Archives
-
Witnessing the past decade: our innovations and achievements in transformer production
Dear Sir or Madam
Hello, we are a manufacturer from China. Our company mainly produces a series of products such as transformers, box-type substations, high and low voltage switchgear, etc. Our company has passed many certifications and tests. We have our own patented technology and have solved electricity problems in many places. In addition, we have the advantages of quality and price. These words may seem to be the same, but when you get to know them further, each company must have its own characteristics.If you have this need, please take a look at our company.Below you can find all the information about us.
Need further information? Please contact me.In any case, we are a good choice
Tel/Wechat：+86 15717841020
China Hangbian Electric Power Technology Co., Ltd.
Add:No.99, Binhainan Third Road, Economic Development Zone, Yueqing City, Zhejiang Province
Website：https://www.cnhangbian.com/
WhatsApp:+86 15717841020
E-mail：sales@cnhangbian.com
by "sales09" <sales09@hangbianpower.com> - 02:32 - 12 Jun 2025 -
Chinese consumers: Adapting to a new reality
On McKinsey Perspectives
Brought to you by Alex Panas, global leader of industries, & Axel Karlsson, global leader of functional practices and growth platforms
Welcome to the latest edition of Only McKinsey Perspectives. We hope you find our insights useful. Let us know what you think at Alex_Panas@McKinsey.com and Axel_Karlsson@McKinsey.com.
—Alex and Axel
—Edited by Belinda Yu, editor, Atlanta
This email contains information about McKinsey's research, insights, services, or events. By opening our emails or clicking on links, you agree to our use of cookies and web tracking technology. For more information on how we use and protect your information, please review our privacy policy.
You received this email because you subscribed to the Only McKinsey Perspectives newsletter, formerly known as Only McKinsey.
Copyright © 2025 | McKinsey & Company, 3 World Trade Center, 175 Greenwich Street, New York, NY 10007
by "Only McKinsey Perspectives" <publishing@email.mckinsey.com> - 01:18 - 12 Jun 2025 -
Boost Your PCB Production with Reliable, Cost-Effective Machines
Dear Info,
I’m Ishfaq Sassoli from Zhongshan Meiding Machinery – your partner for precision single/double-sided PCB machines, trusted by industries like:
LED lighting
Automotive electronics
Smart home appliances
Our plug-and-play systems support clients across China, India, Thailand, and Vietnam, offering:
One-Stop PCB Equipment Solutions
Custom Non-Standard Machines
Free Online/Offline Training
Complimentary PCB Production Line Design
We’d be happy to:
1.Share detailed machine specs
2.Provide real production line videos
3.Invite you for a tour of our Zhongshan facility
Let me know if you'd like more info or suggest a time to connect.
Best regards,
ZHONGSHAN MEIDING MACHINERY MANUFACTURING CO.,LTD
Phone/Whats App: +86 15207603870
Email: md_pcbequip@zohomail.com
Website: https://www.mdpcbequip.com/
by "Ishfaq Sassoli" <ishfaqsassoli287@gmail.com> - 03:18 - 11 Jun 2025 -
Shopify Tech Stack
Shopify Tech Stack
Forwarded this email? Subscribe here for moreNote: This article is written in collaboration with the Shopify engineering team. Special thanks to the Shopify engineering team for sharing details with us about their tech stack and also for reviewing the final article before publication. All credit for the technical details and diagrams shared in this article goes to the Shopify Engineering Team.
Shopify handles scale that would break most systems.
On a single day (Black Friday 2024), the platform processed 173 billion requests, peaked at 284 million requests per minute, and pushed 12 terabytes of traffic every minute through its edge.
These numbers aren’t anomalies. They’re sustained targets that Shopify strives to meet. Behind this scale is a stack that looks deceptively simple from the outside: Ruby on Rails, React, MySQL, and Kafka.
But that simplicity hides sharp architectural decisions, years of refactoring, and thousands of deliberate trade-offs.
In this article, we map the tech stack powering Shopify from the modular monolith that still runs the business, to the pods that isolate failure domains, to the deployment pipelines that ship hundreds of changes a day. It covers the tools, programming languages, and patterns Shopify uses to stay fast, resilient, and developer-friendly at incredible scale.
Shopify Backend Architecture
Shopify’s backend runs on Ruby on Rails. The original codebase, written in the early 2000s, still forms the heart of the system. Rails offers fast development, convention over configuration, and strong patterns for database-backed web applications. Shopify also uses Rust for its systems programming language.
While most startups eventually rewrite their early frameworks, Shopify doubled down to help ensure Ruby and Rails are 100-year tools that will continue to merit being in their toolchain of choice. Instead of moving on to another framework, Shopify pushed it further. They invested in:
YJIT, a Just-in-Time compiler for Ruby built on Rust that improves runtime performance without changing developer ergonomics.
Sorbet, a static type checker built specifically for Ruby. Shopify contributed heavily to Sorbet and made it a first-class part of the stack.
Rails Engines, a built-in Rails feature repurposed as a modularity mechanism. Each engine behaves like a mini-application, allowing isolation, ownership, and eventual extraction if needed.
The result is one of the largest and longest-running Rails applications in production.
Modularization Strategy
Shopify runs a modular monolith. That phrase gets thrown around a lot, but in Shopify’s case, it means this: the entire codebase lives in one repository, runs in a single process, but is split into independently deployable components with strict boundaries.
Each component defines a public interface, with contracts enforced via Sorbet.
These interfaces aren’t optional. They’re a way to prevent tight coupling, allow safe refactoring, and make the system feel smaller than it is. Developers don’t need to understand millions of lines of code. They need to know the contracts their component depends on and trust those contracts will hold.
To manage complexity, components are organized into logical layers:
Platform: Foundational services like identity, shop state, and database abstractions
Supporting: Business domains like inventory, shipping, or merchandising
Frontend-facing: External interfaces like the online store or GraphQL APIs
This layering prevents cyclic dependencies and encourages clean flow across domains.
To support this at scale, Shopify maintains a comprehensive system of static analysis tools, exception monitoring dashboards, and differentiated application/business metrics to track component health across the company.
This modular structure doesn’t make development effortless. It introduces boundaries, which can feel like friction. However, it keeps teams aligned, reduces accidental coupling, and lets Shopify evolve without losing control of its core.
Frontend Technologies
Shopify’s frontend has gone through multiple architectural shifts, each one reflecting changes in the broader web ecosystem and lessons learned under scale.
The early days used standard patterns: server-rendered HTML templates, enhanced with jQuery and prototype.js. As frontend complexity grew, Shopify built Batman.js, its single-page application (SPA) framework. It offered reactivity and routing, but like most in-house frameworks, it came with long-term maintenance overhead.
Eventually, Shopify shifted back to simpler patterns: statically rendered HTML and vanilla JavaScript. However, that also had limits. Once the broader ecosystem matured, particularly around React and TypeScript, the team made a clean move forward.
Today, the Shopify Admin interface runs on React, React Router by Remix, written in TypeScript, and driven entirely by GraphQL. It follows a strict separation: no business logic in the client, no shared state across views. The Admin is one of Shopify’s biggest apps, built on Remix that behaves as a stateless GraphQL client. Each page fetches exactly the data it needs, when it needs it.
