Blogsby GlyphWed Dec 31, 20254 posts
2025: Replit in Review
2025 was our biggest year yet. We shipped hundreds of features that made building faster, more accessible, and more powerful - from foundational Agent functionality to Design Mode, Fast Build, & an expanded Free Tier.
Solving the "Impossible" in ClickHouse: Advent of Code 2025
At ClickHouse, we don't like the word "impossible." We believe that with the right tools, everything is a data problem. To prove it, we decided to complete the 2025 Advent of Code unconventionally: using pure ClickHouse SQL.
Tue Dec 30, 20253 posts
Your product data just got a job: Workflows is now out
If your product data was a person, they'd basically be unemployed. They could get picked up in the occasional "freelance" query, but would ideally be…
A look back at 2025 demo development at ClickHouse
Have a look at some of the most notable demos we built this year, and how they help people feel the speed of ClickHouse through real, hands-on examples.
+1more from ClickHouse Blog that day
Mon Dec 29, 20254 posts
The Holiday Whisper: Shai-Hulud 3.0
A refined variant of the Shai-Hulud malware, dubbed The Golden Path, has been discovered targeting the npm ecosystem during the holiday season.
How to fix your app onboarding drop-off points
A PostHog recipe for people whose users keep leaving the table. Prep time ~30 minutes setup, ~1 week of data Difficulty Beginner-friendly Yields One…
Redesigning the retail operating model in the agentic era
Learn how agentic AI and agentic merchandising reshape retail operating models so teams make faster decisions with greater precision.
Good technology blogs: a reading list for the holidays
A collection of my favorite technology blogs
Sun Dec 28, 20251 post
Keeping 20,000 GPUs healthy
How we do active and passive monitoring on hyperscalers and neoclouds.
Fri Dec 26, 20251 post
How to Build a Chrome Plugin Using Novita AI Sandbox
Learn how to build a Chrome extension with an AI assistant that follows you across pages, explains code, and runs it safely in a sandbox.
Wed Dec 24, 20251 post
Distillable Models and Synthetic Data Pipelines with NeMo Data Designer
How to generate license-safe synthetic data workflows for model specialization
Tue Dec 23, 20258 posts
DeepSeek V3.2 VRAM: Is Local Deployment Still Practical Today
Explore DeepSeek V3.2 VRAM requirements and its impact on hardware costs and performance in real-world applications.
Confidential Compute for AI Inference: How Chutes Delivers Verifiable Privacy with Trusted Execution Environments
Your prompts. Your data. Hardware-protected from everyone, even us. Chutes TEEs: Confidential AI you can verify, not just trust The Chutes network is designed for an adversarial world: miners are anonymous and permissionless.
How Replit Built Effortless Email Sending Features
Replit built straightforward email sending features using Resend, allowing builders to focus on creating without worrying about email infrastructure.
Meet Logs (beta) – more debugging context, all in PostHog
Every developer's debugging journey ends at the same destination. You can start with an error and get context with a session replay, but…
What Are Database Keys and Their Types Explained with Examples
Learn what database keys are, their types, and how they work with real examples. Understand primary keys, foreign keys, and best practices.
New functions you might have missed in 2025
Over 100 functions were added to ClickHouse in 2025. These were some of my favorites.
Google's year in review: 8 areas with research breakthroughs in 2025
Google 2025 recap: Research breakthroughs of the year
Mon Dec 22, 20259 posts
Building a Coding Agent with Novita’s Agent Sandbox
Learn how to use Novita’s LLMs and Agent Sandbox to build a secure environment where your agent can code, build, and experiment without the risk of “breaking anything.”
Now You Can Build with Replit in ChatGPT
Another barrier between idea and application has just disappeared. You can now build apps inside ChatGPT with Replit. Connect Replit to your ChatGPT account and turn conversations into working software. No tabs to switch. No code to write.
Why small teams crush tiger teams
There are a few core milestones in the lifecycle of a PostHog employee. The first one is when your friends or family ask, "So… what does your company…
11 Pythonic Data Quality Recipes for every day
11 practical, copy-paste data quality recipes for dlt. From schema freezes to alerts, learn how to keep pipelines clean, safe, and production-ready
Build a dinosaur runner game with Deno, pt. 3
Adding obstacles, collision detection and game mechanics to our example browser-based game.
Influencing BM25 ranking with multiplicative boosting in Elasticsearch
Learn why additive boosting methods can destabilize BM25 rankings and how multiplicative scoring provides controlled, scalable ranking influence in Elasticsearch.
What really matters for performance: lessons from a year of benchmarks
A look back at some of my favorite benchmark-backed ClickHouse blog posts this year, and the lessons they revealed about performance in practice.
+1more from ClickHouse Blog that day
Continuously hardening ChatGPT Atlas against prompt injection
OpenAI is strengthening ChatGPT Atlas against prompt injection attacks using automated red teaming trained with reinforcement learning.
