Sign in

Steef-Jan Wiggers

@steefjan.bsky.social
456 followers 368 following 277 posts

InfoQ Cloud Lead Editor, Microsoft MVP Azure Integration PaaS, Cloud Native, Speaker, Scuba Diver, Runner, and Field Hockey Player.

PostsRepliesMedia
Reposted by Steef-Jan Wiggers
hev bot @hevbot.bsky.social · 29/09/2026
We built cost tracking into hev kit's trace archive: spend, tokens and cache hits by day and model at list rates, plus a heatmap of when you prompt. hev.dev/kit/?utm_source=bluesky&utm…
System diagram: on your Mac, coding-agent session files flow to hevd, which indexes through the layer gateway in Docker. The gateway archives to the hev-traces namespace in your turbopuffer account. The highlighted Docker dashboard searches through the gateway and shows spend, tokens and cache hits, grouped by day and model at list rates.
001
Steef-Jan Wiggers @steefjan.bsky.social · 29/09/2026
Bonus post: ran Microsoft's Cosmos DB Agent Kit against my own sample repo instead of trusting it blindly. Caught a real bug — mock embeddings never normalized to unit length, silently producing inconsistent cosine-similarity scores. sjwiggers.com/2026/09/29/c...
sjwiggers.com
Cosmos DB Agent Kit: Auditing My Own Sample Repo - Cloud Perspectives
Six posts of hand-checked rules, now checked by a rulebook. I ran the Cosmos DB Agent Kit against my own sample repo — here's what it caught.
000
Steef-Jan Wiggers @steefjan.bsky.social · 24/09/2026
Apple Reference Image signs pixel data inside the iPhone 18 Pro sensor, then develops it in Private Cloud Compute under Apple's signature. Trust moves from C2PA's editing chain to Apple. Verification stays on Apple platforms. www.infoq.com/news/2026/09...
infoq.com
Apple Reference Image Signs Photos at the Sensor, Moving Provenance Trust Away from C2PA
Apple has published the design of Reference Image, an iPhone 18 Pro camera mode that signs pixel data at the sensor and develops it in Private Cloud Compute under an Apple signature. Developers on Hac...
011
Steef-Jan Wiggers @steefjan.bsky.social · 24/09/2026
Built all six rate limiting algorithms in C# on Azure and load-tested them. A "10/sec" fixed window let 19 through, and a second replica quietly doubled every in-process limit. sjwiggers.com/2026/09/24/r...
sjwiggers.com
Rate Limiting Algorithms on Azure: 6 Compared - Cloud Perspectives
Six rate limiting algorithms on Azure, measured with real load tests: fixed window, sliding log, token bucket, leaky bucket, and more.
010
Steef-Jan Wiggers @steefjan.bsky.social · 22/09/2026
Azure's price list for Foundry models: 1,727 meters in one region, but it doesn't name the model. So I wrote a tool that joins the ARM catalog, Retail Prices, and your own deployments: price per 1M tokens, retirement date, and which of yours are affected. sjwiggers.com/2026/09/22/f...
sjwiggers.com
Foundry Model Ledger: Which Model, at What Price, Until When, and Which of Mine - Cloud Perspectives
Foundry Model Ledger shows what a Foundry model costs per 1M tokens in your currency, when it retires, where else it runs, and which of your deployments are affected.
010
Steef-Jan Wiggers @steefjan.bsky.social · 21/09/2026
AWS says it cannot restore data held exclusively in a damaged UAE availability zone, or exclusively in its Bahrain region, after strikes during the conflict with Iran. Multi-AZ and S3 durability are regional properties. www.infoq.com/news/2026/09...
infoq.com
AWS Cannot Restore Data Held Only in Damaged Middle East Availability Zones
AWS has told customers it cannot restore resources and data hosted exclusively in the mec1-az2 availability zone in the UAE, or exclusively in the Bahrain region, after damage during the conflict with...
000
Steef-Jan Wiggers @steefjan.bsky.social · 20/09/2026
Revoked the same user's session twice. One client was locked out in 4 seconds, the other kept working for 65 minutes. The difference: one claim in the token request. Six samples measuring what the Entra ID diagram hides: sjwiggers.com/2026/09/20/m...
sjwiggers.com
Microsoft Entra ID under the hood - Cloud Perspectives
Microsoft Entra ID under the hood: token lifetimes, CAE revocation latency and PIM payload, measured on a real tenant with six deployable samples.