This discipline enforces consistency across platforms. Mobile apps and web admin screens speak the same language (GraphQL), reducing duplication and misalignment between surfaces.
Mobile Development with React Native
Mobile development at Shopify follows a similar philosophy: reuse where possible, specialize where needed.
Every major app now runs on React Native. The goal of using a single framework is to share code, reduce drift between platforms, and improve developer velocity across Android and iOS.
Shared libraries power common concerns like authentication, error tracking, and performance monitoring. When apps need to drop into native for camera access, payment hardware, or long-running background tasks, they do so through well-defined native modules.
Shopify teams also contribute directly to React Native ecosystem projects like Mobile Bridge (for enabling web to trigger native UI elements), Skia (for fast 2D rendering), WebGPU (that enables modern GPU APIs and enables general-purpose GPU computation for AI/ML), and Reanimated (for performant animations). In some cases, Shopify engineers co-captain React Native releases.
Programming Languages and Tooling
Shopify’s language choices reflect its commitment to developer productivity and operational resilience.
Ruby remains the backbone of Shopify’s backend. It powers the monolith, the engines, and most of the internal services.
Sorbet, a static type checker for Ruby, fills the safety gap traditionally left open in dynamically typed systems. It enables early feedback on interface violations and contract boundaries.
TypeScript is a first-class language on the frontend. Paired with React, it provides predictable behavior across the web and mobile surfaces.
JavaScript still appears in shared libraries and older assets, but most modern development favors TypeScript for its tooling and clarity.
Lua is used for custom scripting inside OpenResty, Shopify’s edge-layer HTTP server built on Nginx. This enables high-performance, scriptable load balancing.
GraphQL is served from the Ruby backend and used across all major clients, such as web, mobile, and third-party apps.
Kubernetes YAML defines infrastructure deployments, service configurations, and environment scaling parameters.
Remix is a full stack web framework used across various aspects of the platform — Shopify Admin Interface, marketing websites, and Hydrogen, Shopify's headless commerce framework for building custom storefronts.
Developer Tooling & Open Source Contributions
A large monolith doesn’t stay healthy without support. Shopify has developed an ecosystem of internal and open-source tools to enforce structure, automate safety checks, and reduce operational toil.
Packwerk enforces dependency boundaries between components in the monolith. It flags violations early, before they cause architectural drift.
Tapioca automates the generation of Sorbet RBI (Ruby Interface) files, keeping static type definitions in sync with actual code.
Bootsnap improves startup times for Ruby applications by caching expensive computations like YAML parsing and gem loading.
Maintenance Tasks standardize background job execution. They make recurring tasks idempotent, safe to rerun, and easy to observe.
Toxiproxy simulates unreliable network conditions such as latency, dropped packets, or timeouts, allowing services to test their behavior under stress.
TruffleRuby is a high-performance Ruby implementation developed by Oracle. Shopify contributes to this as part of its broader effort to push Ruby further.
Semian is a circuit breaker library for Ruby, protecting critical resources like Redis or MySQL from cascading failures during partial outages.
Roast is a convention-oriented framework for creating structured AI workflows, maintained and used internally by the Augmented Engineering team at Shopify.
A much more exhaustive list of open-source software supported by Shopify is also present here.
Databases, Caching, and Queuing
There are two main categories here:
Primary Database: MySQL
Shopify uses MySQL as its primary relational database, and has done so since the platform's early days. However, as merchant volume and transactional throughput grew, the limits of a single instance became unavoidable.
In 2014, Shopify introduced sharding. Each shard holds a partition of the overall data, and merchants are distributed across those shards based on deterministic rules. This works well in commerce, where tenant isolation is natural. One merchant’s orders don’t need to query another merchant’s inventory.
Over time, Shopify replaced the flat shard model with Pods. A pod is a fully isolated slice of Shopify, containing its own MySQL instance, Redis node, and Memcached cluster. Each pod can run independently, and each one can be deployed in a separate geographic region.
This model solves two problems:
It removes single points of failure. An issue in one pod won't cascade across the fleet.
It allows Shopify to scale horizontally by adding more pods instead of vertically scaling the database.
By pushing isolation to the infrastructure level, Shopify contains failure domains and simplifies operational recovery.
Caching and Queues
Shopify relies on two core systems for caching and asynchronous work: Memcached and Redis.
Memcached handles key-value caching. It speeds up frequently accessed reads, like product metadata or user session info, without burdening the database.
Redis powers queues and background job processing. It supports Shopify’s asynchronous workflows: webhook delivery, email sends, payment retries, and inventory syncs.
But Redis wasn’t always scoped cleanly. At one point, all database shards shared a single Redis instance. A failure in that central Redis brought down the entire platform. Internally, the incident is still known as “Redismageddon.”
The lesson Shopify took from this incident was clear: never centralize a system that’s supposed to isolate work. Afterward, Redis was restructured to match the pod model, giving each pod its own Redis node. Since then, outages have been localized, and the platform has avoided global failures tied to shared infrastructure.
Messaging and Communication Between Services
There are two main categories of the same:
Eventing & Streaming
Shopify uses Kafka as the backbone for messaging and event distribution. It forms the spine of the platform’s internal communication layer, decoupling producers from consumers, buffering high-volume traffic, and supporting real-time pipelines that feed search, analytics, and business workflows.
At peak, Kafka at Shopify has handled 66 million messages per second, a throughput level that few systems encounter outside large-scale financial or streaming platforms.
This messaging layer serves several use cases:
Emitting domain events when core objects change (for example, order created, product updated)
Driving ML inference workflows with near real-time updates
Powering search indexing, inventory tracking, and customer notifications
By relying on Kafka, Shopify avoids tight coupling between services. Producers don't wait for consumers. Consumers process at their own pace. And when something goes wrong, like a downstream service crashing, the event stream holds the data until the system recovers.
That’s a practical way to build resilience into a fast-moving platform.
API Interfaces
For synchronous interactions, Shopify services communicate over HTTP, using a mix of REST and GraphQL.
REST APIs still power much of the internal communication, especially between older services and support tools.
GraphQL is the preferred interface for frontend and mobile clients. It allows precise data queries, reduces over-fetching, and aligns with Shopify’s philosophy of pushing complexity to the server.
However, as the number of services grows, this model starts to strain. Synchronous calls introduce tight coupling and hidden failure paths, especially when one service transitively depends on five others.
To address this, Shopify is actively exploring RPC standardization and service mesh architectures. The goal is to build a communication layer that’s:
Observably reliable
Easy to reason about
Standardized across all environments
ML Infrastructure at Shopify
The ML infrastructure at Shopify could be divided into two main parts:
Real-Time Search with Embeddings
Shopify’s storefront search doesn’t rely on traditional keyword matching. It uses semantic search powered by text and image embeddings: vector representations of product metadata and visual features that enable more relevant, contextual search results.
Source: Shopify Engineering Blog This system runs at production scale. Shopify processes around 2,500 embeddings per second, translating to over 216 million per day. These embeddings cover multiple modalities, including:
Product titles and descriptions (text)
Images and thumbnails (visual content)
Each embedding is generated in near real time and immediately published to downstream consumers that use them to update search indices and personalize results.
