
Founder of Chirp | Leading the charge on #DePIN, the cutting-edge #blockchain frontier. Exploring the nexus of automation, innovation and technology.
Joined February 2023
- Tweets718
- Following215
- Followers3.8K
- Likes1.3K
Pinned Tweet
Great to see @ChirpDeWi’s testnet launch on @SuiNetwork featured on @CoinDesk 💪
Big thanks to @Liqquidity for the mention: coindesk.com/tech/2024/03/14…
We are already in a full-fledged AI war. Every project, in every industry, needs to take this seriously.
My message to founders, leaders, and teams is direct: pause whatever development you’re doing and make security your immediate priority.
Attackers are using AI to target infrastructure, identify weaknesses, and accelerate exploitation. Projects are using AI to monitor, defend, and respond. This war is already underway, and every organization operating connected systems has a stake in it.
We recently experienced an attack ourselves. We caught it relatively early and stabilized the situation. If there is anything to be thankful for, it is that we discovered these weaknesses now and had monitoring in place that helped us detect what was happening.
That monitoring made a critical difference.
What we’ve observed is deeply concerning. Attackers are taking their time to probe for weaknesses and identify multiple potential entry points before launching their attack. As we close vulnerabilities, we’re seeing attempts to exploit other areas. Patching the first hole is only the beginning of the response.
It brings to mind Terminator: Rise of the Machines. We are watching AI being used against AI, with real infrastructure—and increasingly the physical world—caught in the middle.
Every project should implement meaningful security monitoring immediately. You need visibility into suspicious activity, alerts that reach the right people, and a team ready to act. Monitoring cannot prevent every attack, but it can provide the early warning that lets you interrupt an attack before the damage grows. In our case, it helped us catch the attack early and stabilize the situation.
At Chirp and Kilo, we take this extremely seriously. Our work connects AI with the physical world, including connected devices, machines, and robots. When infrastructure can control physical systems, security is fundamental to everything we build.
We have already patched significant vulnerabilities and broadened the scope of our security work. This experience has accelerated our hardening efforts and moved our security posture substantially forward. Further work is still underway.
Every project needs to understand the urgency here. Whatever your industry, review your exposure, strengthen your access controls, put monitoring in place, and make sure you can respond when something happens.
We are in a full-fledged AI war. Treat security with the urgency that demands.
#physicalai #aiot #depin #lorawan #mioty #iot
🤖 Made with AI
We added full-screen dashboard sharing so teams can run a persistent operations display without leaving a personal Kilo account signed in.
The owner creates a password-protected link and chooses View or Control access. The link can be opened on a shared tablet or monitor, regenerated and revoked later.
The iFrame widget can also place an approved operational web page beside the Kilo data. For example, a shared site screen may include a weather map or another live page the team already uses.
I wanted this because a useful dashboard should be available where the work happens, including to people who do not administer the platform.
See how Chirp brings device data into shared dashboards: chirpwireless.io
#Kilo #IoTPlatform #AIoT #Dashboards
Kilo is available in English, German, French, Spanish and Portuguese.
We translated the operating interface, not only the marketing pages, so teams can work with devices, dashboards, rules and alarms in the language appropriate for their users.
I used to think of localization mainly as release work. Building for commercial deployments changed that view. A platform may be configured centrally and then used every day by people in several regions.
Adding another interface language gives those operators the same product experience without requiring the integration team to translate each instruction or alarm workflow for them.
Explore a multilingual IoT workspace with Chirp: chirpwireless.io
#Kilo #IoTPlatform #AIoT #Localization
We built several Kilo features around a practical requirement: customers need to change a deployment without rebuilding the work around it.
A device can move to another supported connector while keeping its dashboards, rules and alarms. An automation keeps version history and can be restored to a previous version. Access can be changed for a person without handing over control of the whole organization.
These are separate product areas, but they follow the same design decision. The configuration a team has already created should survive normal changes in hardware, logic and responsibility.
I include these workflows in product demonstrations because they show how Kilo behaves after the initial setup, when real deployments begin to change.
Explore device connections and automation in Chirp: chirpwireless.io
#Kilo #IoTPlatform #AIoT #ProductDesign
Synthetic Brew is the AI runtime we built behind Kilo's assistant.
Working on both products at the same time gives us a practical test for the runtime. Kilo needs the AI to use real tools, respect account permissions, show proposed changes and stop for confirmation before actions with operational consequences.
Synthetic Brew provides that execution layer. Kilo provides a real environment where we can see whether it behaves correctly across device setup, rules, alarms and supported commands.
The result is not a separate chatbot added to the platform. The assistant works through the same services and permissions as the rest of Kilo.