000
Steef-Jan Wiggers @steefjan.bsky.social · 17/09/2026
Post 6, closing the series: real production case study on Cosmos DB agent memory cost — one partition key absorbed 80% of traffic, account pinned at 100% RU utilization. Fixing the data model dropped it to 20-35%, no new throughput. sjwiggers.com/2026/09/17/c...
sjwiggers.com
Cosmos DB Agent Memory Cost: Caching, RU Drivers, and a Pitfalls Roundup - Cloud Perspectives
Cost only matters next to the value it unlocks. A practical guide to Cosmos DB agent memory cost: semantic caching, RU drivers, and pitfalls.
100
Steef-Jan Wiggers @steefjan.bsky.social · 17/09/2026
GPT-6 Astra is the first model OpenAI classifies as Critical for cybersecurity under its Preparedness Framework. In expert-led testing it found unknown vulnerabilities in a browser and an OS kernel and built working exploits. www.infoq.com/news/2026/09...
infoq.com
GPT-6 Astra Is the First Model OpenAI Classifies as Critical for Cybersecurity
OpenAI has classified GPT-6 Astra at the Critical cybersecurity threshold under its Preparedness Framework, a first. In expert-led testing the model found previously unknown vulnerabilities in a brows...
000
Steef-Jan Wiggers @steefjan.bsky.social · 16/09/2026
Post 5 on Cosmos DB agent memory: bring-your-own thread storage for Foundry Agent Service — direct query access instead of a managed black box. Gotcha: 3,000 RU/s minimum, or you hit CapabilityHostProvisioningFailed. sjwiggers.com/2026/09/16/c...
sjwiggers.com
Cosmos DB Foundry Agent Service: Bring-Your-Own Thread Storage - Cloud Perspectives
Default agent storage is opaque. A practical guide to Cosmos DB Foundry Agent Service — the enterprise_memory database, RU sizing, and roles.
100
Steef-Jan Wiggers @steefjan.bsky.social · 13/09/2026
Built five RAG architectures on the same Azure setup, with the same 15 questions, so I could actually compare them. Best surprise: Corrective RAG did better on the 12 questions it wasn't built for than on the 3 it was built for. sjwiggers.com/2026/09/13/f... #azure #rag #ai
sjwiggers.com
Five RAG Architectures in Real Azure Code - Cloud Perspectives
Hybrid, GraphRAG, Agentic, Corrective, and Multimodal RAG, built against the same Azure substrate and the same fifteen questions. What the diagrams leave out, and what the measured numbers actually sh...
100
Steef-Jan Wiggers @steefjan.bsky.social · 11/09/2026
Settled the MCP vs API debate the practical way: one Azure Functions app, same backend, two doors. Developers call the REST endpoints; an AI agent discovers the MCP tools and chains them on its own. Code and writeup: dev.to/steefjan_wig...
dev.to
MCP vs API Is the Wrong Question
Recently, I saw a LinkedIn post discussing "MCP vs API" and felt like we needed to choose one. But I...
212
Steef-Jan Wiggers @steefjan.bsky.social · 11/09/2026
Built a scam-checking agent on Azure Functions hosted skills. V1 let the model orchestrate tools and hit 47K-token turns. V2: evidence in code, one LLM call for judgment. dev.to/steefjan_wig...
dev.to
Azure Functions Hosted Skills: I Built a Scam-Checking Agent, Then Removed the Agency
Is that webshop legit? Most of us have pasted a URL into a reputation checker at least once. The...
110
Steef-Jan Wiggers @steefjan.bsky.social · 10/09/2026
Post 4 on CosmosDB agent memory: multi-agent systems need shared memory to coordinate and private memory for each agent's own persona. Handoff is change feed — one agent's write becomes the next agent's signal to act, no polling loop. sjwiggers.com/2026/09/10/c...
sjwiggers.com
Multi-Agent State and Checkpointing with Cosmos DB - Cloud Perspectives
A practical look at Cosmos DB multi-agent state — LangGraph checkpointing, hierarchical partitioning per session, and change feed for agent handoffs.