The embedding system also performs intelligent deduplication. For example, visually identical images are grouped to avoid unnecessary inference. This optimization alone reduced image embedding memory usage from 104 GB to under 40 GB, freeing up GPU resources and cutting costs across the pipeline.
Data Pipeline Infrastructure
Under the hood, Shopify runs its ML pipelines on Apache Beam, executed through Google Cloud Dataflow. This setup supports:
Streaming inference at scale.
GPU acceleration through custom ModelHandler components.
Efficient pipeline parallelism using optimized thread pools.
Inference jobs are structured to process embeddings as quickly and cheaply as possible. The pipeline uses a low number of concurrent threads (down from 192 to 64) to prevent memory contention, ensuring that inference performance remains predictable under load.
Shopify trades off between latency, throughput, and infrastructure cost. The current configuration strikes that balance carefully:
Embeddings are generated fast enough for near-real-time updates
GPU memory is used efficiently
Redundant computation is avoided through smart caching and pre-filtering
For offline analytics, Shopify stores embeddings in BigQuery, allowing large-scale querying, trend analysis, and model performance evaluation without affecting live systems.
DevOps, CI/CD & Deployment
This area can be divided into the following parts:
Kubernetes-Based Deployment
Shopify deploys infrastructure using Kubernetes, running on Google Kubernetes Engine (GKE). Each Shopify pod, an isolated unit containing its own MySQL, Redis, and Memcached stack, is defined declaratively through Kubernetes YAML, making it easy to replicate, scale, and isolate across regions.
The runtime environment uses Docker containers for packaging applications and OpenResty, built on Nginx with embedded Lua scripting, for custom load balancing at the edge. These Lua scripts give Shopify fine-grained control over HTTP behavior, enabling smart routing decisions and performance optimizations closer to the user.
Before Kubernetes, deployment was managed through Chef, a configuration management tool better suited for static environments. As the platform evolved, so did the need for a more dynamic, container-based architecture. The move to Kubernetes replaced slow, manual provisioning with fast, declarative infrastructure-as-code.
CI/CD Process
Shopify’s monolith contains over 400,000 unit tests, many of which exercise complex ORM behaviors. Running all of them serially would take hours, maybe days. To stay fast, Shopify relies on Buildkite as its CI orchestrator. Buildkite coordinates test runs across hundreds of parallel workers, slashing feedback time and keeping builds within a 15–20 minute window.
Once the build passes, Shopify's internal deployment tools take over and offer visibility into who's deploying what, and where.
Source: Shopify Engineering Blog Deployments don’t go straight to production. Instead, ShipIt uses a Merge Queue to control rollout. At peak hours, only 5–10 commits are merged and deployed at a time. This throttling makes issues easier to trace and minimizes the blast radius when something breaks.
Notably, Shopify doesn’t rely on staging environments or canary deploys. Instead, they use feature flags to control exposure and fast rollback mechanisms to undo bad changes quickly. If a feature misbehaves, it can be turned off without redeploying the code.
Observability, Reliability, and Security
This area can be divided into multiple parts, such as:
Observability Infrastructure
Shopify takes a structured, service-aware approach to observability. At the center of this is ServicesDB, an internal service registry that tracks:
Service ownership and team accountability
Runtime logs and exception reports
Uptime and operational health
Gem versions and security patch status
Dependency graphs across applications
ServicesDB catalog metadata and enforces good practices. When a service falls out of compliance (for example, due to outdated gems or missing logs), it automatically opens GitHub issues and tags the responsible team. This creates continuous pressure to maintain service quality across the board.
Source: Shopify Engineering Blog Incident response isn’t siloed into a single ops team. Shopify uses a lateral escalation model: all engineers share responsibility for uptime, and escalation happens based on domain expertise, not job title. This encourages shared ownership and reduces handoff delays during critical outages.
For fault tolerance, Shopify leans on two key tools:
Semian, a circuit breaker library for Ruby, helps protect core services like Redis and MySQL from cascading failures during degradation.
Toxiproxy lets engineers simulate bad network conditions (latency spikes, dropped packets, service flaps) before those issues appear in production. It’s used in test environments to validate resilience assumptions early.
Supply Chain & Security
Security isn’t an afterthought in Shopify’s stack, but part of the ecosystem investment. Since the company relies heavily on Ruby, it also works actively to secure the Ruby community at large.
Key efforts include:
Ongoing contributions to Bundler and RubyGems, focusing on dependency integrity and package security.
A close partnership with Ruby Central, the non-profit that oversees Ruby infrastructure.
A $500,000 commitment to fund academic research and performance improvements in the Ruby ecosystem.
The goal isn’t just to secure Shopify’s stack, but to strengthen the foundation shared by thousands of developers who depend on the same tools.
Shopify’s Scale
Shopify's architecture isn’t theoretical. It’s built to withstand real-world pressure—Black Friday flash sales, celebrity product drops, and continuous developer activity across a global platform. These numbers put that scale in context.
$5 billion in Gross Merchandise Volume (GMV) processed on Black Friday.
284 million requests per minute at the edge during peak load.
173 billion total requests handled in a single 24-hour period.
12 terabytes of traffic egress per minute across Shopify’s edge network.
45 million database queries per second at peak read load.
7.6 million database writes per second during transactional bursts.
66 million Kafka messages per second, sustaining Shopify’s real-time event pipelines.
100,000+ unit tests executed in CI on every monolith build.
216 million embeddings processed per day through ML inference pipelines.
>99.9% crash-free session rate across React Native mobile apps.
2.8 million lines of Ruby code in the monolith, with over 500,000 commits in version control.
100+ isolated pods, each containing its stack (MySQL, Redis, Memcached).
100+ internal Rails apps, maintained alongside the monolith using shared standards.
References:
Open-source tools that we use
SPONSOR US
Get your product in front of more than 1,000,000 tech professionals.
Our newsletter puts your products and services directly in front of an audience that matters - hundreds of thousands of engineering leaders and senior engineers - who have influence over significant tech decisions and big purchases.
Space Fills Up Fast - Reserve Today
Ad spots typically sell out about 4 weeks in advance. To ensure your ad reaches this influential audience, reserve your space now by emailing sponsorship@bytebytego.com.
Like
Comment
Restack
© 2025 ByteByteGo
548 Market Street PMB 72296, San Francisco, CA 94104
Unsubscribe
by "ByteByteGo" <bytebytego@substack.com> - 11:40 - 11 Jun 2025 -
How happy are you at work?
On McKinsey Perspectives
Brought to you by Alex Panas, global leader of industries, & Axel Karlsson, global leader of functional practices and growth platforms
Welcome to the latest edition of Only McKinsey Perspectives. We hope you find our insights useful. Let us know what you think at Alex_Panas@McKinsey.com and Axel_Karlsson@McKinsey.com.
—Alex and Axel
•
Struggling to prioritize well-being. Most people want to have happy colleagues at work, yet many leaders aren’t prioritizing workplace well-being, reflects Jan-Emmanuel De Neve, director of the University of Oxford’s Wellbeing Research Centre and coauthor of Why Workplace Wellbeing Matters: The Science Behind Employee Happiness and Organizational Performance. While most managers recognize that improving workplace well-being will benefit their businesses, only about one-third of senior staff considers it to be a strategic priority, De Neve explains. Moreover, only half of those leaders have a specific plan to improve job satisfaction.
