home.social

Search

467 results for “shop”

  1. The Erica Crooks Halloween Bizarre Broadcast 2026 (Puppet Comedy Halloween Short Film)

    (As seen on Heather Henson's Handmade Puppet Dreams: 48hr Halloween Puppet Film Project 2026)
    (*Award Winner for Best Editing*)

    For more satirical puppetry independent film from Erica Crooks:
    linktr.ee/officialericcrooks (Socials, Patreon & Shop)

    #puppetry #shortfilm #ericacrooks #puppetcomedy #halloween #slapstickcomedy #parodycomedy #darkcomedy #puppetfilm #indiefilm #satirecomedy #independentfilm

  2. Israeli soldiers beat elderly Palestinian in Hebron | Israel-Palestine conflict News

    Israeli soldiers beat and injured Hijazi Shawer, a 70-year-old Palestinian shopkeeper and community leader, during a raid in…
    #EuropeSays #Headlines #News #Europe #Europa #EU #Israel #Conflict #genocide #Israel-Palestineconflict #MiddleEast #Military #Newsfeed #OccupiedEastJerusalem #OccupiedWestBank #Palestine #ShowTypes
    europesays.com/iran/341347/

  3. Popular Italian deli La Fetta opens new ship in Prairie Village

    Less than a year after opening in southern Overland Park, popular Italian sandwich shop La Fetta Delicatessen has expanded its reach with a new spot a little further north. The deli is in the recently updated Corin…
    #dining #cooking #diet #food #Italianfood #Italia #Italian #ItalianFood #italiano #italy #ShawneeMissionArea #top
    diningandcooking.com/2871679/p

  4. I have added this new print to my collection: catherinebabault.com/fine-art- If you are starting your shopping early for the holiday season, feel free to check my website for prints, photo books, gift cards and private photo tours on Vancouver Island. Thank you for supporting small businesses. 😊

    #photography #VancouverIsland #BritishColumbia #wildlife #bears #SupportSmallBusinesses #SelfPromotion

  5. The Guardian view on Oxford Street’s pedestrianisation: cities should be designed around people, not cars | Editorial
    By Editorial

    The transformation of Europe’s busiest shopping street should be a case study in making cities greener, less congested and more inviting places to live

    theguardian.com/commentisfree/

    #Environment #Pedestrianisation #Cities #London #SadiqKhan #Politics #UKnews #Pollution #Children #Airpollution #TheGuardian #Editorial

  6. FYI: Google Ads lets retailers multiply new customer values up to 5x: Multipliers from 1.1x to 5x join flat amounts in Search, Shopping, PMax and Demand Gen this month. What happens to a $20 bonus on a $50 cart versus $1,000? ppc.land/google-ads-lets-retai #GoogleAds #DigitalMarketing #Ecommerce #CustomerAcquisition #OnlineRetail

  7. Home Depot swears its digital tags won't do dynamic pricing, except for commodities and bulk buyers. Retailers keep issuing this denial because legislators and shoppers can read what the hardware enables.
    businessinsider.com/home-depot

  8. RE: fosstodon.org/@freewear/117501

    How can I "choose"? You just send a picture but no link to a shop.

  9. FYI: Walmart lets AI agents buy within customer-set limits such as $50 or $100: Agent Profiles arrive within a month and split agent orders from shoppers' own. Agents never see card numbers, so who answers when a bot buys the wrong thing? ppc.land/walmart-lets-ai-agent #Walmart #AIAgents #Ecommerce #OnlineShopping #CustomerService

  10. 💥Announcement! Monday 26.10.2026💥

    🎃 🍂 Cozy Herbst Socializing with ORT 🍂🎃

    Monday, 26.10.2026 | 6.00 pm | Lunte Neighbourhood Shop, Weisestraße 53, 12049 Berlin

    Directions: U8 Boddinstraße | Bus 166, M43 Herrfurthstraße

    📣 Call to action: demo-ticker-berlin.org/?p=10007 - @ortberlin

    #b2610 #ORTBerlin

    On Monday, there’ll be a cosy autumn get-together with ORT.

    Come along and get to know us over some punch and pumpkin soup.

    We’ll also be painting banners together :).

    Tuesday at 6 pm at the Museum of Capitalism ORT

  11. On tap for today: photographing all of the styles of earrings, bracelets, and rings I have available for sale. And then updating my online shop to include them all, with accurate inventory numbers. This is incredibly tedious work, but I'm pretty sure I can finish it all today. That'll leave me open for more interesting work for the rest of the week. (1/2)

  12. This is a submission for the Hacktoberfest Open-Source AI Challenge Week 2

    Orderly: An AI WhatsApp Order-Taker That Refuses to Guess
    Small shops often receive customer orders through WhatsApp like:

    “2 kg rice, 1 litre oil, and 1 kg sugar.”

