Ordinary Chaos

Tag: tasks

  • CrewAI uses third-party models to automate business tasks

    CrewAI uses third-party models to automate business tasks

    [ad_1]

    Back in 2022, João Moura was directing AI engineering efforts at Clearbit, a startup creating a unified hub for business intelligence tools. There, Moura was responsible for leading the development of AI integrations, as well as defining Clearbit’s AI product roadmap.

    After a year, HubSpot acquired Clearbit, and Moura had the itch to go it alone. He’d founded startups before, including Urdog, which sold a smart collar for pets. But this go-around, Moura had a more technically ambitious concept in mind.

    Moura’s newest company, CrewAI, aims to automate repetitive, back-office tasks like summarizing reports and onboarding employees. Customers can build workflow automations using CrewAI’s platform, then deploy and track them from a dashboard.

    CrewAI doesn’t train AI models itself. Rather, the company taps models from vendors such as OpenAI and Anthropic to drive automations. Companies can build workflows on top of the apps they already use to automate things like enriching marketing databases, analyzing customer feedback, and forecasting trends.

    Moura pitches CrewAI as an alternative to robotic process automation, or RPA. RPA drives workflow automation. But it’s a much more rigid form based on “if-then” preset rules.

    “We have made it easy for teams to build groups of AI ‘agents’ to perform tasks using any model, integrate with more than a thousand different applications, and to do so in a way that protects their data privacy,” Moura said. “We encourage our customers to try multiple models and pick the models that provide the best results for specific business use cases.”

    CrewAI
    Creating automations using CrewAI’s tooling. Image Credits:CrewAI

    RPA is indeed brittle — and error-prone. A 2022 survey from Robocorp, an RPA vendor, found that of the organizations that said they’d adopted RPA, 69% experienced broken workflows at least once weekly. Entire businesses have been made out of helping enterprises manage their RPA installations and prevent them from breaking.

    Of course, AI can break, too — or rather, hallucinate and suffer from the effects of bias. Still, Moura argues that it’s a far more resilient tech than RPA.

    Investors seem to agree. CrewAI has raised $18 million across seed and Series A rounds from backers including Boldstart Ventures, Craft Ventures, Earl Grey Capital, and Insight Partners. Coursera co-founder and AI enterpreneur Andrew Ng has also invested, as has Dharmesh Shah, the co-founder and CTO of HubSpot.

    CrewAI has competition in spades. Orby, Bardeen (which also has funding from HubSpot), Tektonic, 11x.ai, Twin Labs, and Emergence are all developing similar AI-powered, business-focused workflow automation products. Traditional RPA vendors like Automation Anywhere and UiPath, meanwhile, are working to incorporate more AI tech into their tools in an effort to stay relevant.

    To its credit, CrewAI, which is currently valued at around $100 million, has managed to attract a sizeable number of customers — 150 — in its first year. (CrewAI launched in January.) And it’s angling to land more with Enterprise Cloud, a new managed subscription plan.

    Built on top of open source components CrewAI has released over the past year, Enterprise Cloud provides additional access controls and analytics to help secure and audit automations. Subscribers also get “VIP” support and templates for workflows.

    “We are seeing 100,000 groups of multi-AI executions per day across hundreds of different use cases,” Moura said. “Given our current pipeline, we could be cash-flow-positive by next summer.”

    CrewAI, which is based in San Francisco and Brazil, plans to use the cash it has raised so far to grow its 16-person workforce and expand its core automation products.

    [ad_2]

    Source link

  • Karo is a to-do app that lets you assign tasks to your friends and family

    Karo is a to-do app that lets you assign tasks to your friends and family

    [ad_1]

    You can build a reminder and task management system for yourself, and use a service that works for your team. But it might not be easy to get your family members or friends to use the same task management app. iOS app Karo (which means “do it” in Hindi) aims to solve the problem even if the other person doesn’t use the app.

    This isn’t Mustafa Yusuf‘s first task management product. The India-based developer also develops another to-do app called Tasks, for more complex task management. Yusuf told TechCrunch that he developed Karo because other apps didn’t allow him the flexibility of assigning tasks to people in your address book.