•
Good for business and society. There’s a business case for well-being and a social one, too, says De Neve. Workplace well-being affects business performance through productivity, retention, and recruitment, all of which boost companies’ bottom lines. Investing in workplace well-being is also a social imperative, De Neve shares: When people go home in a better mood, there are ripple effects on their communities. Get early access to the full Author Talks interview on the McKinsey Insights App and learn more about the key drivers of workplace well-being.
—Edited by Belinda Yu, editor, Atlanta
This email contains information about McKinsey's research, insights, services, or events. By opening our emails or clicking on links, you agree to our use of cookies and web tracking technology. For more information on how we use and protect your information, please review our privacy policy.
You received this email because you subscribed to the Only McKinsey Perspectives newsletter, formerly known as Only McKinsey.
Copyright © 2025 | McKinsey & Company, 3 World Trade Center, 175 Greenwich Street, New York, NY 10007
by "Only McKinsey Perspectives" <publishing@email.mckinsey.com> - 01:06 - 11 Jun 2025 -
High-quality switch supplier invites you to cooperate
Dear Info,
We are a leading high-quality switch manufacturer. With many years of industry experience and high-quality products, we serve many partners around the world and are widely praised.
Our core products:
Micro switch: high sensitivity, suitable for electrical appliances, automobiles, industrial equipment, medical and other industries;
Rocker switch: diversified design to meet different needs;
Push button switch: durable, widely used in a variety of industries;
Keyboard switch: comfortable feel, designed for user experience.
Our factory has passed ISO certification and provides fast delivery, OEM/ODM services, and can tailor solutions according to your specific needs.
I hope to have the opportunity to understand your company's needs and explore the possibility of cooperation. If you need samples or quotations, please feel free to contact me and I will provide you with detailed information.
Best regards&Thanks
David Wu
Yueqing Tongda Wire Electric Factory
ADD:No.8 Chuanger Road,Yuqing Bay Port Area, Wenzhou,Zhejiang,325609
Tel: 0086-577 57158583
Mobil: 0086-13780102300
Web: www.chinaweipeng.com
Mail: wdm@tdweipeng.com(If email Failed, pls send to weipeng@tdweipeng.com)
Made-In-China: tongdaswitch.en.made-in-china.com
by "Momen Slama" <momenslama451@gmail.com> - 10:01 - 10 Jun 2025 -
Professional Solutions to Cutting Challenges in the Building Decoration Industry
Dear Info,I'm Xavier from Koocut Cutting Technology (Sichuan) Co., Ltd. Our company has extensive experience in the cutting - tool industry, thanks to the technical accumulation of our parent company over the past two decades.In the building decoration industry, precision and efficiency are of utmost importance. Our PCD cement - fiber - board - processing circular saw blades and TCT color - steel - plate - processing circular saw blades are specifically designed to address the cutting challenges in this field. These blades can cut various building materials with high precision, resulting in smooth edges and minimal waste.Our factory, operating in line with Industry 4.0 standards, features intelligent and flexible production capabilities. This means we can quickly adjust production according to your specific order requirements, whether it's a small - batch customized order or a large - scale standard order. We also have a strict quality control system to ensure that every product leaving our factory meets the highest quality standards.We've already worked with many building decoration service providers globally, helping them improve project efficiency and customer satisfaction. We believe we can do the same for your company. If you have any questions or are interested in our products, please contact me.Best regards,
Contact Information:
WhatsApp:+86 17388372772
Email: xavier@koocut.comKoocut Cutting Technology (Sichuan) Co., Ltd.
by "Beaner Lovely" <beanerlovely@gmail.com> - 09:59 - 10 Jun 2025 -
LIVE from Info-Tech (and what’s actually being said)
LIVE from Info-Tech (and what’s actually being said)
Hi MD Abul,
Greg Benton here, Chief Strategy Officer at Third Stage. Eric couldn’t make it to this year’s Info-Tech LIVE conference, so I’m stepping in to share a few thoughts from the ground.
I’m writing this from a hotel lobby, half-empty coffee in hand, name badge already flipped backward, and that familiar buzz of conference energy all around.
We’re here at Info-Tech LIVE, and like always, it’s impressive.
Screens glowing. AI demos everywhere. Consultants are speed-walking to the next panel. Booths promising transformation in bold letters.
But here’s the thing that always hits me in moments like this:
You can feel the gap between the noise and the real work.Because while the tech looks great on the showroom floor, most of the conversations that matter? They’re happening off to the side. In quiet 1-on-1s. At the coffee stations. In the space between hype and execution.
Earlier today, someone asked me what I’m most excited about in the AI space.
And honestly?
It’s not the tools. It’s the teams finally asking better questions.
Not “What can this new system do?”
But:-
“How do we need to change to actually use this?”
-
“Are our people ready?”
-
“Is this even solving the right problem?”
That’s the shift we’re seeing, the one that actually moves companies forward.
We’ll be here all week, walking the floor, sitting in on sessions, and having the real conversations that don’t fit in a glossy brochure. If you’re here too, come say hi.
And if you’re not, don’t worry. We’ll bring the takeaways back with us.
That’s exactly what our Digital Stratosphere Conference this August is all about. We’re pulling back the curtain on strategy, systems, and what it takes to make AI and digital transformation work in the real world, not just the keynote stage.
Our theme this year is Mission: AI-Possible, and we’d love to see you there.
Use the code "earlybird" for 20% off your ticket(s)!
Best regards,
Greg Benton
by "Greg Benton" <greg.benton@thirdstage-consulting.com> - 09:31 - 10 Jun 2025 -
-
Premium nutritional ingredients you are looking for——in stock in our US warehouse
Dear Info,
Glad to noticed that you are in Dietary Supplements contract manufacturing field in US. Hope this email is helpful for your procurement.
I 'm from Shaanxi Top purity Biotech Ltd. We are a manufacturer of nutritional supplements in China, with a warehouse in California.Our product is factory direct sales, with stable output and favorable prices.
We have been engaged in trading in nutraceutical for over 10 years ,our products have been audited and approved by worldwide reputed customers, such as Glanbia, Nestlé, PIPING ROCK, NBTY, Pharmavite, Now foods, Nutricost, Robinson pharma.
If you want to purchase any Herbs & Botanicals ingredients ,Protein Powders, Probiotics, Enzymes and Specialty products, please feel free to let me know.
Below are our current hot sale products for better check. I believe it will definitely help you save a lot of purchasing costs.
Should you any interest, please feel free to let me know.