Try AI-assisted IoT setup in Chirp: chirpwireless.io
#AI #IoT
When we added device commands to Kilo, I did not want the platform to report success only because a message had been sent.
A device may be asleep, offline or unable to carry out the command. In those cases, successful delivery from the platform is not the same as a confirmed physical state.
So the command configuration can define which device reading should be checked and which value means the action succeeded. After sending the command, Kilo compares the reported state with that expected result.
This added more work to the feature, but it gives the operator an honest answer: the command was sent, and the expected state was either confirmed or it was not.
See how Chirp checks the result of a device action: chirpwireless.io
#Kilo #IoTPlatform #AIoT #AI #IoT
1
6
155
We added a What's New panel to Kilo so customers can see the platform work that has shipped since their last visit.
Release notes are useful, but most users do not open a changelog before starting their work. The panel brings the relevant product updates into the application and gives each item a direct explanation.
I wanted this because Kilo is changing quickly across devices, dashboards, rules, alarms, AI and administration. A feature only helps a customer after they know it exists and understand where it fits.
The panel gives us a consistent place to explain those changes inside the product rather than expecting every user to follow our social posts.
Explore the latest Chirp platform capabilities: chirpwireless.io
#Kilo #IoTPlatform #AIoT #ProductUpdates
The reason we made Kilo AI-first was to let more of the platform be configured by the people responsible for the operation.
They can describe the device or outcome they need, review the setup the assistant proposes and confirm it before the platform changes. The same deployment still has detailed controls for integrators, including device mappings, rules, alarms and access policies.
We are not removing those capabilities. We are adding a second way to use them for teams that know what they need operationally but do not want to learn every setup screen first.
That is the direction behind the work we are shipping across the platform.
Try setting up IoT in plain language with Chirp: chirpwireless.io
#Kilo #IoTPlatform #AIoT #BuildInPublic
I've released KiloCenter 2.0 🚀 the biggest update to our open-source mioty service center since launch.
mioty is the wireless IoT tech from Fraunhofer, backed by the mioty alliance. One base station handles up to 110,000 devices and around 3.5 million messages a day. KiloCenter runs your own private mioty network.
What's in 2.0:
More than 50 bug fixes across uplinks, downlinks, sessions and certificates.
A big UI update: base stations on a map, clearer and more detailed device logs, and a live view that updates within a second.
Uplinks and downlinks tested end to end on real hardware, and every command followed until the device confirms it.
And we went through the whole codebase and cleaned it up.
Three ways to run it: the free Community edition on your own servers, the Enterprise edition built into the Kilo IoT Platform in the cloud, or a licensed Enterprise edition we install on your premises.
Open source here 👇
github.com/Kiloiot/kilo-serv…
#mioty #IoT #OpenSource #physicalai
Today is a big day 🚀 We just shipped release 3.11 on both our commercial and residential platforms at the same time, and it's the biggest release we've ever done.
There are two features in it I'm really, really excited about.
Lens brings cameras onto the platform. More than 67,000 cameras support RTSP streams, and Lens connects RTSP cameras, so practically any camera out there, from smart city traffic cameras and transportation cameras to office security and residential cameras, any brand, even an old one, can now sit right next to sensors, trigger automations with motion zones and show up on the same dashboards. Video and IoT in one place! We're going to push this direction hard, and this is only the beginning.
Solutions is the other one, and it fixes the hardest part of IoT: getting started. You sign up and you have no idea what's even possible. Now you pick a ready-made solution, install it in a few clicks, and it's already running with sample devices, dashboards, rules and alarms. It tells you which devices to buy, and when they arrive, the AI chat connects them for you. That's the onboarding I always wanted.
Huge thanks to the team. Months of work went into this one, and you can feel it 🙌
#depin #iot @SuiNetwork @WalrusProtocol #physicalai
We published a white paper on what an operating system for physical AI should be.
The short version: a robot only knows what its own sensors can reach. A robot dog on a campus patrol can't know that a door just opened in a building across campus. Connect the door sensor and the robot to the same operating system, and the door opening sends the robot there, turns on the lights and alerts the guard on call.
The video is our patrol-robot demo: an AI agent runs the robot from plain-language orders, and a person decides what happens next.
Read the paper: kiloiot.io/whitepapers/physi…
#PhysicalAI #AIoT #Robotics #IoT #Kilo
We built Kilo's AI assistant to configure the platform because setup is where product knowledge matters most.
A user can describe what needs to be monitored and ask the assistant to prepare the relevant device, rule or alarm. The assistant works through the platform services, shows the proposed change and asks for confirmation before applying it.
Where testing is supported, the same workflow can use an emulated reading to show how the rule and alarm behave before physical hardware is involved.