    “No app allowed me to simply delegate and track tasks I had for people in my contacts. For example, my accountant, sister, plumber, or electrician. I just want to send them a task, have them receive it on the apps they already use (WhatsApp/Messages), and get notified when they act on it. If they don’t, instead of me nagging them, I’d prefer the app to send automatic reminders,” he said.

    Yusuf said that prior to creating the app, he had to manually remind them through WhatsApp or other chat apps to complete a task. Plus, because there was no tracking system, even he forgot tasks.

    Image Credits: Karo

    The app and the features

    Karo lets you input tasks easily in an interface that looks like a conversation in a messaging app. You can use natural language to add tasks as the app recognizes identifiers, such as “tomorrow” and “10 am.” It then automatically creates a task with a deadline if you mention a specific time and date. Plus, you can @mention a particular person to assign a task to them. Alternatively, there are buttons for adding date, time, and a contact to a task as well.

    Users can optionally attach an image, a video, a PDF document or a voice note to a task. And because it’s 2024, there’s an AI angle too. Users can invoke Karo AI to break down larger tasks into smaller tasks. They can also ask AI to help them plan a trip and list to-dos.

    Once you assign a task to someone, they get a notification on Karo if they have the app. Otherwise, the app first sends them a chat on WhatsApp, and if that fails, a reminder via text message. For scheduled tasks, the person gets a reminder when the task is due as well as two and four days after the due date. For non-scheduled tasks the app sends a reminder two, four, and six days after the task has been created.

    On the main screen, tasks are sorted by contact. If you tap on someone’s name, it opens up a conversation interface with them where you can see past tasks and add more tasks. The app also has an activity tab where you can see updates about all tasks that you are involved in.

    While teams and small organizations might use dedicated solutions for task management, in a country like India, a lot of businesses don’t have a proper system in place. They prefer to handle tasks in WhatsApp groups directly. Yusuf thinks Karo’s integration with the platform and its ability to create groups within the app will appeal to WhatsApp users.

    He mentioned that some firms are already using the app for their work. For instance, a dry fruit distributor in Mumbai uses Karo to assign packaging and delivery tasks to their team. Folks on the team check off tasks through WhatsApp. Yusuf added that this could be useful for study groups or book clubs.

    Karo is available for free, but you have to pay if you want to use features like groups, themes, and attachments. The app is originally priced at $4.99 per month, $39.99 per year, and $99.99 for lifetime unlock with introductory launch offers on all plans.

    Yusuf is planning to make an Android version available to users soon. But he thinks that even if people don’t use the app, the ability to send your contacts a reminder without them being on the app is the main draw.

    [ad_2]

    Source link

  • Iron-shelled robo-snails swarm together for off-road tasks

    Iron-shelled robo-snails swarm together for off-road tasks

    [ad_1]

    Even though snails are slow and slimy-bottomed, they’ve inspired a new type of robot that could be quite useful. Groups of these bots can operate independently or join together in order to perform tasks that would otherwise be impossible.

    In recent years we’ve heard a lot about “swarm robotics.”

    This concept involves utilizing small robots that can work either on their own or as one facet of a group of identical bots. In the latter scenario, all the robots communicate with one another, coordinating their movements in order to get the job done.

    That job might entail searching for survivors at disaster sites, performing reconnaissance in hazardous environments, or even exploring the surfaces of other planets.

    For some applications, the robots have to physically link up with one another.

    Because aerial and underwater bots can move both horizontally and vertically, they’re able to form three-dimensional shapes when joined together. By contrast, because terrestrial (ground-going) robots can only move horizontally, they’re limited to forming two-dimensional shapes – thus limiting their potential uses. What’s more, most of the experimental models created so far can only move across smooth, flat surfaces.

    Things would be different, however, if there were “off-road” terrestrial bots that could climb up on top of one another, stacking themselves into three-dimensional configurations. That’s where the new snail robot comes in.