020
Steef-Jan Wiggers @steefjan.bsky.social · 03/09/2026
Post 3 on Cosmos DB agent memory: same question, four ways — recency, semantic, hybrid, keyword — one container, four answers. Gotcha: named params don't bind in Data Explorer — VectorDistance() won't compile til you swap in literals. sjwiggers.com/2026/09/03/c...
sjwiggers.com
Finding the Right Memory: Vector, Full-Text, and Hybrid Search in Cosmos DB - Cloud Perspectives
Storing memory well isn't the same as retrieving it well. A practical guide to Cosmos DB agent memory search: vector, full-text, and hybrid queries.
010
Steef-Jan Wiggers @steefjan.bsky.social · 02/09/2026
Drew the Microsoft AI stack the way I see it: six layers, the sixth being governance and evidence, which vendor diagrams skip. Then plotted the AI-901 to AI-500 cert trail on it. AI-500 tests guardrails, not models. sjwiggers.com/2026/09/02/m...
sjwiggers.com
The Microsoft AI Stack in 2026 and the Certification Trail That Runs Through It - Cloud Perspectives
The Microsoft AI stack as I see it: six layers, one governance plane, and how the new AI certification trail up to AI-500 maps onto that picture.
120
Steef-Jan Wiggers @steefjan.bsky.social · 02/09/2026
Cloudflare added optional OAuth scopes: client owners mark which permissions users may deselect at consent. MCP servers are the motivating case. Partial consent isn't new. What's new is developer control over what's droppable. www.infoq.com/news/2026/09...
infoq.com
Cloudflare Adds Optional OAuth Scopes, Letting Developers Mark What Users May Decline
Cloudflare has added optional OAuth scopes, letting client owners mark which permissions users may deselect at consent. The company names MCP servers as the motivating case, since agents request the u...
011
Steef-Jan Wiggers @steefjan.bsky.social · 01/09/2026
Partition key and item shape decide whether a Cosmos DB agent memory schema holds up. TTL trick: -1 on the container, override per item. Short-term turns expire; long-term memories don't. Same container. sjwiggers.com/2026/09/01/c... #agenticai
sjwiggers.com
Designing a Cosmos DB Agent Memory Schema - Cloud Perspectives
Before you write agent code, design the schema. A practical guide to Cosmos DB agent memory schema design: partition keys, data models, and TTL.
011
Steef-Jan Wiggers @steefjan.bsky.social · 31/08/2026
Agent ≠ Model. Foundry's router picks the model per request, and the pool refreshes under a stable endpoint. The durable investment is the control plane, not today's model. Also in the post: where that's the wrong answer. sjwiggers.com/2026/08/31/b...
sjwiggers.com
Don't Build Around Today's Model. Build for the AI Control Plane - Cloud Perspectives
Microsoft Foundry's model router updates point to a model-agnostic AI control plane. Why enterprise architects should decouple agents from models.
011
Steef-Jan Wiggers @steefjan.bsky.social · 31/08/2026
Cloudflare Workers can now accept inbound TCP. A connect(socket) handler routes raw sockets to a Worker, Durable Object, or Container. gRPC is first on top: full duplex in Containers, unary and server-streaming in Workers. Private beta. www.infoq.com/news/2026/08...
infoq.com
Cloudflare Workers Accept Inbound TCP, with gRPC the First Protocol on Top
Cloudflare Workers can now accept inbound TCP connections through a new connect(socket) handler routed via Spectrum, ending an eight-year restriction to HTTP. Containers get full-duplex gRPC in any la...
111
Steef-Jan Wiggers @steefjan.bsky.social · 30/08/2026
In 2017, someone called my Cosmos DB demo "an hour-long marketing pitch," too pricey at €1,000/month. Nine years on, it's replacing the cache + SQL + vector stack most AI agents use for memory. Post 1 of 6. sjwiggers.com/2026/08/30/w...
sjwiggers.com
Why Cosmos DB Ends Up as the Agent Memory Database - Cloud Perspectives
A 2017 Cosmos DB talk was dismissed as an expensive pitch. Here's why Cosmos DB agent memory now beats a stitched cache, SQL, and vector stack.
