Computer nerd from Brazil turned engineer, entrepreneur, and investor at @HeyHoVentures. I love to build things and experiment with new technology. HeyHo!
San Francisco, CA
Joined August 2007
- Tweets12K
- Following3.8K
- Followers7K
- Likes9.9K
Hello hardware builders 👋 I'm looking for ideas and prior art.
Has anyone built a sensing companion for AI agents? A friendly way to give your favorite agent vision, and maybe other senses later. 🧵
I loved this design... my idea was something like this, programmable, with better resolution, battery powered, wifi capable and pole mountable. To be honest, I don't need the AI capabilities on the camera, I can offload that to the agents. Image processing is a good to have on the device. overview.ai/products/ov-spar…
Hey @WhatsApp team! Is there any way to connect our WhatsApp accounts to our personal AI agents? I mean, a fully supported and secure way to do that?
And I'm not talking just about @Muse, but any agent a user chooses to use.
Not having that option is starting to feel like we're being deliberately locked out of our own data—or even held hostage by your platform. Why should WhatsApp decide which AI agents we can use?
The best solution would be an official offering from @Meta / @WhatsApp, with proper security, privacy, and access controls. It would make WhatsApp far more valuable to everyday users and even more central to how we communicate today.
The alternatives are far from ideal: either clunky workarounds (I'm currently using Computer Use or Browser Use when needed) or unofficial, unsupported third-party solutions that pose security risks and could get your account banned.
There has to be a better way!
Huge thanks to everyone who replied with tips, workarounds and solutions. This thread has been gold, and it convinced me even more that an official path is needed.
What I’d love to see from @WhatsApp, and would gladly pay for:
An official, opt-in way to connect my personal WhatsApp to the AI agent of my choice (@claudeai , @OpenAI , @bot , @NousResearch , @openclaw , @instinct , etc). Any agent, not only @Muse or other @Meta product.
1. Read access: my agent can read the chats I choose, not all or nothing.
2. Send on my behalf: with modes I pick per chat: draft only (I approve), send with notification, or fully autonomous.
3. Agent identities: agents can have their own accounts, clearly labeled as AI, so people know when they’re talking to my agent and when they’re talking to me. Charge per extra agent or seat.
4. Granular permissions per chat and group: read, reply, start new conversations, access media, manage groups.
5. Rate limits and guardrails: caps on outgoing messages per agent, with paid tiers for higher limits. This also protects the network from spam.
6. Audit log and kill switch: see everything an agent read and sent, and revoke access in one tap.
How it could work: a local MCP server in WhatsApp Desktop would be a great start. A cloud-hosted remote MCP with OAuth-style consent would be even better, so agents keep working when my laptop is closed.
On privacy: encryption between WhatsApp users stays intact. The agent becomes one more endpoint I authorize, like a linked device. If that agent runs in the cloud, my messages leave my device, so make me explicitly accept that for each chat I open to it. Also let the other people in the chat see that an agent is present.
People are already doing this today with unofficial hacks that risk account bans and put no limits on what the agent can do and has an awful experience.
Done well, with strong defaults, real controls and clear consent, this would make WhatsApp much more valuable to its users. Many of us want this today and can’t get it safely. It would also make WhatsApp far stickier in the agent era that’s coming. The messaging app my agents can work with is the one I’ll never leave.
An official path would be safer for users, better for the network, and a new revenue line for WhatsApp.
PS:. This kind of thinking goes for any products and services we use daily. It is huge need, a challenge and a tremendous opportunity.
Questions I don't have good answers to yet:
• Wi-Fi, BLE or wired: what's actually reliable?
• desk-mounted, handheld or both?
• how should the agent "ask" to see?
• lighting without glare?
• which sense comes after vision?
Manoel Lemos retweeted
In just 24 hours 🤯
✓ 𝟲,𝟴𝟭𝟳 vulnerabilities identified
✓ 𝟱𝟴𝟬 critical or high
✓ 𝟭,𝟰𝟭𝟭 already patched by Agent
Vercel Security Reviews are 100% free until Nov 10
Today we are launching Vercel Security Review in public beta.
Run vulnerability reviews on every push, and ship fixes in one click with Vercel Agent.
Read the blog → vercel.com/blog/vercel-secur…
Kudos @PeteShadbolt and the whole team. When you dream big and work hard, the results compound. Keep rocking!
AI is pushing computing infrastructure to its limits. Our photonics can help.
Today, PsiQuantum is unveiling its industry-leading Optical Circuit Switch initiative with an 8×8 OCS demo built on the company’s low-loss silicon photonics platform—the same platform we’ve spent nearly a decade developing for quantum computing.
Manoel Lemos retweeted
AI is pushing computing infrastructure to its limits. Our photonics can help.
Today, PsiQuantum is unveiling its industry-leading Optical Circuit Switch initiative with an 8×8 OCS demo built on the company’s low-loss silicon photonics platform—the same platform we’ve spent nearly a decade developing for quantum computing.
This is hot!
Founders send me pitches every day. Here’s a better way: come to our hackathon!
EVOLVE: 20 hours, overnight at 8090 in Redwood City, Nov 13–14.
Build AI agents and tools that build software. The winning team, beyond recognition and a cool trophy gets two 1:1s with me, to help you refine a pitch or idea and then a follow-up.
Use the time however you want. Ask me for advice, ask me for money or just make fun of me. All allowed if you win!
Apply: 8090-solutions.link/evolve-w…
Manoel Lemos retweeted
Effective Immediately, all organizations using PHP for their systems are advised to be prepared for an update addressing two critical vulnerabilities. PHP is widely used for web applications. These vulnerabilities often provide a direct path to the server's file system or database. If an immediate update is not feasible, it is recommend that organizations take the system offline until it can be fully mitigated. NSA will follow up shortly with more specific guidance.
Creator Micro 2 from @work_louder is awesome, but the Input app sucks. Not a problem... let's build the software I need. Done!
This is hot!!!
We’re hiring a Head of LATAM Product Partnerships at OpenAI, based in São Paulo.
We’re looking for someone who knows the region deeply, loves building longterm, trusted partnerships, and can turn ambitious ideas into things people actually use. This person will help shape how we grow across Latin America. Warning: this is not a cushy tech job. This is a role for an entrepreneurially minded partnerships athlete wanting to make a massive positive impact.
If this sounds like you, or someone you’d love to see in this role, please apply or pass it along.
openai.com/careers/head-of-l…