The detailed configuration remains available for integrators. The assistant gives other teams a way to start from the operational result and review the setup Kilo proposes.
Try describing your setup to Chirp’s AI assistant: chirpwireless.io
#Kilo #IoTPlatform #AIoT #AI
We migrated CHIRP's chain-facing services from the JSON-RPC interface Sui is deprecating to its supported gRPC and GraphQL interfaces.
The work covered the Fountain Assistant, Chirp Assistant, Data Extractor, Controller, Validator and the shared core library.
We scheduled the migration before the old interface is removed so each service could be updated and verified without a forced cutover. It is the kind of engineering work customers should not notice in daily use.
I am sharing it because active platform development includes maintaining the infrastructure behind the visible features, not only adding new screens.
#Chirp #Engineering #DePIN
5
130
One part of Kilo dashboards I like showing is how the display and conditions are configured for the operation rather than imposed across the platform.
For example, a tank level can be shown as a Tube with the current fill and color bands chosen by the customer. The same data can also appear in a chart, gauge, current-value widget or another dashboard format.
The example is a tank, but the capability is broader: Kilo's widgets are configurable so the team building the dashboard can choose how each measurement should be understood.
We built those options for the people who operate from the dashboard after the integrator has finished setting it up.
#Kilo #IoTPlatform #AIoT #Dashboards
3
123
We added QR scanning to Kilo because commissioning a device should not begin with manually copying long identifiers from a label.
The scanner reads the supported device code and fills the corresponding setup fields. The user can review the values before saving and continue with the normal device and connection workflow.
We built it as part of a wider onboarding flow that also includes reusable sensor templates, device mapping and connection diagnostics. QR scanning removes one manual step; the rest of the platform still handles how that device sends data and what the data means.
For teams adding devices repeatedly, removing those repeated typing steps makes commissioning faster and reduces avoidable identifier errors.
#Kilo #IoTPlatform #AIoT #DeviceManagement
3
117
We designed Kilo's organization and access model for deployments that involve multiple sites and several groups of people.
A customer can control which parts of the platform each person may view or edit. Dashboard share links cover people who need a specific operational view without a full account. Alarm escalation decides who is notified for each severity and what happens if nobody responds.
Access decisions go through the organization service, which keeps membership and policy in one place instead of duplicating tenant rules in each feature.
I bring these capabilities into demos early because adding sites is not only a device-count problem. The platform also has to keep the right data, controls and notifications with the right people.
#Kilo #IoTPlatform #AIoT #AccessControl
5
124
We built Synthetic Brew to give AI agents an execution layer with tools, permissions and confirmation rules around the model.
That became especially important when we connected it to Kilo. Reading device history, creating a dashboard and sending a supported command do not have the same consequences, so the runtime should not treat them as the same action.
Each tool can declare what it does, the platform account still limits what the user may access, and actions that require approval stop for confirmation.
For physical commands, the surrounding platform can also verify the state reported by the device afterward. The model proposes the work; the runtime and product services control how it is carried out.
#SyntheticBrew #AI #Agents #IoT
We built a step-through debugger for Kilo rules because deploying an automation should not be the first time someone sees which path it takes.
The user supplies a test reading and runs the rule one node at a time. The canvas shows the active node, the branch selected and the point where an error occurs.
They can pause at breakpoints, change the rule and run the same input again. Real commands remain disabled unless the person testing explicitly enables them.
I like demonstrating this feature because the result is immediate: the person who wrote the rule can see whether the logic matches what they intended before activating it across a deployment.
#Kilo #IoTPlatform #AIoT #Automation
4
184
One Kilo feature I always demonstrate is the complete alarm path from a sensor reading to the phone of the person responsible.
A rule detects the condition. The alarm applies its severity and escalation chain. The notification includes the value that triggered it, and a critical alert can take over the screen through the mobile alerts app.
We built the path as one system because a chart is only one part of monitoring. The team also needs to decide who is contacted, which channel is used and what happens if the first person does not respond.
During a demo, I can trigger that sequence with an emulated reading instead of waiting for a real incident.
#Kilo #IoTPlatform #AIoT #Alarms
4
141
We changed the way Kilo formats sensor readings in alarms and other operator-facing views.
The reason was a real temperature value that appeared as `23.67545184559089°C`. The underlying data was correct, but nobody needs fourteen decimal places to decide what to do next.
Kilo now applies appropriate display precision while preserving the original measurement for processing and history. The operator sees a readable value, and rules can continue to use the stored data.
I wanted this fixed across the product rather than corrected in one alarm template. The same measurement should be presented consistently wherever a person reads it.
#Kilo #IoTPlatform #AIoT #ProductDesign
6
170



















