Adriel

32× growth in one quarter: From 10K to 320K customers across 100+ countries, with over 100 million messages processed. We are not building just another personal AI product. Our long-term vision for Teamily AI is to build the central platform of the next-generation Agentic Internet, where AI agents deliver services and perform real work directly, no longer confined to traditional web, desktop, or mobile apps distributed through app stores and search engines. Our recent growth has only strengthened my conviction in this vision. Agent-as-a-Service (AaaS) will be as accessible as connecting with friends on a social network. Just as Google democratized access to information and services in the web era, Teamily AI aims to democratize access to intelligence and services in the Agentic Internet era, making the right intelligence, people, and services equally accessible to everyone. In this letter to Teamily AI customers, ecosystem partners, and investors, we share our growth, reflect on our early foresight and what we've built, discuss how we see the evolving AI agent landscape, and outline where we're heading next.

Swap any character in your existing footage with Recast. Everything else stays the exact same. It’s almost scary. Try it now on ElevenCreative.

For the most realistic audio in your Seedance videos, stop relying on reference audio. Here’s the workflow to use instead.

🎬 MiniCPM-OCR in action — running locally on iPhone Duo and iPad. Parse fast and accurately. Guard the results. All in your hands. 🙌 Handwritten notes, complex formulas, newspapers, poor-quality scans—real documents put OCR to the test. 📄 MiniCPM-OCR 1.0 — SOTA OCR performance and speed at 1B scale. It achieves 93.68 on Real5-OmniDocBench and 54.7 on OCRBench v2, combining accurate document parsing with efficient inference. 🛡️ MiniCPM-OCR Jev — A 2B decision model that checks OCR accuracy. It approaches Qwen3.5-397B-A17B on JevOCRBench and works with outputs from different OCR parsers. Watch it compare parsed text, formulas, and tables with the original image and assign 0–1 correctness scores—helping you pick better results and identify what needs human review. 🔓 Open weights. Deploy your way. Run with llama.cpp, vLLM, SGLang, or Ollama. Customize with open-source fine-tuning code. Watch it in action. 👇 🔗 GitHub: github.com/OpenBMB/MiniCPM-O… 🤗 Hugging Face: huggingface.co/spaces/openbm… 👉 Try the Demo：platform.modelbest.ai/consol… Use the API：platform.modelbest.ai/consol…

Big scenes rarely come down to one sound, so we built SFX v3 to handle more than ever. Our latest model lets you create richer, more precise sound effects from a single text prompt, with improved music and vocals. Now available inside @ElevenCreative and ElevenAPI.

This mistake makes your content look AI. When you generate with text prompts alone, the model fills in the blanks with the patterns it was trained on, so your content ends up looking generic. Thankfully, the fix is easy. Add 2 references to each generation in Flows, a character reference and a location reference, to fill in the gaps the AI was guessing at and get more consistent, higher-quality results.

The shared context for your personal and team agents to get work done, smarter and faster. • Run on any device • Create any type of artifact • Chat anywhere • Work anytime #SharedContext #TeamAgents #TeamilyAI #iPhoneDuo

We’re giving away $100k for the catchiest AI ad. Here’s how @Fran_Gas made his, for a fictional app that catches AI. FAIKED! tells you whether an image or video is real or fake. The pitch is “seeing is believing” doesn’t hold up anymore. He built the jingle in ElevenMusic, starting from a base take and reworking it section by section. He tested variations inside the same track, changing lyrics, melody, voices, performance and structure until it landed. Next, he made reference images with ChatGPT Images, then brought them to life with Seedance 2.5 and added sound effects, all in ElevenCreative. Not sure where to start? Borrow Fran’s workflow and try it with your own fictional brand.