GlideOS launch recap

“We started Glide with a simple but ambitious idea: business software should be much easier to create, much better to use, and much more closely aligned with the way a company actually operates.” - David Siegel, Glide Co-Founder and CEO

At our GlideOS launch event on August 3, 2026, members of Glide's team showed the next major step toward that vision.

GlideOS is a new AI-first platform for building custom operational software. It brings together apps, data, integrations, workflows, and AI in a shared environment designed not simply for building software, but for getting real day-to-day work done.

During the event, Glide co-founder and CEO David Siegel explained why we decided to rethink Glide for the AI era and demonstrated how GlideOS turns spreadsheets and business context into interconnected systems of apps. CTO Shay Frendt went deeper into the technological changes that made GlideOS possible and explained why advances in coding agents led us to rebuild the underlying Glide platform. Product Manager Josh Johnson then walked through recently launched features and shared our roadmap for making GlideOS more secure, collaborative, observable, and useful for enterprise teams.

You can watch the full recorded event here, or sign up to try GlideOS for yourself. Here’s a detailed recap of everything we covered.

Part 1: Vision & Live Demo with David Siegel, Glide CEO and Co-Founder

David opened by taking the team back to Glide's origins, showing a photo of the four founders (Jason, Mark, David, and Antonio) before any revenue, customers, or even a working prototype. Before Glide, the founding team spent five years building traditional developer tools for people writing mobile business applications by hand.

He described the emotional core of why Glide exists: the team was frustrated by how difficult and clunky business software was to build, especially compared to the "mission control" feel of software in movies. “Uplift everyday work” became their focus.

Building GlideOS for the AI era

"AI has arrived, and now anyone can make software." - David Siegel

David was clear that the goal was never to replace spreadsheets, but to extend them: "We respect spreadsheets, and we love spreadsheets. We don't think spreadsheets are bad or broken, but they have their limits." Glide Classic was built to catch spreadsheets right as they hit those scaling limits, when teams needed automation, mobile access, or multi-user collaboration that a spreadsheet couldn't support.

Glide Classic's success is real and worth restating for the team: millions of users have turned millions of spreadsheets into apps across over 100,000 businesses. Customer examples David highlighted:

A car wash chain manages chemical inventory and application recipes entirely through a custom Glide process.
A retail operator runs pickup, drop-off, pricing, and merchandising in real time across store and back-office teams.
An international coffee supplier that moved its wholesale ordering process off spreadsheet templates and onto a fully branded, real-time digital solution.

“A car is washed every 3 minutes in the United States, using a custom Glide process. And it grew up from spreadsheets into a fully custom solution that helps them operate in real time.” - David Siegel

The idea for GlideOS crystallized last fall at GlideAway, the team's annual offsite in Cabo. The team's key observation: our most sophisticated customers were never building just one app. They were building systems of interconnected apps sharing real-time data. Customers themselves were describing these as "operating systems."

"If that's what our most successful customers are doing, how can we make that a lot more common? How can we make that the default, reflexive way that Glide performs for the customer?" asked David.

At the same time, the team wanted to preserve what made Glide work in the first place, starting from a spreadsheet, even in an AI development process. "We really encourage customers to share their spreadsheets because we think a spreadsheet is like the uber-prompt," he explained. A spreadsheet carries business logic, structured data, and formulas that a typed prompt would never fully capture.

Live Demo Walkthrough

David demoed GlideOS live. Here's how you can get started:

Drop a spreadsheet: Glide's new introspection engine, internally called Sheet Sense, scans the file for embedded structure: formulas, comments, macros, pivot tables, charts, and conditional formatting. Because a full, exhaustive scan of a complex spreadsheet would be expensive, Glide does a lighter first-pass scan to identify likely use cases and asks the user to confirm direction.
Use the App Store: Fully pre-built, ready-to-use multi-app systems, free to copy. Examples shown: CableOS (a multi-app system for wiring data center projects) and CourseOS (a 6-app system for golf course operations). David noted these are meaningfully different from old no-code templates: "Your agent already knows this template inside and out, and it can change it to your business needs." Users are encouraged to drop in their own spreadsheet after copying an App Store project so the agent can rewire it to real data.
"Help me get started" button: No data upload required. The agent interviews the user and researches their company using available signals (e.g., a work email domain), acting as an embedded Glide expert to identify where the platform can help most.

