Today we're launching @mastra `Harness`
`Harness` gives you a `Session` with plan/build modes.
It lets you manage thread lifecycle, spawn or fork subagents, approve tools per-call or session, ask user questions, queue/steer.
With skills, sandbox, browser, heartbeat. channels:
We just moved the entire Frigade marketing site off our no-code CMS back to @nextjs in a week.
We're not using any external services other than @vercel. All content is stored i markdown files so not even any need for a headless CMS. Using markdown also allows to easily leverage Claude Code for authoring and cleaning up content and keeping a consistent voice and tone.
The whole pitch of no-code was solving one problem: engineering attention is too expensive to spend on a marketing site, so route around it with a tool a non-engineer can drive.
That problem is mostly gone now.
Tbh it turned out to be surprisingly easy. And for something as straightforward as a marketing site, the stakes of vibe-coding are way lower than on a production app. We have no issue having non-technical teammates push to main on that repo.
My bet is we're not the only team that's going to make this trip in the coming months.
Hardest part of building AI products is that nobody believes you until they see it.
You can join whatever hype circle is happening on X and talk about how productive AI has made you. But there are so few real raw demos out there.
So here's one. I recorded Frigade working inside @vercel. Not scripted or staged. The AI learns the product, highlights UI elements, and walks users through actual workflows. When the product changes, it adapts.
I've been building this for over a year and it still blows my mind.
Easier to just watch it.
The market has been flooded with AI chatbots, and most of them have the same problem:
They are isolated from the product and the user experience.
So when a user gets stuck, they have to explain everything to the bot from scratch. It's an incredibly clunky experience trying to describe software problems in plain words.
And then because of that, the output is just not that helpful. Usually a wall of bullet points copied directly from the help center.
We ran a simple test: same scenario, same user problem.
One assistant had workflow context. The other didn't.
The usefulness wasn't even close.
Frigade (@FrigadeHQ) uses Browser Use to create superpowered website copilots that actually help users.
"Other companies claim to have stealth mode, but we never got it working. Browser-Use cloud worked instantly and out of the box"
- Christian, CTO @ Frigade
Check out our case study with them below ⬇️
🫡 Released another customer success story for @browser_use today, this time with @FrigadeHQ (YC W23).
When @christianmat gave me a demo live in his office, I was mind-blown. From our transcript: "I don't know how to put into words how amazing this is."
Having an in-app assistant that keeps the user in control and helps them genuinely learn how to use and navigate an app is incredibly helpful.
Much love to the Frigade team, and we're so happy to be on this journey with you guys!
Check out the blog post below ⬇️
The day I signed up for browser-use, @gregpr07 DM'ed me and immediately showed up in person at our office in San Francisco together with @larsencc . Within hours, they helped us fully migrate from our proof-of-concept OSS browser-use setup to a production-ready system.
No one lives and breathes @ycombinator 's "write code/talk to customers" more than this team.
https://t.co/xSGgSC5Tih
Frigade (@FrigadeHQ) learns how your product works and creates in-product guidance from simple prompts. No flows. No selector tagging. No stale docs.
Guidance stays up to date as your UI evolves.
https://t.co/mdmrJKGBLU
Frigade AI started with a simple idea: What if an in-app agent could actually understand your product, not by reading static docs or following rigid scripts, but by using it like a real user?
Our agents explore your product, learn your workflows, and adapt automatically as things evolve.
The result is real-time, contextual support that guides users step-by-step, takes actions on their behalf, answers open-ended questions, and escalates to live support as needed — all with no extra work from your teams.
Check out our new website for more details → https://t.co/4AbHYxw36y
40% Better, 75% Faster AI: How Frigade Slashed Latency & Boosted User Helpfulness 🔥
The short answer: we applied the tips from my "Effective AI Engineering series". Follow me to apply the same tips and supercharge your AI systems!
.@FrigadeHQ is launching an AI assistant that intelligently guides users through your product.
It helps users through workflows, takes actions on their behalf, and answers product questions—all with minimal engineering setup.
https://t.co/zOcFR8wFhw
Hot take: Most AI “copilots” are glorified chatbots. They skim your help center and spit out some bullet points. Basically some “hopes and prayers” that your users will figure it out. The burden is still on the user to follow through. And the burden is also there for companies to keep their help center up-to-date with the product as well.
The solution: Show, don't tell.
That’s why we built Frigade AI: a new kind of assistant that understands product UI and user context. Unlike other copilots that just regurgitate help center instructions, Frigade AI will automatically guide your users in-product and show them how to immediately take action and solve their current problem. In some cases, it will even do it for them (!).
Our agent automatically learns the ropes of your product, ingests your help center and documentation – no engineering or manual work required.
This is a next gen copilot.
Everyone’s freaking out about the so-called @v0 system prompt leak. But here’s the truth: it doesn’t matter.
Writing prompts is the easiest part of building with LLMs. If your system collapses because someone saw your prompt, you were never building anything serious in the first place.
The real challenge: RAG, search, observability, validation, and testing. That’s where the actual moat is. And guess what? You can’t steal or leak those with clever prompt hacking :)
Got a chance to play around with the @OpenAI Operator that just launched. Watch when the agent unprompted clicks and passes he "Verify you are human" CAPTCHA.
Are we cooked? 😅
@garrytan I was at the Mission BART station two days ago and literally feared for my life while waiting for the train to arrive. Who allows this to happen in a civilized country?