    A diagram illustrating the parallels between the White Jade snail and the robo-snail
    A diagram illustrating the parallels between the White Jade snail and the robo-snail

    The Chinese University of Hong Kong

    Developed by Da Zhao and colleagues at The Chinese University of Hong Kong, the device was inspired by the White Jade land snail. Like that gastropod, it can move itself along while doing a pretty good job at sticking to surfaces, but then stop and use suction to form a much stronger bond when needed.

    The robot has a spherical ferromagnetic iron shell, inside of which are a battery, microprocessor and other electronics. On the bottom of that shell is a set of tank-like tracks made of rubber with embedded magnets. A retractable vacuum-powered suction cup is located between the two tracks.

    An underside view of a transparent-shelled version of the robot – an array of tiny polymer stalks inside the suction cup help it adhere to rough surfaces
    An underside view of a transparent-shelled version of the robot – an array of tiny polymer stalks inside the suction cup help it adhere to rough surfaces

    The Chinese University of Hong Kong

    When the robot is moving in “free mode,” the suction cup is kept withdrawn and unpowered. The bot simply moves across both smooth and uneven terrain on its tracks, utilizing the magnets in those tracks to climb up on top of the shells of other snail robots.

    Once it’s in place up there, the bot switches to “strong mode” by lowering and powering up its suction cup. That cup then clings tightly to the shell of the other robot, holding the two bots firmly together. That said, the robot’s shell can still rotate relative to the cup, allowing it to pivot on the spot without losing suction.

    The robo-snails collaborate on climbing over a ledge
    The robo-snails collaborate on climbing over a ledge

    The Chinese University of Hong Kong

    In outdoor tests performed so far, swarms of the snail robots have worked together to perform tasks such as climbing up over ledges, making their way across gaps, and forming a single robotic arm. Although the bots were remotely controlled for these experiments, it is hoped that their descendants could one day do such things autonomously.

    You can see the snail robots in action, in the video bellow. A paper on the research was recently published in the journal Nature Communications.

    And should you be disappointed that the bots don’t look that much like real snails, check out the self-healing robo-snail that was created last year at Carnegie Mellon University.

    [Nature Communications] Snail-inspired robotic swarms

    Source: Nature Communications



    [ad_2]

    Source link

  • Hybrid underwater robot takes aim at scuba-diving tasks

    Hybrid underwater robot takes aim at scuba-diving tasks

    [ad_1]

    The uOne underwater robot may soon be taking on risky, labor-intensive tasks currently performed by human scuba divers. It boasts a unique combination of traits from two different types of undersea bots.

    When it comes to unmanned devices designed to explore the underwater world, they pretty much all fall under the heading of either ROV or AUV.

    ROVs (remotely operated vehicles) are remotely controlled in real time by a surface-located pilot, who is guided by a real-time video feed from the craft’s onboard camera.

    While ROVs are very maneuverable, they’re hampered by an umbilical communications cable running from themselves up to the surface. If that cable gets snagged on a submerged obstacle, the ROV may be permanently lost.

    AUVs (autonomous underwater vehicles) don’t need a cable, as they autonomously navigate the ocean and avoid obstacles while performing marine surveys over a period of hours or even weeks. Because of their built-for-speed torpedo-like shape, however, they can’t do much in the way of hovering in place or turning on the spot.

    That’s where the uOne is intended to come in.

    The uOne is equipped with both a forward-facing stereoscopic HD camera system and a downward-facing camera
    The uOne is equipped with both a forward-facing stereoscopic HD camera system and a downward-facing camera

    uWare Robotics

    Designed by Belgian company uWare Robotics, it’s technically an AUV in that it uses integrated cameras, an IMU (inertial measurement unit) and other technologies to autonomously follow a preprogrammed underwater “flight path.” That data is wirelessly uploaded into the craft’s onboard computer before it’s launched, by its user.

    However, the uOne isn’t shaped like a typical AUV. Its relatively short, squat body incorporates eight vectored thrusters which allow it to hover on the spot or move in any direction as needed. As it does so, there’s no cable to get in the way or get hooked on anything.