“GlideOS is trained as one of our best Glide Experts. It understands the full potential and capabilities of Glide, and it can investigate the nuances of your business to help you zero in on where Glide can help you the most.” - David Siegel

From spreadsheet to app: After dropping the inventory spreadsheet, GlideOS built an initial dashboard and inventory list automatically, faithfully loading all source data. David noted the platform's agentic load and error-correction process. Since spreadsheets are frequently inconsistent or incomplete, the system flags contradictions or missing data and asks the user for clarification rather than guessing silently.

Data trust and transparency: Users need to be able to verify that numbers in the app are real, not hallucinated, a core design principle carried over from Glide Classic. David showed how any calculation or UI element can be inspected, with a natural-language explanation of exactly how a figure was derived, along with direct access to the underlying data. As David put it, the goal is to let customers validate their calculations independently of the interface. “We want to own your most critical business processes and give you the confidence that everything is constructed and is valid,” he explained.

Multi-app systems: David then showed how GlideOS treats work as a system rather than a single app by editing a back-office app and an on-site mobile app side by side, both drawing on shared data, logic, and theming. He noted the agent proactively suggests multi-app structures when appropriate for a customer's use case, rather than requiring the user to architect that themselves.

"Improve my business" button: Once a workspace matures, users are able to use GlideOS as a business consultant. David demoed an "I'm feeling lucky" option where the agent surveys the business and proactively surfaces opportunities. In the live example, the agent identified a compliance gap and an inventory data issue on the golf course project, and David walked through fixing and redeploying the solution.

Agent as operator, not just builder: David's final demo point was that the agent isn't limited to building; it has ongoing access to apps and data once built. He showed the agent being used to query data, write and execute code, and even interact with tools like email and calendar to draft replies. "This agent has full access to the apps and can make changes, but it also can query the data... it can use the apps and not just build and customize them."

Part 2: Engineering Journey & Technical Foundation with Glide STO Shay Frendt

Shay opened with a live stat: in the seven days prior to the event, more than 100,000 new projects had been created on GlideOS. He used a GlideOS-built 3D visualization tool as an example of what the platform now makes possible, built in "just 5 simple chat prompts and around 15 minutes of human effort," a level of output he said he "would have personally never attempted to produce by writing the code myself."

He also cited external context on how AI is changing engineering velocity broadly: PostHog's engineering team reported roughly a 3x increase in delivery volume while growing headcount by only 10%. Ramp used agentic tooling to fix 100 security issues in six days with no human intervention in the execution. “The humans shaped the AI process, and the machines carried it out.”

“We went from being able to ship around 300 pull requests a month to over 2,000 pull requests a month, a nearly 7x improvement in our overall delivery velocity. Notably, we did not grow our engineering headcount at all during this time.” - Shay Frendt

He also noted a broader business trend: companies in the top quartile of AI spend have more than doubled revenue, while bottom-quartile, no-AI-spend companies stayed flat.

The history of AI engineering in Glide

Shay shared his origin story with the company: he joined after meeting David and co-founders Mark and Jason in September 2022, drawn to Glide's mission to put "the power, beauty, and magic of software development into the hands of a billion new creators."

Glide's product approach, going back to that period, has followed three repeated steps:

Break the software development process down into primitives
Watch users interact with those primitives at scale
Iterate on the experience to make development more approachable and effective

Platform primitives shipped over the years included integrations, Glide AI column enrichment, BigTables (a more powerful table/API layer), advanced screen components with adaptive mobile/desktop support, and workflow automation. Adoption data showed Glide AI and Workflows as the fastest-growing capabilities.

The AI column's popularity pushed the team to imagine "a GPT for each part of the product," leading through 2024 hackathon-driven experimentation. During 2024, the team ran hackathons and internal experiments around increasingly conversational forms of software creation. They even experimented with building Glide apps for a phone, on a phone, entirely through chat. Some of those experiments eventually became product features.

One early win: users could type something like "mockup of this product for a marketing flyer" and get a data-connected interface generated automatically. This became one of the platform's top 10 most-used components.

“With the right structure and sequence of chat prompts, we found that LLMs could really produce gorgeous data-driven screens in record time.” - Shay Frendt

By mid-2025, this had evolved into the first-generation Glide Agent, allowing users to build entire apps via chat rather than individual components. That agent still fields over 45,000 messages daily that convert into automated build steps. Two conclusions came out of studying this data: conversational building has clear primacy with users, and, in Shay's words, "LLMs have won" and will produce the majority of the world's software going forward.

Why we rebuilt GlideOS from scratch

Glide Classic represented app logic as an intermediate JSON structure called the app description. Coding agents struggled to reliably produce good applications when limited to working through that layer. When the team instead gave agents direct access to full-stack application code, output quality, such as UI, design systems, and overall fidelity, improved substantially.