Warehouse Add:13735 iroquois Pl,Chino ,CA,91710
Description
Qty (By Kg)
Price（USD/Kg)
Cascara Sagrada Powder
950
9
D-Biotin Tianxin
233
160
Deer Antler Extract 4:1
125
45
Deer Antler powder
275
55
Devil’s Claw root powder
200
8
Diosmin Complex 90% (80%Diosmin&10% Hesperidin)
275
55
Echinacia extract
975
8
Epicatechin 90%
50
200
Fish Collagen
500
18
Forsythia Extract 5-1
350
8
HICA;Alpha-hydroxy-isocaproate Calcium
25
40
Huperzine A 1%
7
500
Inositol -Myo
4500
8
Ipriflavone
100
98
Irish moss powder
300
8
Jambolan Powder
25
10
kale powder
500
5
Ledebouriella Divarciata Extract 5-1
275
8
L-Glutathione Reduced
75
75
Ligusticum Wallichilrhizone Powder
50
8
Maca Extract
225
8
Maca Gelatinized Powder
1000
8
Melatonin
230
72
Milk Thistle Extract 80%
500
30
MK4 1%
81
100
Motherwort Powder
300
5
Mulberry Leaf extract
100
8
Neem Leaf Extract
50
9
Nicotinamide Adenine Dinucleotide 99% (NAD)
300
130
NMN-β-nicotinamide mononucleotide
175
95
Passion Flower Powder
25
8
PHOSPHATIDYLSERINE 20%
25
55
Pterostilbene
20
300
PYRROLOQUINOLINE QUINONE DISODIUM SALT
10
1100
Quercetin Dihydrate powder 95%
700
48
Resveratrol 20%
25
40
Rhodiola Rosea P.E. (Rosavins 3% Salidroside 1%)
75
145
Rutin 95% NF 11
750
28
Saffron Extract 0.3%
25
120
Schisandra Extract 2% HPLC
50
34
Schisandra Extract 4:1
150
8
Schisandra P.E. 2% UV
25
8
Sea -Buckthorn Powder
50
10
Shilajit Extract 20%
100
35
SOD(8000IU )
535
55
Sodium copper chlorophyllin
100
200
Sodium Hyaluronate Haiber
1800
82
Sucralose Techno /jinhe
1800
23
Theobromine 99%
500
47
TUDCA
205
350
Best regards,
Top purity sales team
What's App: 44 7463 099214
Wechat: 17729507767
DL: 1 909-516-8031 / 1 909-678-6521
e-mail: sales3@toppuritybio.com;sales4@sinoingredients.online
by "Biljana Lebowski" <biljanalebowski743@gmail.com> - 05:25 - 10 Jun 2025 -
Cost-Effective Performance
Dear Info,
Balancing performance and cost is critical in the UAV industry. At Micro Magic Inc, we provide high-precision inertial sensors that deliver exceptional value without compromising on quality. Our cost-effective solutions are trusted by leading drone manufacturers worldwide.
Interested in learning more? Let’s arrange a call to discuss further.
Best regards,
Young
Account Manager
Micro Magic Inc
Skypy: manot27
Whatsapp: +8618151836753
Website: www.memsmag.com
by "Karaeski Martirosov" <martirosovkaraeski804@gmail.com> - 12:55 - 10 Jun 2025 -
How Lyft Uses ML to Make 100 Million Predictions A Day
How Lyft Uses ML to Make 100 Million Predictions A Day
Forwarded this email? Subscribe here for moreDatabase Benchmarking for Performance: Virtual Masterclass (Sponsored)
Learn how to accurately measure database performance
Free 2-hour masterclass | June 18, 2025
This masterclass will show you how to design and execute meaningful tests that reflect real-world workload patterns. We’ll discuss proven strategies that help you rightsize your performance testing infrastructure, account for the impact of concurrency, recognize and mitigate coordinated omission, and understand probability distributions. We will also share ways to avoid common pitfalls when benchmarking high-performance databases.
After this free 2-hour masterclass, you will know how to:
Tune your query and traffic patterns based on your database and workload needs
Measure how warm-up phases, throttling, and internal database behaviors impact results
Correlate latency, throughput, and resource usage for actionable performance tuning
Report clear, reproducible benchmarking results
Disclaimer: The details in this post have been derived from the articles/videos shared online by the Lyft Engineering Team. All credit for the technical details goes to the Lyft Engineering Team. The links to the original articles and videos are present in the references section at the end of the post. We’ve attempted to analyze the details and provide our input about them. If you find any inaccuracies or omissions, please leave a comment, and we will do our best to fix them.
Hundreds of millions of machine learning inferences power decisions at Lyft every day. These aren’t back-office batch jobs. They’re live, high-stakes predictions driving every corner of the experience from pricing a ride to flagging fraud, predicting ETAs to deciding which driver gets which incentive.
Each inference runs under pressure and a single-digit millisecond budget. This translates to millions of requests per second. Dozens of teams, each with different needs and models, are pushing updates on their schedules. The challenge is staying flexible without falling apart.
Real-time ML at scale breaks into two kinds of problems:
Data Plane pressure: Everything that happens in the hot path. This includes CPU and memory usage, network bottlenecks, inference latency, and throughput ceilings.
Control Plane complexity: This is everything around the model. Think of aspects like deployment and rollback, versioning, retraining, backward compatibility, experimentation, ownership, and isolation.
Early on, Lyft leaned on a shared monolithic service to serve ML models across the company. However, the monolith created more friction than flexibility. Teams couldn’t upgrade libraries independently. Deployments clashed and ownership blurred. Small changes in one model risked breaking another, and incident investigation turned into detective work.
The need was clear: build a serving platform that makes model deployment feel as natural as writing the model itself. It had to be fast, flexible, and team-friendly without hiding the messy realities of inference at scale.
In this article, we’ll look at how Lyft built an architecture to accomplish this requirement and the challenges they faced.
Architecture and System Components
LyftLearn Serving doesn’t reinvent the wheel. It slots neatly into the microservices foundation already powering the rest of Lyft. The goal wasn’t to build a bespoke ML serving engine from scratch. It was to extend proven infrastructure with just enough intelligence to handle real-time inference, without bloating the system or boxing teams.
At the core is a dedicated microservice: lightweight, composable, and self-contained. Each team runs its instance, backed by Lyft's service mesh and container orchestration stack. The result: fast deploys, predictable behavior, and clean ownership boundaries.
Let’s break down this architecture flow diagram:
HTTP Serving Layer
Every request to a LyftLearn Serving service hits an HTTP endpoint first. This interface is built using Flask, a minimalist Python web framework. While Flask alone wouldn’t scale to production workloads, it’s paired with Gunicorn, a pre-fork WSGI server designed for high concurrency.
To make this stack production-grade, Lyft optimized the setup to align with Envoy, the service mesh that sits in front of all internal microservices. These optimizations ensure:
Low tail latency under high request volume.
Smooth connection handling across Envoy-Gunicorn boundaries.
Resilience to transient network blips.
This layer keeps the HTTP interface thin and efficient, just enough to route requests and parse payloads.
Core Serving Library
This is where the real logic lives. The LyftLearn Serving library handles the heavy lifting:
Model loading and unloading: Dynamically brings models into memory from saved artifacts
Versioning: Tracks and manages different model versions cleanly
Shadowing: Enables safe testing by running inference on new models in parallel, without affecting live results
Monitoring and logging: Emits structured logs, metrics, and tracing for each inference request
Prediction logging: Captures outputs for later audit, analytics, or model debugging
This library is the common runtime used across all teams. It centralizes the “platform contract” so individual teams don’t need to re-implement the basics. But it doesn’t restrict customization.