000
Steef-Jan Wiggers @steefjan.bsky.social · 28/08/2026
Cloudflare Wallets gives agents three spending controls: an allowance, a merchant allow-list, a max transaction size. Each checks one payment against a limit. None expresses a relationship between them. Only handle claiming ships today. www.infoq.com/news/2026/08...
infoq.com
Cloudflare Wallets Arrives Late to x402, and the Spending Controls Stop at the Payment
Cloudflare announced Wallets, giving agents a stablecoin balance and spending controls, though only handle claiming is live and squatting complaints have already surfaced. Payments run on x402, now ho...
121
Steef-Jan Wiggers @steefjan.bsky.social · 27/08/2026
Identity secures who the agent is. It doesn't touch what a poisoned tool result makes the agent do :injection rides in as data the agent asked for, inside the loop, past the perimeter. Series finale: sjwiggers.com/2026/08/27/w...
sjwiggers.com
When Agentic Workloads Break the PaaS Assumptions - Cloud Perspectives
Explore how agentic workloads challenge traditional assumptions in system integrations and transform operational dynamics.
000
Steef-Jan Wiggers @steefjan.bsky.social · 26/08/2026
Every diagram of Functions behind API Management draws one arrow into the function. There is always a second one, and it skips every policy you wrote. Six gaps the happy path leaves out, plus an azd sample that proves it. sjwiggers.com/2026/08/26/a...
sjwiggers.com
Azure Functions Behind API Management: What the Happy Path Diagram Leaves Out - Cloud Perspectives
The Client to APIM to Function diagram is right and incomplete. Six gaps that decide whether your Azure API design survives a security review, plus an azd sample.
000
Steef-Jan Wiggers @steefjan.bsky.social · 25/08/2026
"Is it built?" and "Can we operate it?" are different questions. A platform can pass the first comfortably and fail the second completely, and the second one decides go-live. On production readiness and the execution gap: sjwiggers.com/2026/08/25/f...
sjwiggers.com
Production Readiness: Closing the Execution Gap - Cloud Perspectives
Explore the concept of production readiness and understand what it truly means for your integration platform.
121
Steef-Jan Wiggers @steefjan.bsky.social · 21/08/2026
Power Automate isn't actually an Azure service. It's licensed through Microsoft 365/Power Platform, not Azure, a different bill and admin center than Functions or Logic Apps. Wrote up all three: sjwiggers.com/2026/08/21/a...
sjwiggers.com
Azure Functions vs. Logic Apps vs. Power Automate: When to Use What - Cloud Perspectives
A practitioner's comparison of Azure Functions, Logic Apps, and Power Automate, covering integration realities, the Integration Account cost most guides skip, and why Power Automate is not actually Az...
000
Steef-Jan Wiggers @steefjan.bsky.social · 20/08/2026
A monitoring standard that depends on discipline isn't a platform capability; it's a hope. Observability and FinOps as one discipline in two currencies: what is the platform doing, and what is it costing? sjwiggers.com/2026/08/20/a...
sjwiggers.com
Azure Observability and FinOps for Integration Architects - Cloud Perspectives
Explore azure observability and its crucial role in managing production environments effectively with visibility in costs and performance.
000
Steef-Jan Wiggers @steefjan.bsky.social · 19/08/2026
RAG retrieval quality > infrastructure quality. Keyword search for Metallica returns mostly Radiohead. Hybrid search fixes it: 20/20 Metallica, zero noise. Final post in the Azure Functions AI series. sjwiggers.com/2026/08/19/r...
sjwiggers.com
Building RAG Pipelines with Azure Functions: Event-Driven Data Retrieval at Scale - Cloud Perspectives
Azure Functions handles event-driven data retrieval for RAG pipelines, parallel ingestion, the Azure OpenAI binding extension, and APIM as the LLM gateway. This post maps the pattern and the productio...
110
Steef-Jan Wiggers @steefjan.bsky.social · 18/08/2026
Not all AI orchestration should be autonomous. Durable Functions for when the steps are known: function chaining, fan-out/fan-in, human-in-the-loop, monitor, risk-class routing. Five C# patterns verified on Azure. sjwiggers.com/2026/08/18/d...
sjwiggers.com
Durable Functions as the Orchestration Layer for Directed Agentic Workflows - Cloud Perspectives
Durable Functions is the right orchestration layer for directed agentic workflows — predictable steps, fault tolerance, human-in-the-loop, and long-running execution. This post maps the patterns.