That finding led to the decision, roughly a year ago, to begin building the platform's second-generation agent internally called G2.

The new infrastructure partners supporting the GlideOS runtime include Cloudflare (runtime), Descope (identity), Neon (application data), plus multiple frontier AI labs for inference.

“Jensen Huang, CEO of Nvidia, famously said in 2017. Software is eating the world, but AI is going to eat software. Well, live from 2026, it is now happening.” - Shay Frendt

Shay closed with an internal example: Glide’s own engineering team built a GlideOS app connected to GitHub, Datadog, and PostHog that posts a service health digest to Slack, helping the team spot exceptions and prioritize focus. They were essentially using GlideOS to build GlideOS.

Another example cited was from a customer, Rich from HydroPro, who built a full operating system for his hot tub service business, including scheduling, invoicing, and a CRM, in weeks. “Before, customers said things like, I'm not sure I know how to manage relations between tables. I don't know how to use some of the components. But with GlideOS, they're now telling us things like, I just told the robot to do it, and it works, or I can implement an idea when I leave one hot tub service job, and it's done and working by the time I get to the next job,” explained Shay

Part 3: Recent Ships & Roadmap with Josh Johnson, Product Manager at Glide.

Josh explained his role as helping the team's designers and engineers prioritize what to build next, and noted the team's shipping cadence has been unusually fast. What follows is a curated subset of recent releases:

“This team ships at an incredible cadence unlike anything I've ever encountered.” - Josh Johnson

Recently Shipped

App Store, expanded: Reiterating David's demo, Josh emphasized that App Store templates are fundamentally more usable than old no-code templates because the agent fully understands the underlying structure and can modify it safely. Common requests driving this: AI chat apps and CRMs (Rich's hot tub CRM being a direct example).
Usage dashboard: Lets teams see who's building, what they're working on, which projects/apps/workflows/integrations are consuming the most credits, and overall ROI.
Connect GlideOS to 1,000+ tool integrations: Users no longer need to manage API keys or secrets manually. Connecting to tools like Google Sheets, Notion, Gmail, or HubSpot now just requires asking the agent, which handles OAuth directly.
Workflows + integrations: Enable scheduled, automated data syncing (e.g., pulling new spreadsheet data every morning at 8 am) and downstream actions, such as posting daily summaries to Slack. Josh described the underlying pattern the team has found most effective internally: connect your data → build simple apps on top of it → identify repetitive "busywork" → automate it with workflows.

Q3 Roadmap

"The entire roadmap right now is focused on being safe, secure, and collaborative inside of GlideOS," Josh explained.

Observability & Compliance

Usage dashboard (shipped)
Per-app user lists (in progress)
Audit logging, planned for later this year
SOC 2 compliance, in progress

Auth & User Management

SSO (shipped, rolling out to more teams)
Project-level permissions (currently org-level only; project-level requested by many customers)
Guest seats, let large orgs invite many users to a single app without consuming full seats
Publishing approval workflows allow a trusted approver to gate changes before they go live

Data Integrity

AI provider lock-in setting lets enterprises restrict GlideOS to a single approved provider (e.g., Google, OpenAI, Anthropic) via a toggle, addressing existing vendor agreements
Data backups provide a 30-day point-in-time restore, "probably the most important feature we'll ever ship" according to Josh.
App version control (in progress) rollback for full applications, not just data
Code file visibility, GlideOS "is potentially no longer a no-code platform," and that trust increasingly requires letting teams see the code the agent produces

Automations & Integrations

Build with MCP. Josh's personal favorite feature on the roadmap: lets developers build and edit GlideOS applications from inside their own coding agents/tools rather than only inside GlideOS itself.

Product vision for GlideOS

Josh ties the product roadmap back to a long-standing internal vision board: GlideOS should be "a place to work, not just to build applications." Customer feedback is pushing in the same direction. People want employees actually getting work done, not just vibe-coding all day. He used mocked-up examples (a hypothetical Nike growth team dashboard, a hypothetical IKEA supply chain dashboard) to illustrate what a team-specific GlideOS hub could look like, surfacing exactly the apps, data, and workflows relevant to that team.

Try GlideOS for yourself

The whole Glide team wants to thank everyone who came out, asked questions, and engaged with us during the live event. “We want GlideOS to be a fantastic, beautiful place to work that uplifts the everyday work that people do,” reiterates Josh. That's where we're going with GlideOS.

If this event has piqued your curiosity, head to glideapps.com to start building.