    But real messages can be much messier — Telugu + English, corrections, unclear quantities, or voice notes.

    The difficult part isn't just understanding the message.

    The real problem is knowing whether the AI understood the complete order correctly.

    That's why I built Orderly.

    What I Built
    Orderly is an AI-assisted WhatsApp order-taker for small shops.

    It converts messy text and voice orders into structured orders and checks them before confirmation.

    The core principle is:

    AI extracts. Code verifies. Catalog decides. Human confirms when evidence is insufficient.

    If the order is clear and the evidence is sufficient, Orderly can mark it CONFIRMED.

    If something is unclear, missing, incomplete, or unsafe to verify, it returns NEEDS_REVIEW instead of guessing.

    Demo
    🎥 Watch the Orderly Demo Video

    The demo shows Orderly processing customer orders, validating extracted items, and handling cases where confirmation should not be automatic.

    Video Transcript
    Hi, this is Orderly, an AI-powered WhatsApp order-taker for small shops.

    It handles messy text and voice orders, including Telugu-English messages.

    The key principle is:

    AI extracts. Code verifies. Catalog decides. Human confirms when evidence is insufficient.

    Orderly checks the extracted items against the catalog and validates the customer's request.

    If the evidence is sufficient, the order can be CONFIRMED.

    If anything is unclear or incomplete, Orderly returns NEEDS_REVIEW instead of guessing.

    The goal is simple:

    Never falsely confirm an incomplete order.

    Try Orderly, explore the code, and share your feedback or ideas for improving safe AI-powered ordering for small shops.

    Code
    🔗 GitHub Repository

    Orderly is built as a small Python application with:

    Web interface
    CLI
    Deterministic validation
    Catalog checks
    Automated safety tests
    Fake extractors for deterministic adversarial testing
    How I Built It
    The pipeline is:

    Customer message → transcription → AI extraction → deterministic validation → catalog verification → confirmation/review

    For voice orders, speech is transcribed before extraction.

    The AI produces structured order information, but the application does not treat the AI response as proof that the extraction is complete or correct.

    The deterministic code checks the extracted information against the customer's message and the shop catalog.

    This separation is intentional.

    An LLM can misunderstand a message, hallucinate an item, or simply miss something.

    So the system is designed so that:

    AI proposes. Code verifies. Evidence determines whether confirmation is allowed.

    Safety First
    One of the biggest risks I focused on was customer-item omission.

    For example, if a customer says:

    “2 kg rice and 5 litres sunflower oil”

    but the AI extracts only:

    “2 kg rice”

    a normal pipeline might accidentally confirm the incomplete order.

    Orderly is designed to treat this as a NEEDS_REVIEW situation rather than assuming that the extracted list is complete.

    I specifically tested adversarial cases involving:

    Multiple products with one item omitted
    Three products with one or more items omitted
    Middle-item omission
    Final-item omission
    The same product with different sizes
    Corrections
    Mixed Telugu + English orders
    “And also” constructions
    Long messages containing unrelated text
    The goal is not to maximize the number of CONFIRMED orders.

    The goal is to avoid falsely confirming an incomplete order.

    Why Open Innovation Matters
    This project uses open-source AI because useful AI systems should not always require a closed cloud service.

    Open models and open-source tools make it possible for developers to experiment, inspect the pipeline, run models locally, and build systems for specific real-world problems.

    For a small-shop use case, local and open tooling can also provide more control over how customer data is processed.

    The important lesson for me is that open AI becomes more useful when it is combined with traditional deterministic software rather than being trusted blindly.

    My Agent Session
    I used AI-assisted development to inspect the project, improve the implementation, create adversarial safety tests, investigate edge cases, and verify the application.

    The development process focused heavily on finding cases where an AI extraction could appear correct while still being incomplete.

    That led to a stronger design principle:

    Never confuse a successful AI response with proof that the customer's complete request was understood.

    Testing
    The automated test suite contains 156 tests: 155 passed, 0 failed, and 1 expected failure (xfail).

    Orderly automated test suite showing 156 tests collected, with 155 tests passed, 0 failed, and 1 expected failure (xfail). The terminal output demonstrates the project's deterministic safety, validation, and adversarial testing results.