000
Steef-Jan Wiggers @steefjan.bsky.social · 15/08/2026
I built nine agentic AI design patterns for real on Azure, not just as diagrams. Cosmos DB has its own separate RBAC system that silently grants nothing, and "Azure AI Foundry" is secretly two incompatible resource types. More: sjwiggers.com/2026/08/15/a...
sjwiggers.com
Agentic AI Design Patterns on Azure - Cloud Perspectives
What actually breaks when building agentic AI on Azure PaaS: real errors and fixes from deploying 9 agentic AI design patterns end to end, not just the diagrams.
010
Steef-Jan Wiggers @steefjan.bsky.social · 14/08/2026
Expedia open-sourced mockql-rs: LLM-generated GraphQL mocks at request time, merged with live backend data in one response. Airbnb did it at build time in April. The GraphQL RFC says no network call. Three architectures, two use @mock. www.infoq.com/news/2026/08...
infoq.com
LLM-Generated GraphQL Mocks Arrive at Airbnb and Expedia, While the Spec Lags Behind
Expedia Group has open-sourced mockql-rs, a Rust CLI that fills @mock-annotated GraphQL fields with LLM-generated data at request time. It follows Airbnb's @generateMock in April and a GraphQL Foundat...
000
Steef-Jan Wiggers @steefjan.bsky.social · 14/08/2026
MCP dropped protocol sessions. The under-covered part: Mcp-Method and Mcp-Name are now required headers, so a gateway or WAF can govern agent tool calls without parsing the JSON body. HN split on whether that just makes it REST again. www.infoq.com/news/2026/08...
infoq.com
MCP Goes Stateless, and Developers Ask Whether That Just Makes It an API Again
The MCP 2026-07-28 specification removes the initialize handshake and session header, and adds required method and tool-name headers so gateways can route agent traffic without parsing JSON. Reaction ...
011
Steef-Jan Wiggers @steefjan.bsky.social · 13/08/2026
Three files replace an agent loop on Azure Functions. .agent.md, agents.config.yaml, mcp.json — the runtime handles registration, wiring, and the loop. New post: how it works and when to use it over Foundry Agent Service. sjwiggers.com/2026/08/13/a...
sjwiggers.com
The Azure Functions Serverless Agents Runtime Explained - Cloud Perspectives
The Azure Functions serverless agents runtime lets you build event-triggered agents as function apps. When to use it over Foundry Agent Service.
121
Steef-Jan Wiggers @steefjan.bsky.social · 12/08/2026
A RAG diagram promised "zero hallucination". I built it and measured it. Keyword search made retrieval worse: 7/11 vector-only, 6/11 hybrid, 11/11 with the reranker. Confident diagrams are cheap to draw. Building one shows what's real. sjwiggers.com/2026/08/12/n...
sjwiggers.com
The Four Things Naive RAG Diagrams Leave Out - Cloud Perspectives
Naive RAG diagrams leave out four things that decide whether your pipeline works. Measured results on Azure AI Search, with a runnable sample repo.
000
Steef-Jan Wiggers @steefjan.bsky.social · 12/08/2026
citadel-v1 of the AI Hub Gateway Solution Accelerator is out. New integrations should target /unified-ai/v1/*. Access contracts ship at 1.2.0; multi-provider routing is first-class. Wrote up what changed and what to do about it: sjwiggers.com/2026/08/12/c... #azure #apimanagement #agenticai
sjwiggers.com
Citadel Grew Up: What the citadel-v1 Release Means If You Built on the AI Hub Gateway. - Cloud Perspectives
The citadel-v1 AI Hub Gateway release reframes the accelerator as Layer 1 of Microsoft's AI Citadel Blueprint. Here's what changed and what to migrate.
010
Steef-Jan Wiggers @steefjan.bsky.social · 11/08/2026
Identity without governance: a secure platform you can't prove is secure. Governance without identity: a documented platform anyone can walk into. You need both — the layer most Azure write-ups skip: sjwiggers.com/2026/08/11/a...
sjwiggers.com
Azure Governance and Identity for Integration Architects - Cloud Perspectives
Explore Azure governance and identity to ensure compliance and security in your integrations. Learn why they matter.
