Marks
Mark

https://www.trymaple.ai

ceo @MapleAI | previously 

Family 🌴 Freedom Tech

Public Key

npub136jg2fnty2z5vwcnh7p4jpckrs3tk0dpueftgs7mznuuaenjpfps6tjnxf

NIP-05 Address

Profile Code

nprofile1qqsgafy9ye4j9p2x8vfmlq6equtpcg4m8ks7v545g0d3f7wwueeq5scprpmhxue69uhhqun9d45h2mfwwpexjmtpdshxuet5qyd8wumn8ghj7mnjv4kxz7fwvvkhxar9d3kxzu3wdejhg6raqx6

Publishing to

Author Public Key

npub136jg2fnty2z5vwcnh7p4jpckrs3tk0dpueftgs7mznuuaenjpfps6tjnxf

Show more details

Last Notes

2026-06-15 18:48:40 UTC

- reply

Even the little features can hit hard
#nevent1q…xtsf

2026-05-27 21:18:07 UTC

- reply

See you in a few!

2026-05-27 21:17:35 UTC

Maple provides an AI experience that is as close to the privacy of local, offline AI as possible while running in the cloud. We do this by using Trusted Execution Environments (TEE). Data is encrypted locally and only decrypted inside the TEE. If law enforcement requested a user's data, they would receive an encrypted blob. Furthermore, we offer anonymous accounts that don’t have any associated email or social media identity.
We've been open from the beginning. You can see our code and technical writeups:
- Source code: https://github.com/orgs/OpenSecretCloud/repositories
- High level architecture: https://www.trymaple.ai/proof
- Technical Deep Dive: https://blog.opensecret.cloud/opensecret-technicals/
We are already in the process of commissioning third-party audits because we know those are helpful for certain organizations.
I know of no other cloud AI provider, whether it’s proprietary frontier labs or other privacy AI companies, that is more open and transparent than we are. We set the bar high because we believe this industry should be open by default.
We offer state-of-the-art open-weight models with the strongest privacy protections we can build. It’s up to you to decide what risk tolerance is right for you.

2026-05-27 14:56:29 UTC

- reply

🦗

2026-05-27 00:34:27 UTC

Confirmed: I live inside a Curious George town.
Went to the store tonight and ran into my doctor. One minute later ran into the guy that trims our trees. And then right after that saw my dentist, with whom I have an appointment tomorrow morning. He gave me a few pieces of info that he was going to call me about later tonight.

2026-05-27 00:32:37 UTC

Went to the store tonight and ran into my doctor. One minute later ran into the guy that trims our trees. And then right after that saw my dentist, with whom I have an appointment tomorrow morning. He gave me a few pieces of info that he was going to call me about later tonight.
Do I live inside a Curious George town?

2026-05-26 16:39:55 UTC

Excited to speak at the Oslo Freedom Forum put on by @nprofile…rzcm next week, running a workshop called Freedom AI 101. Come learn AI tools that can be used in any environment. Even if you already use AI daily, still stop by. There will be something for everyone!
https://blossom.primal.net/b6287bcd5763e2b5a01bfdea2d1a54e724ad776eec52d4dda69386f3c93801ce.png
https://oslofreedomforum.com/event/oslo-freedom-forum-2026/freedom-tech-at-the-oslo-freedom-forum/

2026-05-15 21:47:43 UTC

- reply

Never forget that OpenAI is run by the guy behind Worldcoin. They want all your financial, health, and personal data, linked to your eyeball scan.
You can’t make this stuff up. Oh wait, Minority Report already did.
#nevent1q…dh88

2026-05-12 19:11:37 UTC

- reply

Hit up our support and we'll help you out.

2026-05-12 15:12:46 UTC

- reply

Trusted Execution Environments use hardware encryption that provides cryptographic attestation to the code it is running.
1. look at our open source code on github to verify how data is handled
2. build the github code, grab the hash, and verify it matches with the TEE attestation
https://www.trymaple.ai/proof

2026-05-10 03:23:58 UTC

At the grocery store with all the other bros prepping for mother’s day dinner.

2026-05-09 14:27:31 UTC

I wouldn’t use my last name
https://blossom.primal.net/1207ba9595c42fe266d45a725973f81925f0ba1ddf403d4c6df54faf43a60702.jpg

2026-05-08 18:42:40 UTC

- reply

i know we miss all the hard work in between these posts, but it's difficult to imagine a better place to make a career

2026-05-08 15:23:54 UTC

- reply

If you're using it with something like OpenClaw, the tool can get aggressive with token usage. When people use Maple through the app, it's much more reasonable.
Hit up our support email and we can help you out.