    The uOne is designed primarily for tasks like the inspection of underwater structures such as oil rig platforms and seabed pipelines. It can already make 3D point cloud maps of its surroundings, plus plans call for it to be capable of identifying and inspecting any “points of interest” it may come across.

    A point cloud map of an underwater pipeline, recorded by the uOne
    A point cloud map of an underwater pipeline, recorded by the uOne

    uWare Robotics

    All of the gathered data is stored on an onboard solid state drive, and can be wirelessly downloaded once the mission is completed and the AUV is back at its base.

    As far as basic specs go, the uOne measures 40 cm high by 40 cm wide by 50 cm long (15.7 by 15.7 by 19.7 in), weighs 15 kg (33 lb) out of the water, and can carry up to 5 kg (11 lb) of sensors, lights or other items. It has a top speed of 3 knots (5.5 km/h or 3.5 mph) and a runtime of approximately two hours per two-hour charge of its swappable battery.

    The uOne is capable of acoustically transmitting "progress reports" to its user, letting them know the mission is proceeding as planned
    The uOne is capable of acoustically transmitting “progress reports” to its user, letting them know the mission is proceeding as planned

    uWare Robotics

    With its current cast acrylic body, the AUV can descend to a maximum depth of 75 m (246 ft). An aluminum body is in the works, though, which should boost that figure much further. The whole thing can reportedly be deployed by just one person, in 10 minutes.

    Prospective clients can contact uWare via the company website. You can see the uOne in action, in the following video.

    And for another example of a robot that blurs the line between AUV and ROV, check out morphing robotic-armed Aquanaut.

    uWare & Innoceana – Autonomous Seagrass Mapping – Tenerife – October 2021

    Source: uWare Robotics via Uncrewed Systems Technology



    [ad_2]

    Source link

  • Photon mini ROV is big on performing underwater tasks

    Photon mini ROV is big on performing underwater tasks

    [ad_1]

    Canadian firm Deep Trekker is probably best known for its Revolution, a top-of-the-line underwater ROV (remotely operated vehicle). The company is now offering a much smaller, less pricey yet still very capable model, however, called the Photon.

    Designed primarily for tasks such as underwater surveying and inspection, the Photon measures just 481 mm long by 333 mm wide by 228 mm high (18.9 by 13.1 by 9 in).

    It’s compact enough to fit inside a single included Pelican case along with a swappable lithium battery, waterproof joystick controller, and 150-meter (492-ft) communications tether for linking the submersible to its surface-located operator. One battery-charge should reportedly be good for up to two hours of usage.

    The Photon can be equipped with an optional robotic grabber arm, along with other tools
    The Photon can be equipped with an optional robotic grabber arm, along with other tools

    Deep Trekker

    Constructed out of anodized aluminum, stainless steel, carbon fiber and buoyancy foam, the ROV itself weighs 11.6 kg (25.6 lb) when out of the water. It can descend to a maximum depth of 120 m (394 ft), and utilizes six magnetically coupled vectored thrusters to move in any direction while submerged – its top forward speed is about 2.5 knots (4.6 km/h or 2.9 mph).

    Video is captured by an integrated remotely tilt-able 4K camera, which is assisted by LED spotlights along with image-dewarping, clarity-enhancing, underwater-color-correcting software. The operator views a real-time feed of that video via the remote’s 7-inch sunlight-readable screen.

    The Photon comes with a controller, communications tether and custom wheeled Pelican case
    The Photon comes with a controller, communications tether and custom wheeled Pelican case

    Deep Trekker

    Should you be interested, a Base package consisting of a Photon ROV and the described gear can be ordered now via the Deep Trekker website for US$12,000. Buyers willing to pay a bit more can opt for a Nav package that includes a tool skid, robotic grabber arm, 300-m (984-ft) tether, and an enhanced positioning system.

    You can see the Photon in action, in the video below.