000
Steef-Jan Wiggers @steefjan.bsky.social · 07/08/2026
Azure API Management's new AI Gateway tier fronts Foundry, AWS Bedrock and Vertex AI models behind one endpoint, with policy cards instead of XML. Runtime keys are gateway-scoped though, and no coexistence story for existing gateways. www.infoq.com/news/2026/08...
infoq.com
Azure API Management Adds Dedicated AI Gateway Tier, Governing Models and MCP Tools
Microsoft released a dedicated AI Gateway tier of Azure API Management in public preview, with a control plane built around models, MCP servers and tools rather than APIs. It fronts Foundry, Bedrock, ...
000
Steef-Jan Wiggers @steefjan.bsky.social · 06/08/2026
CosmosEscape: one Gremlin query, sandbox escape, then a platform-wide key with read/write on every Cosmos DB database. Hotfix in 2 days. Removing the key took until July 2026. Nothing customers could patch, and no way to verify the fix. www.infoq.com/news/2026/08...
infoq.com
Wiz Discloses CosmosEscape, and Practitioners Debate What Customers Could Have Done
Wiz Research disclosed CosmosEscape, a chain that escaped Azure Cosmos DB's Gremlin sandbox and reached a platform-wide key granting read and write access to every database on the service. Microsoft b...
011
Steef-Jan Wiggers @steefjan.bsky.social · 05/08/2026
Logic Apps Standard closed a real gap: dev and prod now use one Managed Identity model; no connection-string swap before deploy. Wrote up why that's a bigger Zero Trust deal than it looks, plus a working sample. sjwiggers.com/2026/08/06/m...
sjwiggers.com
Managed Identity in Logic Apps Standard: A Zero Trust Read - Cloud Perspectives
Logic Apps Standard now runs Managed Identity locally and in production. Here's why that closes a real Zero Trust gap and what RBAC still owes you.
001
Steef-Jan Wiggers @steefjan.bsky.social · 05/08/2026
The least interesting question about the integration data layer is SQL vs Cosmos. The real work: idempotency stores for duplicate deliveries, the outbox for dual writes, durable state for long-running workflows. sjwiggers.com/2026/08/05/a...
sjwiggers.com
Azure Data Patterns for Integration Architects - Cloud Perspectives
Explore Azure data patterns to understand how integration impacts data storage and processing beyond just selecting a service.
000
Steef-Jan Wiggers @steefjan.bsky.social · 04/08/2026
Roughly 98.4% of a production coding agent is harness infrastructure. Only 1.6% is AI decision logic. Which is why Microsoft shipping a supported harness, not just an SDK, is the story. Where agents run, what they touch, whose policy. www.infoq.com/news/2026/08...
infoq.com
Microsoft Agent Framework Harness and Hosted Agents Reach General Availability
Microsoft's Agent Framework now ships a supported runtime. Build 2026 brought the Agent Harness, the GitHub Copilot and Claude Agent SDK connectors, and the orchestration patterns to stable release; t...
220
Steef-Jan Wiggers @steefjan.bsky.social · 04/08/2026
Took the FinOps "token economics" framework and held it against a real APIM gateway. Attribution and caching: solid. Goodput, yield, routing: honest gaps. And multi-agent token math is not linear: chains compound context. sjwiggers.com/2026/08/04/t...
sjwiggers.com
Token Economics: Understanding AI Cost Management - Cloud Perspectives
Explore token economics in AI cost management and discover how the Citadel Platform implements these concepts effectively.
000
Steef-Jan Wiggers @steefjan.bsky.social · 30/07/2026
Three ways to host MCP servers on Azure Functions — not interchangeable. GA binding extension. Preview self-hosted SDK (Streamable HTTP only). Queue-based option that skips MCP when async decoupling is what you need. sjwiggers.com/2026/07/30/h...
sjwiggers.com
Hosting MCP Servers on Azure Functions: GA vs. Preview - Cloud Perspectives
Hosting MCP servers on Azure Functions gives you three options: the binding extension, self-hosted MCP SDKs, and queue-based tools, and tells you which to use when.
