We're going live on YouTube next Thursday, Oct 29 at 10am PT, for a spooky edition of the Netlify Community Showcase.

Four builders from the community will demo what they've made on Netlify, and applications to be featured are open now. @thisiskp_ and @taniacryptid are hosting.
a

Sometimes you need an answer before you need another code change.

@taniacryptid shows how Ask Mode in Netlify Agent Runners lets you troubleshoot errors and ask questions about your project without jumping to another AI tool or triggering a build.

@thisiskp_ @taniacryptid @deepbane @gehrigkunz We're live at 1 PM ET with @thisiskp_, @taniacryptid, @deepbane, and @gehrigkunz. Watch on YouTube: youtube.com/watch?v=PbVs-p…

Testing AI models shouldn't require juggling provider accounts, credit balances, and API keys.

@taniacryptid explains how Netlify AI Gateway makes it easier to explore different models and experiment with what works best for your app.

On Thursday, Oct 22 at 1 PM ET, @thisiskp_ and @taniacryptid are going live with two of the people who built Interactive Agent Runners, Vanessa Ramos (@deepbane) and Gehrig Kunz (@gehrigkunz).

They'll walk through how each change came together and demo it live, from Ask mode to lines of code you want the agent to change.

youtube.com/watch?v=PbVs-p…

Seeing what you're building while you're building it makes a difference.

@taniacryptid shows how Netlify Preview alongside Agent Runners makes it easier to review changes and iterate, using a copy review app she built for her marketing and community work.

How do you run an AI workflow every morning without leaving your laptop open?

@mr_rtl wanted to publish a document automatically at 6 a.m., so he used Netlify Agent Runners to create a scheduled serverless function.

He'd never even heard of a cron job before.

The changelog example sorts a support ticket into bug, feature, or question. You hand Jev the ticket text as state, declare the three labels you'll accept, and the answer comes back as one of them. There's nothing to parse on your end.

netlify.com/changelog/type…

What happens when business leaders can work directly with their own data?

@mr_rtl shares how access to platforms like @salesforce, @HubSpot, and @Snowflake has helped him move beyond spreadsheets and build tools tailored to his team's workflows.

Last-minute livestream tomorrow: @taniacryptid is walking through AI Gateway, and she's keeping it beginner-friendly.

Set a reminder on YouTube! youtube.com/live/vRuFHqlGk…

Agent Runners gets Gemini 4 Argon alongside AI Gateway, so an agent building in Netlify can run on it. The changelog example is six lines, and the only model-specific piece is the gemini-4-argon string.

netlify.com/changelog/gemi…

AI isn't just for work.

@mr_rtl built a mobile-friendly site to organize everything his son needed for a sporting event, then updated it from the sidelines using Agent Runners.

Sometimes the best projects solve everyday problems.

Four models landed here, and the changelog says what each one is for.

→ Muse Spark 1.1 covers coding, computer use, and multimodal perception.

→ Spark 1.2 is tuned for agentic coding workflows.

→ Spark 1.3 handles long-horizon work with cleaner code and fewer tokens.

→ one for complex coding and long-running agents.

netlify.com/changelog/meta…

"How long did it take to build?"

"Maybe an hour."

@mr_rtl walks through a custom sales dashboard he built that morning, complete with Salesforce links, live metrics, and interactive views tailored to the way he works.

Build or buy?

@mr_rtl shares how Netlify approaches this decision in the AI era. If a tool helps the team move faster and fits naturally into the workflow, buying it can be the right choice.

The changelog example is a Function that takes a prompt off the query string and returns the generated image itself, with the right content type on it. Aspect ratio and image size go in the config, and AI Gateway handles the connection to Google.

netlify.com/changelog/gemi…

Netlify is returning to Barcelona tomorrow!

Wednesday, October 7, with @HackBarna and Happy Operators. We'll have a live Agent Runners demo from @deepbane, a fireside chat with Dana Lawson and @WadeWegner, and then an open mixer.

Every attendee gets 3,000 credits and swag. 4mZXp5Vo

The change most SvelteKit projects will hit is the lib alias. It moves from a dollar prefix to a hash prefix and now runs on Node's subpath imports, so a directory import no longer resolves on its own. The migration tool rewrites them. If a build fails afterward with a module for lib imports with no file name on the end.

netlify.com/changelog/2026…

The edge node picks a compute node with rendezvous hashing, so a function lands on the same node every time. Its MicroVM stays warm and its code stays cached. When one site takes a big share of a region's traffic, we spread it across a slice of nodes.

The request lands on the Netlify edge node closest to the client. The node terminates TLS and checks the path against the deploy's Edge Functions routes. Anything without a matching route goes on to the cache and origin as usual.

A warm request to a Netlify Edge Function now spends about 6ms on our side before the response heads back out. It passes through an edge node, a compute node, and a Firecracker MicroVM without leaving our network.

Agent Runners gets GPT-6.1 Sol alongside AI Gateway. In a Function, the changelog example runs it through the Responses API with the model set to gpt-6.1-sol, and AI Gateway takes care of the authentication.

netlify.com/changelog/gpt-…

Here's the program for our Barcelona meetup on October 7:

Product Manager Vanessa Ramos will demo building with Agent Runners live.

After that, Happy Operators will host a fireside chat with our CTO, Dana Lawson, and CPO @WadeWegner about where the web is heading.

