Building APIs for Big Data Services

Data volumes are increasing at an unprecedented rate, exploding from terabytes to petabytes and sometimes exabytes of data. Traditional API approaches are laid out for transactional data and can’t handle these data volumes because they don’t scale well enough and are too expensive.

In this session, we will first look at requirements of big data applications and the boundary conditions of building a data service. Then will discuss how to build a data lake including data ingestion and ETL. And, mainly, we will see different patterns of how to build APIs for such services that can be used by internal and external consumers, including aspects such as semantics, security, cost and performance.

Embeddable Agentic SaaS with A2UI and Microfrontends

Today’s interfaces often expose agents as simple chats or side panels, while real products are organized around clear domains and responsibilities. This session explores how to use A2UI to deliver complete agentic experiences as microfrontends, each aligned to a specific product context and owned by a dedicated team. Instead of sprinkling agent calls across many components, an agentic microfrontend becomes the primary interface for a given task space, with server-side agents describing UI updates via A2UI and the microfrontend rendering them using its own trusted design system.

The talk walks through the core patterns for this approach: structuring an agentic microfrontend around a domain boundary; using A2UI as the contract between agents and UI; handling streaming updates and state transitions over time; and integrating with existing navigation, auth, and observability. Particular emphasis is placed on production concerns: how to test agent-driven flows, how to constrain and validate A2UI payloads, how to instrument interactions end-to-end, and how to keep agent behavior evolvable without breaking the surrounding product.

Attendees gain clear mental models for where agentic microfrontends make sense, practical guidance on designing the A2UI contract and lifecycle, and concrete patterns they can apply when introducing agent-driven experiences into complex applications and organizations.

Interface Contracts in Microfrontend Architectures

Microfrontends are a popular approach to building distributed web architectures, and there’s no shortage of frameworks and strategies. But these common approaches require all-in commitment to specific frameworks, which makes a smooth migration of legacy platforms very difficult. In this talk, we will take a look at a frameworkless approach based on native browser technology such as custom elements and the CustomEvent API, and we will learn how this can be used to gradually modernise a legacy web application. We will still use advanced tools for building and delivering our solutions, but we avoid any shared runtime dependencies, which allows for maximum portability of our components. In addition, we will see how we can ensure encapsulation, testability and evolvability through clearly defined interface contracts. As infrastructure, we can use containers or serverless cloud technology, deployed via CI/CD and Infrastructure as Code.

The TSAI protocol: An open industry standard for trust signals in agentic ecosystems

The rapid proliferation of AI agents in 2025-2026 has created a fundamental trust problem: platforms cannot distinguish legitimate agents from malicious actors, leading to blanket blocking or expensive bot detection, while agents cannot prove their legitimacy, resulting in unreliable access and legal disputes. Trust Signals for Agentic Interactions (TSAI) addresses this challenge through an open, standards-based protocol that enables agents to present verifiable credentials issued by independent Trust Authorities and verified by platforms. Built on W3C Verifiable Credentials and Decentralized Identifiers, TSAI provides cryptographically signed credentials containing multi- dimensional trust signals—including verified operator identity, behavioral reputation, economic stake, and authorized constraints—while preserving user privacy through zero personally identifiable information in credentials.

TSAI employs a four-tier trust model (T0-T3) that balances performance and security across diverse risk levels. Lower tiers (T0-T1) provide basic identity and reputation signals with offline verification under 5 milliseconds, serving 95% of agent traffic including browsing, search, and public API access. Higher tiers (T2-T3) add economic accountability through posted collateral and insurance, plus fine-grained authorization constraints, with challenge-response verification and real-time Trust Authority queries for high-value transactions and regulated operations. This tiered approach enables incremental adoption—starting with simple identity verification and progressively adding reputation, economic stake, and constraints as the ecosystem matures—while optimizing for the common case of low-risk, high-volume interactions.

The protocol operates through a three-party model where professional Trust Authorities evaluate agents and issue short-lived credentials, agents present these credentials when accessing services, and platforms verify authenticity and make risk-calibrated access decisions based on their own policies. TSAI explicitly signals rather than enforces: the protocol defines what trust signals exist and what they mean, but platforms retain full autonomy in interpreting signals and setting access policies. With governance transferring to the Linux Foundation and early adoption from major European retailers representing over €80 billion in annual gross merchandise value, TSAI provides the trust infrastructure necessary for agentic commerce to scale while maintaining accountability, transparency, and interoperability across the emerging agent ecosystem.

IMPORTANT: This is still a confidential cross-industry project under the leadership of AWS. Please do not publish on the website before discussing with me. Thanks!

Micro-SaaS: Turning Microservices into Scalable Products

Micro-SaaS is an emerging trend in the software industry, supported by an ongoing fragmentation of the software landscape and increased readiness of enterprises to purchase software as a service. Micro-SaaS offerings are coming from two angles: First, established software vendors who have mastered scalable distributed architectures and are now turning microservices and microfrontends into standalone offerings. Second, startups or individual developers with a highly specific business model centered around a single capability. In this talk, participants will learn about architectural and engineering patterns that are needed to turn microservices, APIs and microfrontends into full-fledged commercial products: Self-service onboarding, identity/access management, consumption metering, tenant isolation, invoicing and billing. All of this with scalability and evolvability in mind which are essential for successful SaaS.