    Photon Micro-ROV

    Source: Deep Trekker



    [ad_2]

    Source link

  • Figure’s humanoid can now watch, learn and perform tasks autonomously

    Figure’s humanoid can now watch, learn and perform tasks autonomously

    [ad_1]

    Figure’s Brett Adcock claimed a “ChatGPT moment” for humanoid robotics on the weekend. Now, we know what he means: the robot can now watch humans doing tasks, build its own understanding of how to do them, and start doing them entirely autonomously.

    General-purpose humanoid robots will need to handle all sorts of jobs. They’ll need to understand all the tools and devices, objects, techniques and objectives we humans use to get things done, and they’ll need to be as flexible and adaptable as we are in an enormous range of dynamic working environments.

    They’re not going to be useful if they need a team of programmers telling them how to do every new job; they need to be able to watch and learn – and multimodal AIs capable of watching and interpreting video, then driving robotics to replicate what they see, have been taking revolutionary strides in recent months, as evidenced by Toyota’s incredible “large behavior model” demonstration in September.

    But Toyota is using bench-based robot arms, in a research center. Figure, like Tesla, Agility, and a growing number of other companies, is laser-focused on self-sufficient full-body humanoids that can theoretically go into any workplace and eventually learn to take over any human task. And these are not research programs, these companies want products out there in the market yesterday, starting to pay their way and get useful work done.

    Figure's 01 was walking within 12 months of development – a record, Adcock believes
    Figure’s 01 was walking within 12 months of development – a record, Adcock believes

    Figure

    Adcock told us he hoped to have the 01 robot deployed and demonstrating useful work around Figure’s own premises by the end of 2023 – and while that doesn’t seem to have transpired at this point, a watch-and-learn capability in a humanoid is indeed big news.

    The demonstration in question, mind you, is not the most Earth-shatteringly impressive task; the Figure robot is shown operating a Keurig coffee machine, with a cup already in it. It responds to a verbal command, opens the top hatch, pops a coffee pod in, closes the hatch and presses the button, and lets the guy who asked for the coffee grab the full cup out of the machine himself. Check it out:

    So yes, it’s fair to say the human and the Keurig machine are still doing some heavy lifting here – but that’s not the point. The point is, the Figure robot took 10 hours to study video, and can now do a thing by itself. It’s added a new autonomous action to its library, transferrable to any other Figure robot running on the same system via swarm learning.

    If that learning process is robust across a broad range of different tasks, then there’s no reason why we shouldn’t start seeing a new video like this every other day, as the 01 learns to do everything from peeling bananas, to putting pages in a ring binder, to screwing jar lids on and off, to using spanners, drills, angle grinders and screwdrivers.

    It shouldn’t be long before it can go find a cup in the kitchen, check that the Keurig’s plugged in and has plenty of water in it, make the damn press-button coffee, and bring it to your desk without spilling it – a complex task making use of its walking capabilities and Large Language Model AI’s ability to break things down into actionable steps.

    So don’t get hung up on the coffee; watch this space. If Figure’s robot really knows how to watch and learn now, we’re going to feel a serious jolt of acceleration in the wild frontier of commercial humanoid robotics as 2024 starts to get underway. And even if Figure is overselling its capabilities – not that any tech startup would dream of doing such a thing – it ain’t gonna be long, and there’s a couple dozen other teams manically racing to ship robots with these capabilities. This is happening.

    Make no mistake: humanoid robots stand to be an absolutely revolutionary technology once they’re deployed at scale, capable of fundamentally changing the world in ways not even Adcock and the other leaders in this field can predict. The meteoric rise of GPT and other language model AIs has made it clear that human intelligence won’t be all that special for very long, and the parallel rise of the humanoids is absolutely designed to put an end to human labor.

    Things are happening right now that would’ve been absolutely unthinkable even five years ago. We appear to be right at the tipping point of a technological and societal upheaval bigger than the agricultural or industrial revolutions, that could unlock a world of unimaginable ease and plenty, and/or possibly relegate 95% of humans to the status of zoo animals or house plants.

    How are you feeling about all this, folks? Personally, I’m a little wigged out. My eyebrows can only go so high, and they’ve been there for a good while now. I’m getting new forehead wrinkles.

    Source: Figure



    [ad_2]

    Source link