The night s of drinks and snacks with the Netlify team.

A few more numbers from the rollout: Invocations are 47.4% faster at p99, and availability sits at 99.998%. Edge function logs now arrive 5x faster. Cold starts happen on about 1.2% of invocations and average around 9ms.

Agent Runners gets Sonnet 5.5 alongside AI Gateway, so an agent building in Netlify can run on it. In a Function, the changelog example sets the model to claude-sonnet-5-5 with a max_tokens of 4096, and the rest is a normal Anthropic SDK call.

netlify.com/changelog/clau…

Netlify is coming back to Barcelona on Wednesday, October 7. We're co-hosting an evening community meetup with @HackBarna and Happy Operators.

Everyone who comes gets 3,000 Netlify credits and some Netlify swag to take home!

Don't rewrite the same prompt every time.

@JJEnglert explains how AI skills turn repeatable workflows into reusable building blocks that you can share with your team and use again and again.

Agent Runners picks these up along with AI Gateway, so an agent building in Netlify can use them too. The changelog example calls Sol through the Responses API with the model set to gpt-6-sol.

netlify.com/changelog/gpt-…

Opus 5.5 is available in Agent Runners and AI Gateway, so the agent building your app and the app it builds can both run on the same model. Set the model string to claude-opus-5-5 and use the effort setting on the output config when you want it to work harder on a given request.
W

A simple line can make a big difference:

"If you have any questions, ask me."

@JJEnglert explains why inviting AI to clarify requirements leads to better results, and then demonstrates how connecting Claude to Netlify allows it to create projects and deploy directly into your account.

The build is only as good as the plan.

@JJEnglert shows how creating a detailed implementation plan first helps AI agents identify dependencies, parallelize work, and build more efficiently.

Natural language is becoming part of the developer workflow.

@JJEnglert and @thisiskp_ discuss how AI is reducing the amount of time spent on technical setup so builders can focus more on creating and iterating.

Here is the full walkthrough if you want to watch it work.

A real support ticket goes in, and Jev comes back with the team that owns it, a frustration score, and a probability that it reads as urgent. The confidence value on each answer decides what gets handled automatically

TypeSafe's Jev answers typed questions about your program state instead of writing prose. You hand it the state and declare the answers you are willing to accept. A routing question can come back as sales, support, or spam and nothing else, which means you can branch on it g anything.

netlify.com/changelog/type…

International Generalist Day started in Melbourne and Amsterdam yesterday and has been moving west ever since.

If you are the person at your company who somehow owns the frontend, the deploys, and the changelog, today is for you!

luma.com/generalist.eve…

Feeling like you're behind on AI?

@JJEnglert says we're all still early. Businesses are learning, workflows are changing, and there's still plenty of room to experiment, build, and figure out what works.

AI lets you build almost anything. The better question is: *Should you?*

@JJEnglert from @tenex_labs explains why his team focuses on building what strengthens their competitive advantage and buys the rest.

Generalists, September 16 is your day! Volunteers are hosting more than 100 International Generalist Day events across 40+ countries.

Find one near you: luma.com/generalist.eve…

Describe what you want to build before you think about how to build it.

Seán Mc Ternan explains how staying in the ideation phase longer helped one hackathon team finish the core of their project in about 45 minutes and spend the rest of their time improving it.

Cursor Origin repos can now deploy on Netlify.

Push, preview, and ready to rollback in a click. Live in beta.

netlify.com/blog/cursor-or…

ChatGPT Images 2.5 ships as two models, and Netlify supports both. OpenAI points to Flare for fast everyday generation and Sunburst for work that needs precise edits. The changelog includes a Function that returns the generated image as a JPEG.

netlify.com/changelog/chat…

Making it easier to build has brought more people into software development.

Seán Mc Ternan talks about Netlify's rapid community growth and why developer platforms have to keep evolving alongside increasingly capable AI models.

Your first prompt doesn’t have to be perfect. Agent Runners now prioritizes what to build first, giving you a useful starting point to build on.

Same ambitious prompt. Different result:

Before - Cancelled with no preview
████████████████████ 100% credit use
+ a plan
███████████████████░ 95% credit use

netlify.com/changelog/2026…

What if your AI agent could use a service without you clicking through a dashboard?

Seán Mc Ternan and @thisiskp_ talk about how MCP is changing software workflows by giving agents direct access to the capabilities they need, with less manual setup.

There are more AI models to choose from than ever.

@jherr demonstrates how OpenCode makes it easier to compare models, understand estimated costs, and choose the right one for the task at hand.

"Everybody needs this easy one-click deploy."

Chris Watts from @OpenRouter reflects on using Netlify in its early days and why simple deployment quickly became something developers expected, not just appreciated.

Colin Wren from @stripe built and deployed a Tic-Tac-Toe app in three minutes. Provisioning it by hand would have taken another 10 to 20.

Stripe Projects created the Netlify account and project first, so Codex only had to write the app.

You don't need to be tied to your laptop while AI is working.

@thisiskp_ and Chris Watts from @OpenRouter discuss why cloud-based AI agents are becoming a natural fit for longer-running development tasks.

Point the Google GenAI SDK at gemini-3.8-flash and the Netlify Function you already wrote keeps working the way it did before. AI Gateway handles caching, rate limiting, and auth underneath it. Available now.

netlify.com/changelog/gemi…