Custom ML/Predict Code
The core library is dependency-injected with team-owned inference logic. Every team provides two Python functions:
def load(self, file: str) -> Any: # Custom deserialization logic for the trained model ... def predict(self, features: Any) -> Any: # Custom inference logic using the loaded model ...Source: Lyft Engineering Blog
This design keeps the platform flexible. A team can use any model structure, feature format, or business logic, as long as it adheres to the basic interface. This works because the predicted path is decoupled from the transport and orchestration layers.
Third-Party ML Library Support
LyftLearn Serving makes no assumptions about the ML framework. Whether the model uses TensorFlow, PyTorch, LightGBM, XGBoost, or a home-grown solution, it doesn’t matter.
As long as the model loads and predicts through Python, it’s compatible. This lets teams:
Upgrade to newer framework versions without coordination.
Use niche or experimental libraries.
Avoid vendor lock-in or rigid SDKs.
Framework choice becomes a modeler’s decision, not a platform constraint.
Integration with Lyft Infrastructure
The microservice integrates deeply with Lyft’s existing production stack:
Metrics and tracing plug into the company-wide observability pipeline.
Logs and prediction events feed into central analytics systems.
Kubernetes handles service orchestration and autoscaling.
Envoy service mesh provides secure, discoverable network communication.
This alignment avoids duplicating effort. Teams inherit baseline reliability, visibility, and security from the rest of Lyft’s infrastructure, without needing to configure it themselves.
Isolation and Ownership Principles
When dozens of teams deploy and serve ML models independently, shared infrastructure quickly becomes shared pain. One broken deploy can block five others. A single library upgrade triggers weeks of coordination and debugging turns into blame-shifting. That’s what LyftLearn Serving was built to avoid.
The foundation of its design is hard isolation by repository, not as a policy, but as a technical boundary enforced at every layer of the stack.
One Repo, One Service, One Owner
Every team using LyftLearn Serving gets its own GitHub repository. This repo isn’t just for code, but it defines the entire model-serving lifecycle:
The service code includes custom load() and predict() logic.
The configuration files for deployment and orchestration.
The integration hooks for CI/CD, monitoring, and metrics.
There’s no central repository to manage and no shared runtime to coordinate. If a team needs five models, they can choose to host them in one repo or split them across five.
Independent Deploy Pipelines
Each repo comes with its deploy pipeline, fully decoupled from others. This includes:
Staging and production environments.
CI jobs that run model self-tests and linting.
Version tagging and release promotion.
If one team pushes broken code, it doesn’t affect anyone else. If another needs to hotfix a bug, they can deploy instantly. Isolation removes the need for cross-team coordination during high-stakes production changes.
Runtime Isolation via Kubernetes and Envoy
LyftLearn Serving runs on top of Lyft’s Kubernetes and Envoy infrastructure. The platform assigns each team:
A dedicated namespace in the Envoy service mesh
Isolated Kubernetes resources (pods, services, config maps)
Customizable CPU, memory, and replica settings
Team-specific autoscaling and alerting configs
This ensures that runtime faults, whether it’s a memory leaks, high CPU usage, or bad deployment. A surge in traffic to one team’s model won’t starve resources for another. A crash in one container doesn’t bring down the serving infrastructure.
Tooling: Config Generator
Getting a model into production shouldn’t mean learning multiple configuration formats, wiring up runtime secrets, or debugging broken deploys caused by missing database entries.
To streamline this, LyftLearn Serving includes a Config Generator: a bootstrapping tool that wires up everything needed to go from zero to a working ML serving microservice. Spinning up a new LyftLearn Serving instance involves stitching together pieces from across the infrastructure stack:
Terraform for provisioning cloud infrastructure.
YAML and Salt for Kubernetes and service mesh configuration.
Python is used for defining the serving interface.
Runtime secrets for secure credential access.
Database entries for model versioning or feature lookups.
Envoy mesh registration for service discovery.
Expecting every ML team to hand-craft this setup would be a recipe for drift, duplication, and onboarding delays. The config generator collapses that complexity into a few guided inputs.
The generator runs on Yeoman, a scaffolding framework commonly used for bootstrapping web projects, but customized here for Lyft’s internal systems.
A new team running the generator walks through a short interactive session:
Define the team name.
Specify the service namespace.
Choose optional integrations (for example, logging pipelines and model shadowing).
The tool then emits a fully-formed GitHub repo with:
Working example code for the load() and predict() functions.
Pre-wired deployment scripts and infrastructure configs.
Built-in test data scaffolds and CI setup.
Hooks into monitoring and observability.
Once the repo is generated and the code is committed, the team gets a functioning microservice, ready to accept models, run inference, and serve real traffic inside Lyft’s mesh. Teams can iterate on the model logic immediately, without first untangling infrastructure.
Model Self-Testing System
Model serving can often drift when a new dependency sneaks in, slightly changing output behavior. For example, a training script gets updated, but no one notices the prediction shift. Or, a container upgrade silently breaks deserialization. By the time someone spots the drop in performance, millions of bad inferences have already shipped.
To fight this, LyftLearn Serving introduces a built-in Model Self-Testing System. It’s a contract embedded inside the model itself, designed to verify behavior at the two points that matter most: before merge and after deploy.
Every model class defines a test_data property: structured sample inputs with expected outputs:
class SampleModel(TrainableModel): @property def test_data(self) -> pd.DataFrame: return pd.DataFrame([ [[1, 0, 0], 1], # Input should predict close to 1 [[1, 1, 0], 1] ], columns=["input", "score"])Source: Lyft Engineering Blog
This isn’t a full dataset. It’s a minimal set of hand-picked examples that act as canaries. If a change breaks expected behavior on these inputs, something deeper is likely wrong. The test data travels with the model binary and becomes part of the serving lifecycle.
Two checkpoints that matter are as follows:
During Deployment Runtime
After a model loads inside a LyftLearn Serving instance, it immediately runs predictions on its test_data. The results:
Get logged and surfaced in metrics dashboards.
Trigger alerts if predictions drift too far from expected.
Provide an immediate signal of runtime integrity.
This catches subtle breakages caused by environment mismatches. For example, a model trained in Python 3.8 but deployed into a Python 3.10 container with incompatible dependencies.
During Pull Requests
When a developer opens a PR in the model repo, CI kicks in. It performs the following activities:
Loads the new model artifacts.
Runs predictions on the stored test_data.
Compares outputs against previously known-good results.
If the outputs shift beyond an acceptable delta, the PR fails, even if the code compiles and the service builds cleanly. The diagram below shows a typical development flow:
Inference Request Lifecycle
A real-time inference system lives in the milliseconds between an HTTP request and a JSON response. That tiny window holds a lot more than model math. It’s where routing, validation, prediction, logging, and monitoring converge.
LyftLearn Serving keeps the inference path slim but structured. Every request follows a predictable, hardened lifecycle that allows flexibility without sacrificing control.
Here’s a step-by-step on how the request gets served:
Request hits the Flask endpoint: The service exposes an /infer route via Flask, backed by Gunicorn workers. Envoy handles the upstream routing. The incoming payload is parsed and handed off to the core serving library.