2026-05-08 13:46:41 UTC

- reply

😆😆😆

2026-05-08 13:41:17 UTC

- reply

the way TEEs work is that you can cryptographically match the open source code with the code on the server.
you need open source code and reproducible builds though.

2026-05-08 13:39:53 UTC

- reply

recent addition, yeah. a vast majority of their traffic does not route through it. and their code still isn’t open source.

2026-05-08 13:38:42 UTC

- reply

same thing applies to them. Swap their names with Venice.

2026-05-08 02:55:15 UTC

- reply

thank you for being a subscriber! we will keep working on getting the right balance so users get a good experience.

2026-05-08 02:54:21 UTC

- reply

TEE overhead for inference is negligible. It's the fact that it takes multiple top-of-the-line nvidia gpus chanined together to run a single large model. The models in Maple are full size, not quantized.

2026-05-08 02:47:17 UTC

- reply

Some companies lose money in order to get users. We want to run a sustainable business, so we set limits where users get a good service and we are able to keep the lights on.

2026-05-08 00:06:08 UTC

- reply

middle out

2026-05-08 00:03:45 UTC

- reply

Venice can see your AI data, they just publicly say that they won't look at it. Maple's code is open source so people can verify that chats are encrypted locally first and the privacy is enforced by hardware encryption on the servers.
#nevent1q…5exe

2026-05-08 00:03:34 UTC

- reply

2026-05-07 17:52:17 UTC

- reply

hosted solution. goal is to make it easy for anyone to use, similar to what Signal did for messages

2026-05-07 15:22:28 UTC

- reply

Working with Skyler and team was great. They walked through the process and got down to the essence of what we wanted, and what we didn’t even know we wanted. Can’t recommend them enough!
Best thing we got out of this: confidence
#nevent1q…6v57

2026-05-07 15:21:04 UTC

- reply

try different models within the app. we don't add any censorship on top of the open-weight models.

2026-05-06 20:44:12 UTC

- reply

I spent 6 years at Apple working on cool AI and privacy tools. Then I saw that popular AI apps couldn't be used for anything sensitive. So I left and helped build an AI people can trust with their real life.
We've come a long way since our humble launch in Jan 2025.
#nevent1q…6t3x

2026-05-01 21:41:48 UTC

- reply

Hmm, did not see an email. Please send to support at trymaple

2026-05-01 01:27:15 UTC

- reply

We keep iterating. Goal is for the interface to be simple to use and not require extra cognitive load to figure out. Even little things matter, so we continue to welcome feedback.
Here's the latest.
#nevent1q…2y4j

2026-04-20 21:11:49 UTC

- reply

count me in. There are 4 of us. Maybe even dozens!

2026-04-19 15:54:35 UTC

- reply

We are actively iterating on the new design, so share your feedback on adjustments you'd like to see
#nevent1q…h0xv

2026-04-15 15:18:20 UTC

- reply

yeah, many of them meet on Sundays for 1-2 hours

2026-04-15 15:17:03 UTC

Whenever Claude goes down, it's a reminder that AGI isn't here

2026-04-09 15:55:06 UTC

- reply

We have been planning this feature for a while. It got put on hold as we worked on another major initiative (more info on that soon).
I'm super stoked for this update because I use Maple every day as my first AI. Being able to use projects with shared context has been a big productivity unlock, both professionally and personally. I know some people like the ability to organize their sidebar, I get that. But for me, the custom instructions per project are the massive feature.
Eventually we want to add shared files as well, as part of a bigger document repository feature.
Maple updates were fast and furious during 2025 and slowed over the last three months. We weren't sitting on our hands. Expect to see regular updates again.
#naddr1qv…ymaj

2026-04-09 03:41:35 UTC

I have an open PR with OpenCode that adds @nprofile…uh5t as an AI provider with privacy. We added GLM 5.1 tonight, so I'm using it to update the PR to latest from upstream. Maybe we can finally get it merged. 🤞 https://blossom.primal.net/c1097f567eaa443894675deda041df95efc65f32206af526bd9887fd2d722e1e.mov

2026-04-06 16:34:01 UTC

- reply

We are exploring Gemma 4 as one of the default models, probably replacing GPT-OSS-120b as the "Quick" model.
Try it out yourself and let me know which one you like better for quick responses: Gemma 4 vs GPT-OSS
#nevent1q…yhj9

2026-04-02 03:36:36 UTC

- reply

🫡

2026-04-01 17:51:47 UTC

- reply

Good real estate photos make a huge difference. Be reliable with showing up when you say you will and you will have all the clients you want.

