![Code snippet using generateText with the web fetch tool.
const { text } = await generateText({
model: 'openai/gpt-6.1-sol',
prompt: 'Read https://ai-sdk.dev and summarize.',
tools: {
webFetch: createWebFetchTool({
allowedDomains: ['ai-sdk.dev', 'vercel.com'],
requireUrlInUserMessage: true
})
}
})](/pic/orig/media%2FHVVDEE9aYAYLPk8.jpg)
great advice - personally, where possible, i also find it to be helpful to own the surface where artifacts are created / maintained for portability reasons.
something like justhtml.sh, a 3rd party note taker app, or google docs work fine. it’s easy enough to give whatever personal agent you choose access to these surfaces and that way if you decide to switch agents (or let multiple agents collaborate) you don’t silo all your generated data across multiple agent offerings.
something like justhtml.sh, a 3rd party note taker app, or google docs work fine. it’s easy enough to give whatever personal agent you choose access to these surfaces and that way if you decide to switch agents (or let multiple agents collaborate) you don’t silo all your generated data across multiple agent offerings.
After 40 days of onboarding an agent to my personal life, here's the advice I'd give you all for getting started.
1) Choose a product that you are OK giving a lot of data to (luckily we have many great options. I still think @Muse is best for personal agents atm.)
2) Give your your calendars, and have it make a central calendar that shows all of the events together. Then have your agent analyze your calendar, and share it's thoughts with you. Then chat with it deeply until you two are aligned. Your agent will now be in charge of your calendar. Pro Tip: have it monitor your daily routine, and alert you when there is traffic on your route in advance.
3) Give your agent access to your email. Go into the connector settings, and clearly define all of the permissions you want it to have. I recommend "Allow" for all "read actions" and "ask" for all write / delete actions. I also recommend having it log all of its activity in a spreadsheet that you can audit in the beginning, and have it send you an end of day recap for the actions it took. Also, tell it to triage your inbox how you prefer.
4) At this point, your agent will have enough details about you to be helpful. Now, you need to identify 2-3 things that it can help you with. Mine are grocery shopping / meal prep / fitness tracking / selling items on marketplace / etc. Spend time onboarding your agent to each process.
5) For anything that needs permissions, add your username + password to the secret store only (don't share it in chat)
6) Stop using all other note taking / reminder tools. Everything like this now goes to your agent. That context is gold for your agent to help you more.
7) When working with your agent, especially in the first 30 days, treat every convo like a learning opportunity. If the result is bad, tell it why. If it's not helping enough, tell it how to help more. This is how it actually learns what GOOD is for you.
Enjoy the process. At first, it feels laborious, and that's because it is. You're teaching a computer about how you like to live / work / etc. And you're very special, and complex. It takes a while for anything to learn this. Have patience! The results will follow.
Want more? I shared it all in the video below
0
0
3
185
feels like 1password for agents is a big unlock to bring a whole cohort of new customers into the personal agent space securely
Does your agent need to sign in? With 1Password for Agents, agents can sign in and complete tasks securely without seeing or storing passwords, secrets, or credentials.
The 1Password for Agents platform is building with companies @Muse, @bot, @AnthropicAI @claudeai, @OpenAI, @Browser_Use, @heydana_ai, Instinct, @usekernel, @Perplexity_ai, @tryramp, @Tomo, and @wajo_ai, with more integrations on the way 🎉
0
0
2
207
since i am already using @thelinqapp connector via vercel for my diy personal agent, it was a one-shot prompt in codex to add voice notes. linq has lots and lots of functionality already built in if your agent needs it.
1
0
9
508
I prefer to get my daily updates via gpt-image-2 generated manga style panels
ebook I bought on parenting from ebooks.com turned out to be drm-locked, which is annoying cause i want to have my personal agent review the pdf and be able to develop a weekly guide to send me on what i should be doing with baby, development milestones to hit and d also, so my agent can answer questions from the authoritative source i want and not just from the general internet.
cool workaround i have implemented is giving my agent access to the website through a @usekernel browser + vaulted credentials. this way i can access the info i want without needing to utilize a shady drm-to-pdf converter online.
0
0
1
120
let your personal agents do work across your web logins without seeing your passwords, supported via @1Password x @usekernel
Today we announce 1Password for Agents platform, building with companies @Muse, SpaceX Super Intelligence Grok @bot, @AnthropicAI’s @claudeai, @OpenAI, @TownAI, @browserbase, @browser_use, @heydana_ai, Instinct, @usekernel, @perplexity_ai, @tryramp, @tomo, and @wajo_ai, with e way. 🎉
We've heard from so many of our customers who want to use these new personal AI agents to help them get more done, but want more control over what the agent can access with the same trust and ease of use they get when using 1Password directly.
1Password for Agents gives your agents access without giving up your passwords. When an agent reaches a sign-in form, it requests a credential from your vault. If you approve the request, 1Password directly auto-fills the credential in the destination site so the agent can’t see or take possession of the credentials.
The agent gets the task done, but it doesn’t get your passwords..
You stay in control. 1Password for Agents free and optional for personal and family accounts. Unless you opt-in to an agent connector, agents cannot touch your vault.