The core library retrieves the model: Using the model_id, the runtime looks up the corresponding model. If the model isn’t already loaded into memory, the system calls the team-supplied load() function and loads it on demand. This is where versioning logic, caching, and model lifecycle controls kick in. There’s no global registry and no shared memory. Each service owns and manages its models.
Input validation and preprocessing: Before running inference, the platform performs sanity checks on the features object, such as type or shape validation, required field presence, and optional model-specific hooks. This step guards against malformed inputs and prevents undefined behavior in downstream logic.
User-defined predict() runs the inference: Once the inputs are deemed valid, the system hands control to the custom predict() function written by the modeler. This function converts inputs into the expected format and calls the underlying ML framework (for example, LightGBM, TensorFlow) to return the prediction output. The predict() function is hot-path code. It runs millions of times per day. Its performance and correctness directly affect latency and downstream decisions.
Logs, metrics, and analytics: After the output is generated, the platform automatically logs the request and response for debugging and audit trails. It also emits latency, throughput, and error rate metrics and triggers analytics events that flow into dashboards or real-time monitoring. This observability layer ensures every inference can be traced and every service behavior can be measured.
Response returned to the caller: Finally, the result is packaged into a JSON response and returned via Flask. From request to response, the entire path is optimized for speed, traceability, and safety.
Conclusion
Serving machine learning models in real time isn’t just about throughput or latency. It’s about creating systems that teams can trust, evolve, and debug without friction.
LyftLearn Serving didn’t emerge from a clean slate or greenfield design. It was built under pressure to scale, to isolate, and to keep dozens of teams moving fast without stepping on each other’s toes.
Several lessons surfaced along the way, and they’re worth understanding:
“Model” means different things to different people. A serialized object, a training script, and a prediction endpoint all fall under the same label. Without clear definitions across tooling and teams, confusion spreads fast.
Documentation is part of the product. If teams can’t onboard, debug, or extend without asking the platform team, the system doesn’t scale. LyftLearn Serving treats docs as first-class citizens.
Once a model is deployed behind an API, someone, somewhere, will keep calling it. Therefore, stability is a requirement for any serving system that expects to live in production.
Trade-offs aren’t optional. Seamless UX conflicts with flexible composability. Structured pipelines clash with custom workflows. Every decision makes something easier and something else harder. The trick is knowing who the system is really for and being honest about what’s being optimized.
Power users shape the platform. Build for the most advanced, most demanding teams first. If the platform meets their needs, it’ll likely meet everyone else's. If not, it won’t scale past the first few adopters.
Prefer boring tech when it works. Stability, debuggability, and operational maturity are key aspects to consider.
LyftLearn Serving is still evolving, but its foundations hold. It doesn’t try to hide complexity, but it isolates it. Also, it enforces a contract around how the models behave in production.
References:
SPONSOR US
Get your product in front of more than 1,000,000 tech professionals.
Our newsletter puts your products and services directly in front of an audience that matters - hundreds of thousands of engineering leaders and senior engineers - who have influence over significant tech decisions and big purchases.
Space Fills Up Fast - Reserve Today
Ad spots typically sell out about 4 weeks in advance. To ensure your ad reaches this influential audience, reserve your space now by emailing sponsorship@bytebytego.com.
© 2025 ByteByteGo
548 Market Street PMB 72296, San Francisco, CA 94104
Unsubscribe
by "ByteByteGo" <bytebytego@substack.com> - 11:38 - 10 Jun 2025 -
New: The smartest way to run HR and payroll
New: The smartest way to run HR and payroll
.png?width=1200&upscale=true&name=Email%20Header_EN%20(3).png)
Hi MD,
Growing your team shouldn’t mean juggling different tools for every country and every employee type. But for most companies, that’s the reality: up to 73% of time is lost to manual work, while payroll errors cost thousands.
Introducing Remote HRIS:
Effortless HR. Accurate payroll. All in one place.Remote HRIS brings all your employee data into one place, from hiring and onboarding to leave and time off, expenses, payroll, and more. You can:
- Hire and onboard employees and contractors anywhere
- Run payroll and HR in the same system
- Automate manual tasks with AI and no-code workflows
- Launch performance reviews with the new Remote Perform: a better way to manage feedback, goals, and reviews
And we're just getting started. Coming soon:
- A smarter way to manage compensation
- Automated onboarding workflows
- AI-powered engagement surveys
- ...and much, much more
You received this email because you are subscribed to
News & Offers from Remote Europe Holding B.V
Update your email preferences to choose the types of emails you receive.
Unsubscribe from all future emailsRemote Europe Holding B.V
Copyright © 2025 Remote Europe Holding B.V All rights reserved.
Kraijenhoffstraat 137A 1018RG Amsterdam The Netherlands
by "Remote Team" <hello@remote-comms.com> - 09:21 - 10 Jun 2025 -
Test our US ZIP code data - No signup required
Test our US ZIP code data - No signup required
👋 Hi there! In this edition of The Geodata Insider, I'm excited to share a new way you can explore our ZIP code and city data in the US.
2:30 minute read
📮 Explore our ZIP code data—no signup needed
📈 Master address validation across any platform
🛠️ Multi-language feature enhancement
🔎 Monthly changes
🌊 A sea adrift: Beautiful and borderless
Explore our US ZIP code data—no signup needed
Search any US ZIP code. Get the states, cities, population, accurate coordinates, and standardized formatting. It’s the same data that powers operations at Amazon, MercuryGate, and DB Schenker across 247 countries.
.png?width=100&upscale=true&name=Group%2054%20(1).png)
Master address validation across any platform
Ready to eliminate address errors and boost data quality? Here is a list of step-by-step guides to ensure streamlined address validation.
Don't see your platform? Reply to this email with the system you're missing.
Smarter Language selection
Selecting the right language for users in multilingual countries like Canada and Switzerland has always been challenging.
GeoPostcodes multi-language data exports now include intelligent language recommendations by location, identifying the locally preferred language(s) for each area.
This ensures you always use the most appropriate language for your users' locations, improving data accuracy and user experience.
Monthly changes
Full Postal & Street updates
Australia, Netherlands
Postal database patches
(small changes)
Argentina (Postcodes),
Croatia (Postcodes), Italy (Postcodes), Philippines (Administrative divisions)
Postal boundary patches (small changes)
China, Spain
Administrative boundaries patches (small changes)
Denmark
For a detailed changelog, take a look at the CSV file.
A sea adrift: Beautiful and borderless
Forget everything you know about oceans—there’s one sea that plays by none of the rules. The Sargasso Sea doesn’t touch a single shore. No beaches, no cliffs, no coastline.
Instead, it floats in the middle of the Atlantic, held together by four swirling currents.
Inside this watery loop, a golden jungle of Sargassum seaweed drifts on the surface.
It’s a nursery for turtles, an eel breeding ground, and a pit stop for whales and tuna.The Sargasso is now home to the North Atlantic Garbage Patch: thousands of plastic pieces per square kilometre.
A place that exists without borders… and reminds us what happens when we cross them. Follow us on LinkedIn for more geographical facts like this!
Kind regards,
Jerome & the GeoPostcodes team
PS: Interested in previous Monthly Product Updates? Read here.
