Till | gotoHuman

If you have used AI nodes in @n8n_io you have probably generated markdown at some point. LLMs use it to show headers, tables or code. Want AI to update your internal docs? Make sure to approve it first. Markdown editing in @gotoHuman just got a lot better. n8n wflow template👇

@hwchase17 Great to see what Agent Inbox evolved into. For anyone curious about alternatives, e.g. for richer human reviews, have a look at gotoHuman

Inbox is a way to manage and oversee your agents in LangSmith Fleet Every time they are stuck and need human approval for an action, or have a questions for you, it will show up here and you can quickly cycle through and unblock them

@BrandGrowthOS @hwchase17 you might want to have a look at gotoHuman 's n8n integration

@hwchase17 this is what my typescript agents need badly. they get stuck on edge cases in my n8n workflows and just... wait. having a proper approval queue instead of digging through logs to find where they failed would save hours of debugging

@kimmonismus Not random at all. They started I think 2-3 years ago and already had a sweet list of angels on board. Will be exciting to watch what they ship for sure

A random 10-person team in Paris just dropped what looks even superior to Clawdbot! Twin is everything Clawdbot should've been: - Zero setup (sign up and go) - Runs in cloud, not your laptop - Scales infinitely - Built secure from day 1 Watching this one closely!

@joaomdmoura Absolutely, often both in the same workflow

The autonomy vs control debate in AI agents is a false dichotomy. The best agentic systems give agents autonomy where exploration matters, structure where precision is required. Not either, but both

@jesse_miller @joaomdmoura @gotoHuman You are talking about AMP? But crew ai <> gotoHuman would also work right?

@till_simon @joaomdmoura @gotoHuman Potentially - we support webhooks and an API callback. An integration would be pretty simple.

@joaomdmoura I like "Orchestrated autonomy", might use this

Orchestrated autonomy is what you want in prod: structure + shared state + recovery baked in. Put the “intelligence” where it counts.

🤦‍♂️Bad HITL: agent runs, you approve, done. That’s just latency + pager fatigue lol HITL isn’t “for compliance.” In prod, stuff breaks: bad tools, stale data, edge cases. You need a human checkpoint + replay/recovery so agents can ship safely, not just “be smart.” 🧵

@dexhorthy Interesting pivot Dex. Will definitely follow the new journey even if we're not on the same path anymore. All the best 💪

the best thing about low key pivoting is I can just set up an email filter for "human in the loop" and send all that marketing slop straight to the trash cause we don't do that any more

@GeminiApp @GoogleLabs Is it possible to create plugins/integrations for opal to offer pre-built steps? (as I understand these mini apps are built using opal)

We’re rolling out a new type of Gem from @GoogleLabs that lets you create interactive, AI mini-apps on desktop. These experimental Gems turn your prompts into actionable tools that can help you complete everyday tasks or explore new ideas. Here’s what you can build 🧵

@ctatedev I'm looking into building a plugin for @gotoHuman. Does it support wait-and-resume for human-in-the-loop steps or a webhook trigger step?

Right now there are plugins for AI Gateway, v0, Resend, Linear, Firecrawl and Slack They let you add workflow steps to generate text and apps, send email and messages, create tickets, scrape or search the web, and more Which plugins are you most excited to see next?

@leerob Maybe your take on #2 answers #1 and this is a bit like "kids these days don't even know how to fix a car anymore"

My biggest worries about coding with AI: 1. Beginners not actually learning 2. Atrophy of skills I’m seeing #1 happen and I don’t have a good answer yet. Leveling up as an engineer requires grinding and it’s not always fun. If AI can solve most of the problems for you, when do you lean into the healthy friction? When do you embrace the suck? Coupled with fewer opportunities for pair programming, it’s definitely tougher for those starting their engineering career. It’s not all bleak though. Those with high agency are figuring it out and learning extremely fast. I just worry about the industry as a whole outside these folks. We need better products and better education. I’m hoping to try and do my part here. For #2, I’m definitely paranoid about this for myself. What will it feel like to build software in 5 years? Will I have forgotten someone of the skills I used to rely on? Maybe that won’t even matter because we will truly be operating at a higher level of abstraction. Even if that pans out, it’s always been important to deeply understand the systems/dependencies you’re building on. I normally talk about the stuff I’m optimistic for but think it’s good to have a healthy skepticism here.

@cramforce I like that both keep a human in the loop 👌

① How to find an agent to build that is high business impact and actually works ② We open-sourced the internal ones we built, so you don't have to start from zero

You can also check input against broader policies. Like flag prompt injection attempts or verify that messages stick to a certain topic. Some of these guardrails are LLM-based btw., others use regex pattern matchers or keyword lookups.

The @n8n_io Guardrails node is out (v1.119.0) It helps sanitize LLM input and check its output, but can also be used to hide sensitive data from human supervisors, like - credit card numbers - email addresses, or - phone numbers (This and many more supported out of the box)

I love these disposable one-off workflows. This is how I migrated @gotoHuman from one email provider to another. Sure, I could have vibe-coded a simple script, but having a nice viewer for the data (that I can pin) to inspect responses is just super handy. #n8n #Automation

The OpenAI node in @n8n_io now allows using built-in tools like web search and code interpreter. Pretty handy if you want to fetch some data, generate code to do some calculation or processing on it and then run it - all in a single node.

@n8n_io n8n.io/workflows/7310…

Ok ok, this @n8n_io AI workflow might need a little tweak. But it's still super valuable for anyone sharing posts and articles from their favorite creators and companies. Whatever you post - you write it. But why not have a content assistant that helps out a bit. Workflow below

@n8n_io @gotoHuman Full interview: youtu.be/NG9bFFNNmQg

Built your AI workflow. Presented a magical POC. But how to turn this into a trustworthy solution? When I sat down with Max from @n8n_io, we talked about how a human in the loop can help with reliability and the HITL principles @gotoHuman is built around.