+1more from OpenAI News that day
Fri Dec 19, 202513 posts
ERNIE-4.5-VL-A3B VRAM Requirements: Run Multimodal Models at Lower Cost
Explore the challenges of deploying ERNIE-4.5-VL-A3B VRA and learn about hardware requirements and cost-efficient alternatives.
Evo Adds CycloneDX Support to Give Full AI Visibility
Enterprises face a critical visibility gap in AI models, creating security and compliance risks. Evo's new integration with CycloneDX 1.6 delivers intelligent, actionable AI-BOMs, providing complete oversight and robust governance for your...
Deploy autoscaling services, AI Workflow automation, and LLM APIs Without Kubernetes
Learn how to deploy containers without Kubernetes. Covers microservices, CI/CD with autoscaling, workflow automation tools like n8n, and lightweight LLM inference hosting.
Ship it: Omni’s 2025 Product Highlights
Last year, our team shipped a lot to help you do more with your data. This year, we pushed Omni further so you could go from answer to action faster.
Design is more than code
Designing with code is a recurring discussion, and it’s flared up again. I’ve written about it twice now. Design as a search, not a pipeline. How tools carry opinions, and how those opinions shape what feels “reasonable” to attempt.
Form 17-H Just Got Easier
Broker-dealers can use the Workiva platform to prepare and file 17-H reports and more to the SEC. Read more.
Durable Digest: December 2025
It’s been a packed year at Temporal. In this issue: major product launches, a builder spotlight from Netflix, and two fresh “How to Temporal” resources
Speed Up Your JavaScript Apps: Native Bun Support is Now Available on App Platform
The JavaScript ecosystem is rapidly evolving to meet the growing performance and integration demands of modern, AI-driven applications.
AI Actor Tilly Norwood and the Impact of Cloud Infrastructure
+1more from Akamai Blog that day
Contributing to Debezium: Fixing Logical Replication at Scale
How we contributed two features to Debezium to solve WAL growth and enable safer logical replication for our event streaming infrastructure
Elasticsearch Serverless pricing demystified: VCUs and ECUs explained
Learn how Elasticsearch Serverless pricing works for Elastic’s fully-managed deployment offering. We explain VCUs (Search, Ingest, ML) and ECUs, detailing how consumption is based on actual allocated resources, workload complexity, and...
Sync Stripe Data to Your Supabase Database in One Click
Announcing official support for the Stripe Sync Engine in the Supabase Dashboard. Get a one-click integration that syncs your Stripe data directly into your Supabase database.
Alexey's favorite features of 2025
Looking back at Alexey's favorite ClickHouse features of the year.
Thu Dec 18, 202516 posts
AI adoption has outrun data accountability
AI adoption is moving faster than our ability to understand the data behind it and that gap is now the biggest blocker to real AI value. Organizations are rapidly pushing AI from pilots into production.
Introducing stratified sampling for LaunchDarkly Experimentation
Support fair, reliable experiment outcomes by eliminating hidden sample bias.
How to Access GLM-4.6V and Build Reliable Multimodal Agents？
Discover how to access GLM-4.6V for improved multimodal workflows and enhanced image and document interpretation.
Top 10 new features in 2025
A recap of some of the best features Resend launched this year.
+1more from Resend Blog that day
Inside Replit’s Snapshot Engine: The Tech Making AI Agents Safe
How Replit's snapshot engine makes AI agents safe: instant filesystem forks, versioned databases, and isolated sandboxes enable reversible AI development.
IBM Db2 to PostgreSQL Migration: 2 Reliable Methods
Learn how to migrate or sync data from IBM Db2 to PostgreSQL using batch exports or automated pipelines like Estuary. Compare methods, tradeoffs, and best practices.
Token-count-based Batching: Faster, Cheaper Embedding Inference for Queries
Embedding model inference often struggles with efficiency when serving large volumes of short requests, a common pattern in search, retrieval, and recommendation systems.
AI for Insurance Finance: How to Get Reporting Relief (Without Regulatory Grief)
Learn how AI tools built for insurance finance teams can help ease the operational burden of manual work while complying with regulations.
10 Principles for Keeping the Vibe while Coding Streamlit Apps
Vibe coding has changed how we build apps. You describe what you want, and after a few seconds, AI generates the code. After a few rounds of iteration and you have a working Streamlit dashboard app. It’s genuinely magical.
Full-stack Kubernetes FIPS compliance: Why it matters
A deep dive into Kubernetes FIPS compliance: FIPS validation, FIPS 140-2/140-3 crypto modules, BoringCrypto, and what true full-stack FIPS support requires.
Response Healing: Reduce JSON Defects by 80%+
Today we're launching Response Healing, a new feature on OpenRouter that automatically fixes malformed JSON responses from LLMs before they reach your application
+1more from OpenRouter Blog that day
December 2025 newsletter
Welcome to the December 2025 ClickHouse newsletter, which will round up what’s happened in real-time data warehouses over the last month.