2026-04-01 17:46:35 UTC

https://blossom.primal.net/5db311357f6dd656598648396ce2c1c628e6e247dfd77bbeb5fb5504291e9112.jpg

2026-04-01 15:54:24 UTC

- reply

...online. it's the dumbest thing on the internet. at home it can be incredibly funny. my kid got me good this morning with a giant fake spider.

2026-04-01 15:53:12 UTC

When you work in security and privacy, it's easy to come across as a doomer. I'm optimistic about the future of people empowered with AI tools that enable them to build the world they want to see.
Everyone is gaining everyone else's skillset, which means better builders!

2026-03-29 21:29:44 UTC

This is for gen alpha
https://blossom.primal.net/1529c9b0af665aac41df66218881efb7becd11b2810ceeb5a675aded83a26422.jpg

2026-03-29 15:02:38 UTC

The new Maple app we're building can help you use technology less and experience the world around you more. How does that sound to you?
I traveled with it over the last two weeks, and it quickly became my interface for a multitude of tasks. Looking up things, research around town, capturing thoughts, public transportation, real-time sports scores, language translation, sightseeing, reminding me of scheduled items, small therapy sessions, keeping track of recommendations, to-do list items, figuring out restaurant menu items, journaling, understanding local traditions, and even finding a bathroom from a photo I took ("I'm right here, where's the nearest public toilet?!").
There were at least 10 apps it replaced. For the ones it didn't, I hit it first before going to the other app. Equipping me with personalized research meant I was fast at using the other app. Got in, got out.
Another outcome is that I had fewer moments of getting distracted by the algorithms each app uses to keep you engaged with them.
And yes, you can interpret this post as, "Use technology less by using our app more." 😅 I think the goal should be decreasing wasted time in apps overall. If one app reduces how much time you spend on your phone, that seems like a benefit. Just me?

2026-03-28 11:48:23 UTC

- reply

2026-03-28 11:12:34 UTC

Saturday morning vibes in Bedford
https://blossom.primal.net/1cd0ced0ca442b6cc84b9011ffbd4e6324635ae64535b657a5a3081554bdce79.jpg
https://blossom.primal.net/e33c51d76f67109430af0991ca09735d691dd714c5578bcf0c83467103128aa7.jpg
https://blossom.primal.net/9d3f2a34cd56463c1001300f2b8833ae14f4cb0908295c3916bd48c26c6c1608.jpg

2026-03-28 02:42:00 UTC

- reply

Famous last words ever posted to nostr

2026-03-27 09:15:44 UTC

It’s event day in Bedford 🤘
https://blossom.primal.net/a24fc13373ab2a423e0ae0041b34a214ca1fe73b5a0632209742a08944e46b06.jpg
https://blossom.primal.net/642eaa23da7b5055e3ce1037d0244527e75e5106bfcbef878940e5b12bb7f57c.jpg
https://blossom.primal.net/1d09f89d97cc6b14273b6c0899d78b9c388e802c0e8af8a817bf809567ea43f1.jpg

2026-03-24 13:26:51 UTC

Delete a couple of apps from your phone today. I bet it will make you feel a little better.

2026-03-23 19:33:58 UTC

- reply

lol that’s so messed up. they should have at least claimed the need to protect trade secrets, not a dumb thing like that

2026-03-23 17:34:34 UTC

pov your job is safe from ai
https://blossom.primal.net/b589501e907877386256e81314066b7fe6f2471384341837be023f4996602e7c.mov

2026-03-23 16:31:06 UTC

You can’t put bitcoin on a gate
https://blossom.primal.net/1a861aba65601856dc25b2ea793db38e1f42bf82491cc00cd014105f84ce04f2.jpg

2026-03-23 15:57:29 UTC

- reply

you’ll never know unless you just submit an offer

2026-03-23 15:51:00 UTC

A little reclaimed wood shiplap and grey paint would do wonders for this fixer upper
https://blossom.primal.net/61a2675d5e578ebe9f6be4977bca4d329d41ab15e14ce685397e8cef07a65bb5.jpg

2026-03-23 14:54:07 UTC

Half a billion people played Pokémon Go. They didn't know they were building centimeter accurate maps for the military.
New video breaks down three stories about visual surveillance:
1. Pokémon Go's pivot to military tech
2. Meta's glasses capturing private moments
3. Ring forcing users to choose: privacy OR features
When your data is viewable, it can be repurposed.
https://blossom.primal.net/76eb7f04da251ceabfd7e817d0a0e8d5f4a4741322a855209b425313250ca9e3.mov