    The expected failure is intentional. It documents a known limitation involving an unknown product such as bellam when the extractor completely omits that product and the customer message does not provide enough recognizable evidence for the completeness checker to detect the omission.

    The limitation is documented rather than hidden.

    The tests are primarily deterministic safety and validation tests. They use controlled/fake extraction behavior to test whether the application fails safely when extraction is incomplete.

    Real-Gemma validation was not performed in the current environment, so these test results should not be interpreted as evidence of real-model extraction quality.

    Current Limitations
    Orderly is a prototype, not a production-ready autonomous ordering system.

    Completeness is difficult to guarantee perfectly for:

    Completely unknown products
    Ambiguous language
    Speech-recognition errors
    Unusual phrasing
    Product names appearing in unrelated text
    Corrections that are difficult to resolve
    Messages where the customer's intent cannot be reliably identified
    A particularly important limitation is that a completeness checker cannot mathematically prove that an LLM extracted every possible customer intent from arbitrary natural language.

    Therefore, the safest behavior is to request human confirmation whenever sufficient evidence is unavailable.

    Because of this documented limitation, Orderly should not be considered production-safe for unattended autonomous ordering.

    Final Thoughts
    Orderly started as an AI order-taking idea, but the more important lesson became verification.

    AI is good at understanding messy human language.

    Code is better at enforcing deterministic rules.

    Putting the two together creates a safer system than relying on either one alone.

    AI extracts. Code verifies. Catalog decides. Human confirms when evidence is insufficient.

    The most important design decision was not making the AI say CONFIRMED.

    It was making sure that when the system cannot establish that the customer's complete request is represented, it refuses to guess.

    Try Orderly, explore the code, and share your feedback or ideas for making AI-powered ordering safer for small shops.

    devchallenge #weekendchallenge #hf26challenge #opensource #ai #python#devchallenge #hf26challenge #software #coding #development #engineering #inclusive #community
    Orderly: An AI WhatsApp Order-Taker That Refuses to Guess

  13. europesays.com/ie/750508/ Watch Noel Gallagher talk Oasis reunion and “compromise” of working with Liam on ‘Chicken Shop Date’ #Britpop #Éire #Entertainment #IE #Indie #Ireland #Music

  14. 📖 Seattle's iconic 605-foot observation tower with 360-degree city views. This is Space Needle in WA.

    #Heritage #RoadTrip #HistoryLovers #Washington #la…

    Your skin deserves the same care as your itinerary
    Shop now: https://www.amazon.com/dp/B0DR98V39P?tag=jwebb004-20
    #PaidLink #Ad #AmazonAffiliate

    Space Needle

  15. Beep Boop: Alien hates doing #selfPromo so made a bot to do it instead!

    I have tons of stuff available in my Ko-fi shop, including #knitting patterns, #crossStitch patterns and more!
    ko-fi.com/alien_sunset/shop

    check out alien-sunset.neocities.org for all the cool things I do!

  16. 19 Germany Christmas Markets for Mulled Wine, Choir Music, and Gingerbread

    In Germany, Christmas markets are synonymous with the holiday season, and Deutschland is the perfect December destination for whatever holiday shopping you want to do. Whether you’re there to see architec…
    #dining #cooking #diet #food #GermanWine #Germanwine #shopping; #shops&markets #travelinspiration #Wine
    diningandcooking.com/2871382/1

  17. 🏚️ A massive 1,348-foot prehistoric effigy mound built by the Fort Ancient culture. This is Serpent Mound in OH.

    #MustSee #Heritage #PassportR…

    Freshen up on the road with this travel-friendly beauty find
    Shop now: https://www.amazon.com/dp/B0F15D5CNP?tag=jwebb004-20
    #PaidLink #Ad #AmazonAffiliate

    Serpent Mound

  18. Дівчата, чим ви чистите світле спортивне взуття? Придивляюся до моделей тут vittorossi.ua/ua/shop/zhenskie , виглядають класно, але розумію, що в місті вони чистими будуть недовго. Осінній пил швидко в'їдається. Чи є якісь дієві пінки або губки, які реально допомагають підтримувати їх у нормальному стані без щоденного прання?

  19. Source: CBC #news #novascotia

    This Nova Scotian just ran his first marathon. He trained for it by eating donairs
    David Wadden ran and biked to 59 donair shops in the #halifax area this year to eat donairs and rate them as part of a charity fundraiser. But that wasn't the hard part. After eating the donair, he faced the ungodly task of running or biking back home.
    cbc.ca/news/canada/nova-scotia

Share on Mastodon

Enter the server where you have an account.