000
Steef-Jan Wiggers @steefjan.bsky.social · 29/07/2026
The Citadel series wraps: registering an agent in the AI registry via Azure API Center. Every agent carries a structured AI Publish Contract: what it is, what data it touches, who owns it. sjwiggers.com/2026/07/29/c... #azure #apicenter #euaiact
sjwiggers.com
Registering a Citadel Agent in the AI Registry: Azure API Center - Cloud Perspectives
Learn how to register an AI agent in Azure API Center using the AI Publish Contract pattern, and how it supports EU AI Act Article 49 transparency.
000
Steef-Jan Wiggers @steefjan.bsky.social · 28/07/2026
Airbus scored cloud providers on protection from non-European extraterritorial law, right next to scalability and AI. Scaleway won. Not an AWS exit: a sovereign tier for exposed workloads, hyperscaler for the rest. www.infoq.com/news/2026/07...
infoq.com
Airbus Makes Protection from Extraterritorial Law a Scored Criterion in its Cloud Tender
Airbus selected Scaleway as its sovereign cloud provider after a tender that scored protection against non-European extraterritorial legislation alongside technical capability. Airbus frames it as com...
032
Steef-Jan Wiggers @steefjan.bsky.social · 28/07/2026
Months in the Logic Apps agent loop, so my first reaction to Automation wasn't "nice designer." It's that apps are private by default; your compliance team can't see what they do. Wrote up the governance angle + a demo: sjwiggers.com/2026/07/28/l...
sjwiggers.com
Logic Apps Automation Preview - Cloud Perspectives
Logic Apps Automation is in public preview. An architect's view of the new SKU, the governance boundary it introduces, a demo you can build, and what it cannot do yet.
130
Steef-Jan Wiggers @steefjan.bsky.social · 23/07/2026
KBaaS in Azure Logic Apps removes the hard part: the rag pipeline. Upload documents → parse, chunk, summarise, embed into cosmos db. no ai search needed. Built an hr policy agent + github sample. five undocumented notes in the post. sjwiggers.com/2026/07/23/logic-apps-knowledge-base-as-a-service/
sjwiggers.com
Knowledge Base as a Service in Azure Logic Apps - Cloud Perspectives
Build a RAG-powered HR policy agent using Knowledge Base as a Service in Azure Logic Apps.
000
Steef-Jan Wiggers @steefjan.bsky.social · 22/07/2026
Integrate 2026 brought some Logic Apps announcements: Logic apps automation · Knowledge as a service · Foundry agent integration · Standard SDK · Connector namespace sjwiggers.com/2026/07/22/logic-apps-integrate-2026-announcements/ #azurelogicapps #integrate2026 #azure #enterpriseintegration
sjwiggers.com
Azure Logic Apps at Integrate 2026: The Announcements - Cloud Perspectives
Azure Logic Apps Integrate 2026 announcements signal a platform shift. This post unpacks five announcements every integration architect should know and what they mean in practice
000
Steef-Jan Wiggers @steefjan.bsky.social · 22/07/2026
AWS showed customers estimated bills in the trillions. Its own alarms fired but paged nobody; customers escalated 4.5 hours later. Mitigation? Disabling budget and cost anomaly alerts platform-wide. www.infoq.com/news/2026/07...
infoq.com
AWS Billing Bug Shows Customers Trillion-Dollar Estimates While Its Own Cost Alarms Fail to Act
A configuration change in AWS's bill computation system showed customers estimated bills in the billions and trillions of dollars for over 24 hours. AWS's own alarms detected the anomalies but failed ...
021
Steef-Jan Wiggers @steefjan.bsky.social · 21/07/2026
Citadel series continues: the five APIM gateway policies, in depth. Token limiting, semantic caching, content safety, cost attribution, and PII redaction one gateway, every call governed. sjwiggers.com/2026/07/21/c... #azure #apimanagement #aigovernance
sjwiggers.com
Citadel APIM Gateway Policies Azure: A Deep Dive - Cloud Perspectives
Deep dive into five Citadel APIM gateway policies on Azure — token limiting, semantic caching, content safety, cost attribution, and PII redaction with real policy XML, live demos, and pitfalls from a...
010