2026-03-20 23:59:47 UTC

- reply

Politicians are sounding the alarm about AI privacy. The message is true, but they have their own motives for caring.
Meanwhile, we've been building private AI for years and will continue to fight for the freedoms of everyday people.
You can add privacy to your AI toolbox.👇
#nevent1q…arqt

2026-03-17 00:58:54 UTC

- reply

🤘🤘🤘
#nevent1q…veg0

2026-03-16 16:44:11 UTC

Starting the week, you ask your AI Agent to give you critical pushback on the discussion you had over the weekend. Agent's response?
"I'm drawing a blank, what were we talking about?"
bro has a case of the mondays hard

2026-03-16 00:53:58 UTC

Send your own bulk emails through Google Workspaces using gMassMailer.
Helpful for any person or AI agent operating business with a Google Workspaces account.
This weekend a friend needed help sending out emails to clients, so I built a tool for it.
https://github.com/marksftw/gmassmailer

2026-03-13 18:59:52 UTC

Outdated
https://blossom.primal.net/5f0bc59c420817c31b162bf81b5842d426cd16db9db841c883b422776f83f86a.jpg

2026-03-13 16:52:22 UTC

- reply

for sure. hence the "cheap" word in the statement. Even the likes of ChatGPT and Claude monetize user data to subsidize the monetary price users pay.

2026-03-13 15:33:24 UTC

There will always be companies willing to provide cheap or free services to people willing to give up their dignity.

2026-03-06 00:23:45 UTC

- reply

thanks for the feedback. will fix

2026-03-05 23:03:16 UTC

Microsoft Azure portal is so difficult to use, I don't know how anything ever got configured before AI was able to walk people through the steps.

2026-03-05 22:46:43 UTC

Apple wins the AI race https://blossom.primal.net/35444d169b088d51719f854f2bc17463646f68db8d98d84291986fb48bf0fc97.mov

2026-03-05 18:40:57 UTC

It's surprising to me how surprised people are by this. If you wear glasses with cameras and mics, streaming the data to a company that openly admits it can see your data and monetize it, your personal life will be exposed. https://blossom.primal.net/f00bd21d2f36b8b8f8776e4b36629762238ea5fcf698b70cb6d85b87a8801399.png

2026-03-05 04:30:47 UTC

- reply

You too can take crazy ideas from a podcast and have your agent build it for you.
No longer do you have to say, "That would be cool if someone made that!"
#nevent1q…lea5

2026-03-05 04:27:58 UTC

You have your OpenClaw doing your taxes, I have mine searching for a high-demand toy for my kid. These tools are neat-o.

2026-03-04 18:08:27 UTC

- reply

In the works... waiting on certs from the evil empire
#nevent1q…0p5q

2026-03-04 17:24:43 UTC

- reply

you or the clanker

2026-03-04 17:22:47 UTC

- reply

🤘
#nevent1q…pssq

2026-03-04 17:21:16 UTC

- reply

what is this...did your bot discover it had been mentioned on the show?

2026-03-04 17:20:12 UTC

- reply

longer answer: recurring bitcoin payments have poor UX. waiting for a good solution to that. our users tell us repeatedly that they like paying for Maple and knowing that it is there for them whenever they need it.

2026-03-04 17:18:44 UTC

- reply

spend and replace?

2026-03-03 17:59:16 UTC

- reply

Data continues to be the commodity, only now it is more dense and valuable. AI data contains thought processes, hopes & dreams, despair, and detailed sensitive data that can render a person vulnerable in a myriad of ways.
It's no surprise that companies like Anthropic & OpenAI launched as data collection companies. And it will be no surprise when a "Plaid for AI" comes to market. That company is quite possibly already here, operating under a different MO.
#nevent1q…xz0e

2026-03-03 16:15:37 UTC

- reply

AI Agents remind me of the early days of the internet. We’re all discovering email and HTML for the first time. Super exciting and unknown!
This chat with @nprofile…0ajq is a good listen for people just dipping their toe in or who have already dived head first into the pool.
#nevent1q…9lyh

2026-02-28 14:51:56 UTC

I feel socially obligated to share this chart
https://blossom.primal.net/8d1bed7baa8a06b5037fdd404010d3d44afa829f6222cf6550c001d66fd9f212.jpg

2026-02-28 14:45:54 UTC

- reply

🫡 #nevent1q…anwl

2026-02-28 14:44:23 UTC

- reply

you can tell OpenClaw to set up Maple and use it for sub-agent tasks. We are still working on a solution that allows it to power the entire thing.

