.png&w=256&q=75)
🤝 Top Collaborators
.png&w=640&q=75)
.png&w=640&q=75)
🤓 Latest Submissions
.png&w=828&q=75)
A automated customer support
High-volume e-commerce support often faces severe bottlenecks, causing lost sales and poor customer satisfaction. The AMD Smart Customer Support & KB Routing Agent resolves this by automating initial email handling, intent routing, and product troubleshooting using high-performance AI inference on AMD Instinct GPUs. How It Works Orchestrated in n8n, the workflow listens for unread customer support emails via Gmail OAuth 2.0. Incoming messages are dispatched to GLM-5.3-Flash hosted on the AMD Developer Cloud, which extracts structured JSON metadata—identifying intent, sentiment, product SKUs, and clean query summaries. Key Capabilities Intent & Sentiment Categorization: Instantly sorts inquiries into distinct channels (Product Issues, Returns/Refunds, or General Support). Automated Knowledge Base (KB) Checks: Queries internal product documentation to answer technical questions automatically. Smart Self-Evaluation & Escalation: If the model cannot guarantee an accurate answer (UNKNOWN_ANSWER), it avoids hallucinating and gracefully routes the ticket to human agents while informing the customer. Multi-Branch Action Routing: Triggers specific workflows depending on the category, ensuring smooth operational handoffs. Technical Stack AI Model: GLM-5.3-Flash (AMD Developer Cloud API) Orchestration: n8n Workflow Automation Triggers & APIs: Gmail OAuth 2.0 & OpenAI-compatible Chat Completions Endpoint Business Impact By offloading repetitive product inquiries and enforcing strict human guardrails for complex cases, merchants reduce average first-response times from hours to seconds while cutting operational support costs significantly.
18 Oct 2026
.png&w=828&q=75)
bandisworstforme
ive been trying for so long but the damn thing jsut wont no agents are working like bruh they just stuck acp error fixed but still not working ok i dont know iam not giving up but right now this is what ill have to submit nothing because i literally cant submit ..................................................................................................................................................................................................................................................................................................................................................
6 Oct 2026
Agentic Eclipse
We are living in a time where there is a lot of video data. The tools we use to look at videos are not working well because they are slow and do a lot of extra work. They also do not really understand what is going on in the videos. Agentic Eclipse is trying to fix this problem. It is using a way of working that is, like a team of agents. This was made for the AMD Developer Hackathon. The main new idea is something we call the 'Storyboard Strategy'. This makes it easier to look at videos by choosing the important parts using OpenCV. It also helps us not waste the keys we need to use the video tools.. It makes sure the story of the video still makes sense. The part of the system that works in the background is controlled by Python and the Fireworks API. This lets users get reports that tell them what they want to know about the videos. They can even choose how they want these reports to sound.
13 Jul 2026