➡️Learn more: bit.ly/487da6J
1
0
11
642
Danny Prevoznik retweeted
Today we announce 1Password for Agents platform, building with companies @Muse, SpaceX Super Intelligence Grok @bot, @AnthropicAI’s @claudeai, @OpenAI, @TownAI, @browserbase, @browser_use, @heydana_ai, Instinct, @usekernel, @perplexity_ai, @tryramp, @tomo, and @wajo_ai, with e way. 🎉
We've heard from so many of our customers who want to use these new personal AI agents to help them get more done, but want more control over what the agent can access with the same trust and ease of use they get when using 1Password directly.
1Password for Agents gives your agents access without giving up your passwords. When an agent reaches a sign-in form, it requests a credential from your vault. If you approve the request, 1Password directly auto-fills the credential in the destination site so the agent can’t see or take possession of the credentials.
The agent gets the task done, but it doesn’t get your passwords..
You stay in control. 1Password for Agents free and optional for personal and family accounts. Unless you opt-in to an agent connector, agents cannot touch your vault.
➡️Learn more: bit.ly/487da6J
We've heard from so many of our customers who want to use these new personal AI agents to help them get more done, but want more control over what the agent can access with the same trust and ease of use they get when using 1Password directly.
1Password for Agents gives your agents access without giving up your passwords. When an agent reaches a sign-in form, it requests a credential from your vault. If you approve the request, 1Password directly auto-fills the credential in the destination site so the agent can’t see or take possession of the credentials.
The agent gets the task done, but it doesn’t get your passwords..
You stay in control. 1Password for Agents free and optional for personal and family accounts. Unless you opt-in to an agent connector, agents cannot touch your vault.
➡️Learn more: bit.ly/487da6J
637
1,083
10,551
1,772,280
Danny Prevoznik retweeted
KERNEL now supports @1Password for agents. your agent can sign in without ever seeing your password.
it asks for the login it needs. you approve the request in 1password. the password goes straight into the KERNEL browser and never to the agent.
it asks for the login it needs. you approve the request in 1password. the password goes straight into the KERNEL browser and never to the agent.
4
7
28
5,646
favorite part of this news is that sharp tech will be in video format going forward!
Ben has for many years been the best analyst in tech. I’m very much looking forward to this new @stratechery + @stripe partnership.
stratechery.com/2026/presented…
0
0
0
128
another practical use case for computer-use agents: tourist visa paperwork.
congrats to @alymoursy for launching @VeezaAI on yc today.
powered by @usekernel browsers.
congrats to @alymoursy for launching @VeezaAI on yc today.
powered by @usekernel browsers.
Today we're launching @VeezaAI (YC F26) - an AI tourist visa agent for people with weak passports.
Yes, you read that right. Weak passports.
If you have a US or EU passport, you book a flight and go. For the rest of the world, every trip starts with forms, bank statements, and ment.
We've suffered through this for 30 years. Those days are coming to an end.
Pick your destination. Veeza AI does the paperwork and books the appointment.
start at: veeza.ai
1
3
16
814
ebook I bought on parenting from ebooks.com turned out to be drm-locked, which is annoying cause i want to have my personal agent review the pdf and be able to develop a weekly guide to send me on what i should be doing with baby, development milestones to hit and d also, so my agent can answer questions from the authoritative source i want and not just from the general internet.
cool workaround i have implemented is giving my agent access to the website through a @usekernel browser + vaulted credentials. this way i can access the info i want without needing to utilize a shady drm-to-pdf converter online.
cool workaround i have implemented is giving my agent access to the website through a @usekernel browser + vaulted credentials. this way i can access the info i want without needing to utilize a shady drm-to-pdf converter online.
0
0
6
296
minted a new domain just to accompany this cookbook i helped write: browserrepl.com
a common browser use pattern emerging involves defaulting to structured tools (like webmcp) w/ code mode as a fallback via an on-vm persistent node.js runtime we call browser repl
a common browser use pattern emerging involves defaulting to structured tools (like webmcp) w/ code mode as a fallback via an on-vm persistent node.js runtime we call browser repl
sites without an api don't have to stay that way. on KERNEL, an agent writes a helper once in the browser repl, reuses it, and saves it as a webmcp tool.
0
0
10
276
amazing startup program i recommend applying for, if you are planning to add a browser primitive to your product capabilities.
get started with the best browser infra out there
get started with the best browser infra out there
introducing KERNEL for start-ups. you build agents, we cover the browsers.
early-stage teams shipping web agents get up to $5k in credits. available for start-ups backed by yc, a16z speedrun, or accel.
1
0
12
332
Danny Prevoznik retweeted
personal shoppers that index every product in real time. job searches built around you. our ceo @juecd__ on why agents need reliable access to act on the live web, on @redpandadata's hello, agent!
1
3
24
2,264
practical example: my personal agent keeps track of weekly kernel image repo changes in longer-form docs using a tiktok-style scroll
i do the same thing right now for our docs repo changes too
i do the same thing right now for our docs repo changes too
been using justhtml.sh so my diy personal agent can choose to render complex answers as proper html instead of a wall of imessage text. great for cron job summaries too
1
0
10
479