fine-tuned on 50,000 historical classifications from the company's customs database. Accuracy: 91.3% exact match (correct to 8-digit CN level), 97.8% at 6-digit HS level. Low-confidence classifications (below 85%) flagged for specialist review. (2) Duty Calculation Agent: computing applicable duties based on classification, origin, and trade agreements. The agent: applying the EU Common Customs Tariff duty rate, checking eligibility for preferential trade agreements (EU-UK TCA, EU-Canada CETA, EU-Japan EPA, EU-Vietnam FTA — 40+ agreements), verifying origin requirements (rules of origin — sufficient processing, value-added thresholds, product-specific rules), calculating anti-dumping and countervailing duties where applicable, and computing import VAT (21% Dutch BTW, with margin scheme calculations where applicable). Trade agreement utilisation increased from 60-65% to 89% of eligible shipments — the agent systematically checking every agreement for every shipment. (3) Document Generation Agent: preparing customs documentation from shipment data. The agent: generating customs declarations in ATLAS-compatible format (import declarations — AGS aangifte, export declarations — ECS uitvoeraangifte), preparing certificates of origin and EUR.1 movement certificates with correct origin statements, generating transit documents (NCTS/T1/T2) for intra-EU and international transit, and compiling document packages for customs broker submission. (4) Sanctions and Export Control Agent: screening every shipment against: EU consolidated sanctions list (updated daily from the EU Official Journal), Dutch national sanctions (Sanctiewet implementation), dual-use goods regulation (EU Regulation 2021/821 — checking goods descriptions against Annex I control lists), and country-specific export restrictions (arms embargoes, sectoral sanctions). The agent using both exact name matching and fuzzy matching with phonetic algorithms to catch transliteration variations. Processing: 30-45 seconds per shipment (vs. 8-12 minutes manual). Zero false negatives in 6 months of operation (validated through monthly retrospective audits). (5) Audit and Compliance Agent: maintaining the audit trail required by Dutch customs law (AWR — Algemene wet inzake rijksbelastingen) and preparing for post-clearance audits. The agent: logging every classification decision with reasoning (supporting AEO — Authorised Economic Operator — audit requirements), tracking duty savings from trade agreement utilisation (documentation for origin verification requests), generating monthly compliance reports for management, and flagging potential compliance issues proactively (classification patterns that might attract customs scrutiny). The orchestrator: managing the pipeline from shipment data receipt through declaration submission, with a human gate for high-value shipments (over EUR 100K), novel product classifications, and sanctions screening matches.
Declaration processing time: from 35 minutes average to 6 minutes (AI processing + specialist review of flagged items) — 83% reduction. The compliance team of 14 processing the same 4,200 monthly declarations in 40% less time — the freed capacity handling growing shipment volumes and proactive compliance activities. Classification accuracy: 91.3% exact match at CN 8-digit level — exceeding the team's estimated 87% manual accuracy. Classification disputes with Dutch customs: reduced from 12 per quarter to 2 per quarter. Trade agreement utilisation: increased from 60-65% to 89% of eligible shipments — recovering EUR 780K in annual duty savings (previously unclaimed preferential treatment). Duty overpayment: reduced by EUR 290K annually through more accurate classification and consistent duty calculation. Sanctions screening: processing time reduced from 8-12 minutes to 30-45 seconds per shipment. Coverage: 100% of shipments screened (vs. estimated 85% previously due to time pressure). Zero sanctions violations in 6 months. AEO status: the company's Authorised Economic Operator certification renewal streamlined — the AI system's comprehensive audit trail exceeding customs' documentation expectations. Annual value: EUR 780K (recovered duty savings) + EUR 290K (reduced overpayment) + EUR 420K (compliance team productivity) + EUR 180K (sanctions risk reduction) = EUR 1.67M annual value. Development cost: EUR 185,000. Payback: under 7 weeks.
Design teams of specialized agents with defined roles, tools, and communication protocols.
Sequential, parallel, and conditional agent execution with dependency management and state passing.
Structured message passing, shared memory, and context management between agents in the system.
Supervisor agents that monitor quality, resolve conflicts, and ensure output meets requirements.
Real-time dashboards showing agent activity, step latency, token usage, and cost per workflow.
Automatic retries, fallback agents, graceful degradation, and human escalation for production reliability.
Amsterdam's multi-agent AI market includes major consulting firms (Deloitte, PwC, EY, KPMG — all with significant Dutch practices), Dutch IT services companies (Atos, Ordina, Capgemini Netherlands), and a growing AI startup ecosystem. The Big 4 offer AI strategy and implementation but typically deliver single-model solutions at premium rates (EUR 200-400/hour). Dutch IT services companies build enterprise systems but most lack the LLM engineering expertise required for production multi-agent orchestration. AI startups in the Amsterdam ecosystem tend to build products, not custom enterprise orchestration systems. The gap: production-grade multi-agent systems require expertise at the intersection of LLM engineering, distributed systems architecture, and deep domain knowledge of Dutch and EU regulatory frameworks. ZTABS provides this combination with demonstrated production deployments and competitive rates — our multi-agent experience spanning financial services, logistics, and enterprise operations within the EU regulatory landscape.