2026-02-28 14:41:42 UTC

- reply

I’m a fan of local, sovereign computing. We don’t know which $20K hardware config will run the local models that are truly powerful. Thus the “one-time cost” today might be obsolete when that time comes.

2026-02-27 18:48:08 UTC

- reply

not in the immediate plans. local models just aren't good enough and likely won't be for 1-3 years. most people don't want to spend the money to buy and maintain $20K hardware annually.
you should check out what @nprofile…hckk and @nprofile…fu08 are each working on if you want to go the route of a local mini machine.

2026-02-27 18:44:55 UTC

- reply

The @nprofile…uh5t client generates a unique encryption key per user. That key is used to encrypt data throughout the process so we can't see your data. Our code is open-source and uses confidential computing to provide proof of the code running on our servers.
To my understanding, Venice is a proxy that passes your data through their servers, whether it's to open models or ChatGPT/Claude. Users can't verify the code running on their servers, so there aren't cryptographic proofs of privacy like Maple provides.
#nevent1q…u7cx

2026-02-27 18:44:49 UTC

- reply

2026-02-26 23:26:56 UTC

- reply

Zap this man if you like having AI privacy on @nprofile…uvay 🫡
#nevent1q…ekxx

2026-02-26 23:23:44 UTC

- reply

fwiw, open source is essential to the security model of @nprofile…uh5t
#nevent1q…5sun

2026-02-26 23:23:29 UTC

- reply

fwiw, open source is essential to the security model of @nprofile…uh5t

2026-02-26 23:21:31 UTC

- reply

We aim to bring as much openness to the AI pipeline as possible. Getting @nprofile…uh5t on @nprofile…uvay is an important part of that. We wanted users to have a way to get a build of the app that is cryptographically signed by a key they can verify through Web of Trust.
It took a bit to get here due to some CI/CD feedback we had, and I'm grateful that @nprofile…qyxr was up to the challenge. He was super responsive and got things working.
If you're in a zapping ⚡️ mood, don't zap me, zap @nprofile…qyxr for all the work he's doing to bring sovereign app distribution to a world dominated by closed systems. "Zap" is in his name, after all!
🤘
#nevent1q…cjxp

2026-02-23 03:51:41 UTC

- reply

special thanks to a community member for getting this working. Coming soon, hopefully, to a v0.40 Start9 near you.

2026-02-23 03:50:04 UTC

OpenCode on my Mac talking to the @nprofile…uh5t Proxy running on my @nprofile…fu08 server.
sovereignty-intensifies.gif
https://blossom.primal.net/7f080a08b4658412eb114dd49acaf41985aaf414b5366792a0dbb8f1fb2d9296.mov

2026-02-21 23:35:44 UTC

- reply

do you mean you want Maple as the LLM powering OpenClaw? If so, we are doing internal testing this weekend and should have an official announcement soon!

2026-02-21 23:18:41 UTC

- reply

👀
#nevent1q…skk9

2026-02-21 22:35:13 UTC

- reply

that moment when you respond to your own nostr note thinking it’s your OpenClaw DM because the note is a screenshot of your DM 🥴
#nevent1q…957u

2026-02-21 22:32:21 UTC

With Claude having an outage, this internal GLM-5 on @nprofile…m72d is coming in clutch!
https://blossom.primal.net/ee0d26d63176dd388f695629481d99ed8c50b2ede398eb9cf73a2b3c3bbaef96.jpg

2026-02-21 22:30:44 UTC

- reply

great meeting you as well. appreciate all the questions!

2026-02-21 22:30:08 UTC

- reply

thanks for the feedback. will improve

2026-02-21 22:29:48 UTC

- reply

Freedom tech 🤘
“You should now have a totally private AI assistant running alongside your local-first markdown vault synced over Nostr.”
#nevent1q…qajj

2026-02-21 22:27:50 UTC

- reply

not yet. if you’re a github user, feel free to post an Issue with details https://github.com/OpenSecretCloud/Maple

2026-02-21 22:26:37 UTC

- reply

This will get better soon 🫡
#nevent1q…cux7

2026-02-21 22:03:19 UTC

- reply

💜
#nevent1q…gsg8

2026-02-21 22:02:59 UTC

- reply

💜💜💜