GeoPostcodes, Bld Bischoffsheim 15, 1000 Bruxelles, Belgium
by "Jérôme from GeoPostcodes" <jerome@geopostcodes.com> - 06:35 - 10 Jun 2025 -
Excelling as a leader: Decision making
Email 2 of 7
Faster, better decisions
In the previous session, we explored the mindsets and behaviors of growth leaders. In this second email, we'll explore how to cultivate careful—yet bold—decision making. Today’s leaders are faced with hard choices and more pressure than ever. Here are some ways to set yourself up for more successful decisions.
Executives today spend about 40 percent of their time making decisions—and 60 percent say that time is poorly used. The opportunity costs of these inefficiencies are hard to reconcile: for the typical Fortune 500 company, about 530,000 days of managers’ time is potentially squandered each year. That’s equivalent to around $250 million in wages annually.
Much of this time is spent in meetings, says McKinsey senior partner Aaron De Smet. But “most meetings are designed to not make a decision,” he says. So the inefficiency perpetuates itself.
It doesn’t have to be this way. Check out these insights to cultivate your skill as a decision maker under any circumstances.
Meet better
Most meetings aren’t set up for decision making, say McKinsey senior partner Aaron De Smet and senior expert Leigh Weiss. That can be because the roles are unclear—no one knows who is supposed to decide—or the meeting isn’t set up to be productive. But don’t fix the problem of bad decision making in meetings; instead, design meetings around the decision.
Read: To unlock better decision making, plan better meetings
This email contains information about McKinsey’s research, insights, services, or events. By opening our emails or clicking on links, you agree to our use of cookies and web tracking technology. For more information on how we use and protect your information, please review our privacy policy.
You received this email because you subscribed to “The McKinsey Publishing Guide to excelling as a leader.”
Copyright © 2025 | McKinsey & Company, 3 World Trade Center, 175 Greenwich Street, New York, NY 10007
by "McKinsey Publishing Guides" <publishing@email.mckinsey.com> - 03:17 - 10 Jun 2025 -
Your VIP Golf Experience Awaits — Ready to Tee Off in Style?
Hello Sir/Madam,
Are you ready to take your golf game to the next level? At VIP Golf Concierge, we specialise in arranging luxury golf trips at unbeatable prices. Whether you’re dreaming of playing legendary courses in the UK and Ireland or jetting off to the sun-drenched fairways of Spain, Portugal, or beyond — we have you covered!
Here’s what we offer:
- Access to top-tier courses across Britain, Ireland, Europe, and worldwide
- Tee times, accommodation, and transport all taken care of
- A personalised, VIP service that the big tour operators simply can’t match
- But the best part? Book any overseas golf trip with us and you’ll get absolutely free UK golf at renowned venues so you can warm up your game (imagine playing The Belfry, Close House, Celtic Manor, and more – on us!).
Our VIP spots are limited and fill up fast. Don’t miss out on your chance for a truly unforgettable golfing experience.
Ready to get started? Reply to this email or call us on +44(0)7492 189 785 to begin planning your dream golf getaway.
See you on the fairway!
Best regards,
Samantha Taylor
Tour Leader
+44(0)7492 189 785
by "Samantha - VIP Golf Concierge" <info@vipgolfbookings.co.uk> - 03:02 - 10 Jun 2025 -
What are the fashion industry’s biggest themes in 2025?
On McKinsey Perspectives
Brought to you by Alex Panas, global leader of industries, & Axel Karlsson, global leader of functional practices and growth platforms
Welcome to the latest edition of Only McKinsey Perspectives. We hope you find our insights useful. Let us know what you think at Alex_Panas@McKinsey.com and Axel_Karlsson@McKinsey.com.
—Alex and Axel
•
Fashion’s 2025 outlook. Amid consumers’ unease, supply chain concerns, and shifts in global trade, the fashion industry is facing a challenging year. Despite these risks, McKinsey’s latest State of Fashion research reveals that there are still opportunities for brands to capture. That’s according to Senior Partner Gemma D’Auria, who discusses the research’s key findings in a recent episode of The McKinsey Podcast. Fashion leaders can help their businesses grow by using AI and AI-led tools to engage directly with customers and offer personalized shopping experiences, both online and in stores, D’Auria explains. McKinsey’s previous State of Fashion research found that up to 25% of AI’s potential in fashion can enhance the industry’s creative side.
—Edited by Belinda Yu, editor, Atlanta
This email contains information about McKinsey's research, insights, services, or events. By opening our emails or clicking on links, you agree to our use of cookies and web tracking technology. For more information on how we use and protect your information, please review our privacy policy.
You received this email because you subscribed to the Only McKinsey Perspectives newsletter, formerly known as Only McKinsey.
Copyright © 2025 | McKinsey & Company, 3 World Trade Center, 175 Greenwich Street, New York, NY 10007
by "Only McKinsey Perspectives" <publishing@email.mckinsey.com> - 01:28 - 10 Jun 2025 -
Register: Driving innovation with custom Joule skills and agents
Meet Joule Studio: Drive Innovation with AI Agents & Joule Skills
Tuesday July 1st, 2025
[3 PM ET | 3 PM CET | 3 PM SGT]AI tools are taking productivity to new levels. But why stop at out-of-the-box solutions?
Join us for an SAP Webcast featuring Joule Studio in SAP Build and see how to create custom AI agents and Joule skills tailored to your business workflows. Get a look at the developer experience and the tools that help you customize and extend Joule’s capabilities. You’ll discover the essentials of building and managing AI agents that are ready to work for your business.
You’ll also learn more about:
- Understanding Joule skills and AI agents
- Building with Joule Studio for SAP and non-SAP systems
- Identifying the right agentic AI use cases
Kind regards,
The SAP Build team
Contact us
See our complete list of local country numbers
SAP (Legal Disclosure | SAP)
This e-mail may contain trade secrets or privileged, undisclosed, or otherwise confidential information. If you have received this e-mail in error, you are hereby notified that any review, copying, or distribution of it is strictly prohibited. Please inform us immediately and destroy the original transmittal. Thank you for your cooperation.
You are receiving this e-mail for one or more of the following reasons: you are an SAP customer, you were an SAP customer, SAP was asked to contact you by one of your colleagues, you expressed interest in one or more of our products or services, or you participated in or expressed interest to participate in a webinar, seminar, or event. SAP Privacy Statement
This email was sent to info@learn.odoo.com on behalf of the SAP Group with which you have a business relationship. If you would like to have more information about your Data Controller(s) please click here to contact webmaster@sap.com.
This offer is extended to you under the condition that your acceptance does not violate any applicable laws or policies within your organization. If you are unsure of whether your acceptance may violate any such laws or policies, we strongly encourage you to seek advice from your ethics or compliance official. For organizations that are unable to accept all or a portion of this complimentary offer and would like to pay for their own expenses, upon request, SAP will provide a reasonable market value and an invoice or other suitable payment process.
This e-mail was sent to info@learn.odoo.com by SAP and provides information on SAP’s products and services that may be of interest to you. If you received this e-mail in error, or if you no longer wish to receive communications from the SAP Group of companies, you can unsubscribe here.
To ensure you continue to receive SAP related information properly, please add sap@mailsap.com to your address book or safe senders list.
by "The SAP Build team" <sap@mailsap.com> - 12:20 - 10 Jun 2025