Multi-agent orchestration projects in Amsterdam typically range from EUR 55,000 for focused 2-3 agent systems to EUR 300,000+ for enterprise-grade platforms with 5+ agents, complex integrations (Douane ATLAS, KvK, DNB systems), and EU AI Act compliance documentation. Amsterdam AI development rates: EUR 120-200/hour for quality firms. Big 4 consulting: EUR 200-400/hour. ZTABS delivers at competitive rates with the advantage of production multi-agent experience — reducing development time versus teams learning orchestration patterns on your project. Every engagement includes architecture documentation, comprehensive test suites, monitoring configuration, and EU AI Act compliance documentation.
Verified reviews from real client engagements — sourced from our public testimonial archive and Clutch profile.
My experience is throughout positive. Communication, service, the short response times and the flawless execution of a challenging topic was absolutely great. ZTABS is definitely my first choice again.
Christian Neff
Bank Software Advisory · Bank Software Advisory
Fintech
Fantastic Agency! I couldn't fault them even if I tried. They always go above and beyond to meet your expectations and always produces quality work. Thank you ZTABS.
Stephanie Kal
CEO · Beauty Finder Australia
Marketplace
It has been great working with ZTABS. They bounce off the ideas along the way. Amazing Experience.
Joel Rowe
CEO · Drill Quoter
Marketplace
We don't just contract. We ship and operate our own software. 23 products in production.
Common questions about multi-agent orchestration for Amsterdam businesses
We design multi-agent systems with EU AI Act compliance from the architecture phase. For high-risk AI systems (most financial services and certain logistics applications), this includes: risk management system documentation (Article 9), data governance and quality measures (Article 10), technical documentation and record-keeping (Article 11-12), transparency and information provision (Article 13), human oversight mechanisms (Article 14), and accuracy, robustness, and cybersecurity measures (Article 15). We prepare conformity assessment documentation and support the Declaration of Conformity process required before deployment.
ZTABS is a remote-first multi-agent orchestration agency serving Houston businesses — including agent team architecture, workflow orchestration, agent communication protocols. We work with Energy & Oil/Gas, Healthcare & Biotech, Aerospace & Defense companies in Houston, TX via timezone-aligned engineers and async workflows; we do not have a local office, and we are explicit about that with every client.
Multi-Agent Orchestration in New York, NYZTABS is a remote-first multi-agent orchestration agency serving New York businesses — including agent team architecture, workflow orchestration, agent communication protocols. We work with Finance & Fintech, Media & Advertising, Fashion & Retail companies in New York, NY via timezone-aligned engineers and async workflows; we do not have a local office, and we are explicit about that with every client.
Multi-Agent Orchestration in Los Angeles, CAZTABS is a remote-first multi-agent orchestration agency serving Los Angeles businesses — including agent team architecture, workflow orchestration, agent communication protocols. We work with Entertainment & Media, E-commerce & DTC Brands, Gaming & AR/VR companies in Los Angeles, CA via timezone-aligned engineers and async workflows; we do not have a local office, and we are explicit about that with every client.
Web Development in Amsterdam, NetherlandsZTABS is a remote-first web development agency serving Amsterdam businesses — including full-stack development, progressive web apps, api development. We work with FinTech, Logistics Tech, AgriTech companies in Amsterdam, Netherlands via timezone-aligned engineers and async workflows; we do not have a local office, and we are explicit about that with every client.
Web Design in Amsterdam, NetherlandsZTABS is a remote-first web design agency serving Amsterdam businesses — including ui/ux design, responsive design, custom interfaces. We work with FinTech, Logistics Tech, AgriTech companies in Amsterdam, Netherlands via timezone-aligned engineers and async workflows; we do not have a local office, and we are explicit about that with every client.
AI Development in Amsterdam, NetherlandsZTABS is a remote-first AI development agency serving Amsterdam businesses — including llm integration & fine-tuning, ai agents & automation, rag & knowledge systems. We work with FinTech, Logistics Tech, AgriTech companies in Amsterdam, Netherlands via timezone-aligned engineers and async workflows; we do not have a local office, and we are explicit about that with every client.
Multi-Agent OrchestrationLearn more about our multi-agent orchestration services nationwide.
Partner with ZTABS for expert multi-agent orchestration in Amsterdam. Get a free consultation today.
