Hey Tohju community! It’s been a busy few months at tohju.com, and we’re excited to share what’s new. We’ve focused on making your AI conversations smoother, more collaborative, and packed with practical features. Whether you’re using our web app, Telegram bots, or diving into e-commerce integrations, these updates are designed to help you get more done with less hassle. Here’s a roundup of the key changes—grouped by category so you can skim what’s most relevant to you.
UI and Mobile Improvements
We’ve polished the app to feel more intuitive on every device. Mobile users will notice the side navigation now auto-closes after clicks, fixing those annoying overlaps. The “New Model” and payment top-up pages have been optimized for smaller screens, with cleaner layouts and better touch interactions. We’ve also fixed the thread footer getting blocked by the “jump to end” button, and the textarea positioning now stays put during typing. For a more personalized vibe, your selected dark mode, theme colors, sidebar state, and default model preferences are now saved across sessions.
Chat and Conversation Enhancements
Chatting with AI just got smarter and more organized. You can now pin important messages to build a personal knowledge base right in your conversations, and these pins are even included in prompts sent to the LLM for better context. We’ve added a conversation minimap (like in ChatGPT) with dots for your messages, pagination for easy browsing, and highlighted pins—tap an avatar to jump between them. Drag-and-drop file uploads are live, making it effortless to add images or documents mid-chat. The send message widget now supports image and audio uploads too. For longer threads, we’ve introduced collapsed views, query for older messages, and pagination on scroll-up or via a “see previous messages” button. Shared sessions have been fixed and improved: you can now include (or exclude) conversation messages and uploaded documents when sharing, and there’s a switch for granular control.
Agent and Model Features
Agents are at the heart of Tohju, and we’ve made them more powerful. After uploading a document, it’s automatically added to your agent’s context—no manual steps needed. We’ve improved image generation agents with clearer model names and better error displays for any issues. The models menu on the Tohju bot is back in action, and we’ve enabled memory and RAG (Retrieval-Augmented Generation) for more accurate responses. HITL (Human-in-the-Loop) flows are smoother, with chat history now properly influencing AI decisions, and user original messages included in responses. We’ve added new models like GPT-5.1 and Kimi 2.5, with announcements in the app. For sharing, you can now include conversation messages when sharing an agent.
Integrations and Onboarding
Telegram, Slack (Closed Beta) and WhatsApp (closed beta) integrations are live, letting you connect bots to sessions for seamless messaging. When linking Telegram, you can now create new models (image, video, text) right in the flow. The onboarding wizard has been refined: select existing agents or create custom ones, with sessions shared automatically. Shared sessions as x402 endpoints are enabled, tying into our agent micropayments work. For payments, we’ve rewritten our Crypto payment, replaced external widgets with our in-house one, and fixed top-up links and codes. The blog article page now keeps images within the viewport too.
Usage, Credits, and Monetization
We’ve added a Usage and Settings page showing your execution history and credit top-ups—grouping reports for better insights. A “No Credit” component guides you if you’re low, and we’ve enabled a free tier where you can watch videos or complete tasks for more points. Ads are now integrated into Tohju, with a dashboard connecting to the latest money-making posts. Flipcash submission is complete, including a logo redesign. We’ve also connected the latest “money-making incredible” posts to your dashboard for inspiration.
Bug Fixes and Reliability
We’ve squashed several bugs to keep things running smoothly: multiple images/responses no longer duplicate, wrong inputs to Replicate are fixed (no more multiples), the gallery view in chats is restored, and markers now show for attachments in prompts (included or stripped). Shared chat sessions are fully operational again, and the recent sessions list no longer shows “no sessions” errors. We’ve also turned off unnecessary info logs in production for cleaner performance.
These updates build on our core features like client-side databases for local SQL queries on chat history, message pinning for knowledge bases, shared sessions with access controls, smart image galleries, and real-time sync. We’re shipping fast to make AI conversations truly useful—whether for personal productivity, team collab, or e-commerce.
What’s your favorite new feature? Drop a comment or hit us up on @tohjuapp. More coming soon—stay tuned!
