SOFTWARE AGENCY - COLOGNE
E-Commerce Development for the Age of Agentic Commerce
Spun off our own AI commerce product (Mandelbaum.ai) · Partner to igus since 2017 · Full-service from Cologne, Germany
VALUE PROPOSITION
Why shop architecture now decides who gets found
AI-generated answers now appear in a substantial share of search queries. Customers ask ChatGPT and Perplexity for product recommendations before they ever open a store. And AI agents don't look at your homepage; they read structured data and APIs.
Older shop systems, where product data has grown for years without ever being cleaned up, struggle in this environment. They get misread, or they don't get found at all. Not because the products are worse, but because the technology behind them was built for humans, not for machines that shop on behalf of humans.
We build e-commerce platforms that work for both.
WHAT WE BUILD
Online stores that do more than sell
Shop development
Fast, scalable platforms that are still maintainable five years from now.
02 Service
AI search for online stores
Semantic search that understands what customers mean, not just what they type.
API-first architecture
Clean interfaces for AI agents, PIM, ERP, and CRM. The foundation for everything else.
04 Service
Composable Commerce
Modular systems where new commerce interfaces plug in without touching the core.
Product data machines can read
Structured, consistent product data is the ticket into agentic commerce. We build the data model for it.
Conversion optimization
Based on data, not gut feeling.
System integration
Connecting existing PIM, ERP, and CRM landscapes, including the messy, grown ones.
PROCEDURE & METHODOLOGY
How an e-commerce project runs at dynabase
Commerce audit
We look at what's there: architecture, data quality, interfaces. An honest assessment, not a sales pitch.
Step 2
Architecture decision
Replatform, migrate, or rebuild selectively? We recommend what makes economic sense, not what makes the biggest project.
Step 3
Agile development
Short sprints, measurable milestones, full cost transparency.
Step 4
Integration & testing
System connections, quality assurance, load tests. Before launch, not after.
Step 5
Go-live & operations
Supported rollout with monitoring until everything runs stable.
Continuous development
We add new AI interfaces and commerce protocols once they're production-ready. Not sooner, not later.
TECHNOLOGY
The Technology Stack Behind Future-Ready Online Stores
We build on open architectures instead of closed shop systems. The reason is simple: closed systems decide for themselves which interfaces they support. With open architectures, you decide.
Headless Commerce
Frontend and backend separated, so new interfaces don't mean a rebuild
GraphQL & REST APIs
Standardized interfaces for AI agents, marketplaces, and internal systems
Semantic search & LLMs
Product search that understands natural language
Kubernetes & cloud
Reliable operations, even when Black Friday gets serious
GDPR-compliant architecture
Data protection built in from the start, with fully European hosting available on request
INDUSTRIES & USE CASES
Who we build stores for
1. Classic online stores
Scalable shop platforms with clean system integration, well-thought-out UX, and stable performance. Without any AI features, if that's the right solution.
2. B2B & industry
Configurators and product advisory systems that understand technical specifications and industry language.
3. Retailers with large catalogs
The more products, the bigger the leverage of an AI search that actually understands customers.
4. Stores with complex product data
Technology, spare parts, accessories: anywhere customers can't quite name the product they need.
CASE STUDIES
E-commerce and AI search in practice
Mandelbaum.ai – AI search for online stores
Semantic search that lets customers find products in natural language, without knowing the product name. Developed at dynabase, now an independent SaaS company.
E-commerce | AI | Search
RBTX Machine Planner – B2B configurator for industrial products
A configurator that brings technical consultation and purchasing together for industrial automation.
E-Commerce | Configurator
WHY DYNABASE
Agentic commerce from practice, not from a whitepaper
A lot of people are talking about agentic commerce right now. We built our own AI commerce product, Mandelbaum.ai, which now operates as an independent company. That experience — from model selection to data architecture to running under real load — goes into every client project.
As an e-commerce development agency based in Cologne, Germany, we work with you from the architecture decision to day-to-day operations. And if a small rebuild is all you need, we'll tell you that too.
Do you have any other questions? Ask us directly during your free initial consultation.
What is agentic commerce?
Agentic commerce is online retail where AI agents take over parts of the buying process: finding products, comparing options, making recommendations, and increasingly, completing purchases. For store owners, that means one thing: the technology has to be built so machines can read and understand it.
Do I need to replace my existing store entirely?
No, often not. In many cases, a targeted rebuild of the data layer and interfaces is enough. Whether replatforming or rebuilding makes more sense is something we clarify in a commerce audit, before any proposal gets written.
What does an agentic-commerce-ready store cost?
New builds start in the mid five-figure range; rebuilds of existing systems often come in below that. The exact cost depends on your current system, catalog complexity, and the integrations involved. You'll get a first estimate in a free consultation.
How do I make my store visible to AI agents?
Three things decide it: structured, complete product data; open APIs instead of closed systems; and machine-readable markup of your content. That's exactly the foundation we build.
Is all of this GDPR-compliant?
Yes. We build data protection into the architecture from day one and can host entirely within Europe. Which setup fits your case depends on your requirements and is defined before the project starts.