n8n adds shared credits, Threads courts podcasts: Sept. 17
n8n simplifies AI service billing, Threads adds podcast tools, and Jev tests a faster browser agent. Here are today's practical stack moves.

This sheet contains partner links. A purchase through one earns Runbook a commission at no additional cost to you. How we make money.
n8n has removed eleven account setups from the first run of an automation, Threads has turned podcast links into measurable posts, and a new browser agent points to quicker web tasks. Give this dispatch 15 minutes. You'll leave with one billing control to set, one distribution test to run, and one risky experiment kept away from customer systems.
n8n puts eleven services behind one balance
Eleven AI and web services became available through n8n Gateway credits on September 16. Gateway credits are prepaid funds held inside n8n, so a workflow can use another company's service without a separate account or secret software password. n8n says the launch covers six model providers and five tool services, including OpenAI, Anthropic, Google Gemini, Brave Search, Browserbase, Firecrawl, LlamaParse, and PDF.co.
The immediate gain is testing speed, not cheaper production. n8n says it aims to match each provider's published price without adding a convenience premium. The selected provider still receives the prompt, document, or other content sent by your workflow. It does not receive your identity or n8n account details. Gateway credits are available on n8n Cloud Starter and Pro, while self-hosted installations are not included yet.
Open Cloud Admin > Workspace settings and decide whether Gateway credits should be available to the whole workspace. Then open the billing view, set a small balance, and inspect spending by workflow and service after each test. For a live workflow with sensitive customer records, keep your existing direct provider connection until its privacy terms and bill have been reviewed. The guarded n8n agent setup shows how to keep the first version read-only.
Your move
Threads adds podcast cards and click reporting
Meta rolled out six podcast features on Threads on September 16. Its announcement lists profile banners, tappable episode cards, transcript and guest cards, sharing reminders, and Podcast Insights. Insights is the reporting screen. It now shows podcast reach, engagement, clicks, and saves.
This makes Threads testable as a distribution channel. A plain episode link used to tell you little about what happened between a post view and a listen. The new card and click count give you a basic trail. They do not prove that somebody finished the episode or became a customer, so keep the claim narrow.
Pick one recent episode. Add it as the featured episode, publish one native episode card, and note its clicks and saves in Podcast Insights after seven days. Use the same episode title and destination link elsewhere so the comparison is fair. If your team turns transcripts into social posts, send every excerpt through the AI content approval workflow before it reaches Threads. Businesses without a podcast can ignore this release.
Jev cuts browser work to a smaller decision loop
Jev Ultrafast published its first public repository on September 16 with a 7.1-second Google Flights demonstration. A browser agent is software that reads a web page and clicks or types on your behalf. The project's evidence shows a median time reduction from 9.450 seconds to 7.092 seconds across three runs of each version. It reduced browser protocol calls, or the small instructions sent to the browser, from 1,092 to 101.
The useful design change is restraint. Jev reads a numbered list of visible controls, chooses an allowed action, and only asks a small language model to write text when typing is required. Model output cannot become page code, screen coordinates, or shell commands. That is a sensible pattern for repetitive research tasks such as checking public pricing pages.
Do not connect this early release to logins, purchasing, or customer records. Ask a technical owner to run the published demo in a separate Chrome profile with invented data, pause with Choose next, and inspect every proposed click. Three successful runs of one travel task are not a reliability record. For a safer reporting job you can deploy now, use the weekly marketing report build.
On the bench
- Claude's merged Chat, Cowork, and Artifacts interface is rolling out first to Pro and Max accounts. Test the routing before changing team instructions.
- Cloudflare released a six-stage security-audit skill for coding agents. It belongs with a qualified technical reviewer, not as an automatic approval stamp.
- Compare Gateway credits with direct provider billing after one month, especially if several n8n workflows draw from the same balance.
Get the next build from the automation archive.
About Runbook
AI tools and automation builds for marketers. What to use, how to wire it, and the workflow to copy this week. How we work
GET THE NEXT DISPATCH
Run the next build before your competitors read about it.
One short email when an AI tool or automation actually changes the work, with the build to copy.
No send unless there is a build worth running.
// keep_reading
Related builds

ChatGPT texts, n8n adds shared memory: Aug. 21
ChatGPT can send Apple Messages, n8n adds shared customer memory, and Google lets people tune Discover. Here is what to rewire today.

n8n opens MCP, Google adds Ads insights: Aug. 11
n8n opens one-click AI connections, Google adds Ads insights, and Claude marks generated work. Here is what to rewire today.

Google agents expand, GitLost warns: July 8 dispatch
Google's Gemini agent update, GitLost's repo-leak warning, and n8n's memory guidance change how you wire AI agents.