Have something to say?
Add native chat export functionality (.md, .pdf, .docx, .json)
Please add a native export feature to Remalt Chat that allows users to download chat conversations and generated outputs in standard file formats: Markdown (.md), PDF (.pdf), Word Document (.docx), and JSON (.json). Why this is needed Currently, saving or sharing long chat outputs requires manual copying and pasting, which often breaks formatting (such as tables, bullet points, and code blocks) or strips valuable metadata (like linked context nodes and model parameters). Adding direct export options will streamline workflows for documentation, report generation, and archival. Proposed Export Options: Markdown (.md): Clean text formatting for immediate use in Obsidian, Notion, GitHub, or documentation vaults. PDF (.pdf): A formatted transcript ready for printing, client presentation, or quick visual sharing. Word (.docx): Editable document format for drafting reports and collaborating in Microsoft Office or Google Docs. JSON (.json): Full structured export containing raw message roles, timestamps, and linked context node references for data backups or developer pipelines. Suggested Location Add an Export dropdown button directly next to the existing Copy and Branch action controls on the chat output card, or within the main conversation options menu.
Private Desktop Version
I would love to see a native Remalt desktop application for macOS and Windows, designed around local-first privacy. The main reason is simple: as Remalt becomes more useful for RAG, personal knowledge, research, documents, MCP and AI workflows, it will inevitably contain some of a user’s most valuable and sensitive information. I would like the option to keep that knowledge on my own computer rather than storing everything in the cloud. Local-first Second Brain Ideally, the desktop version could run the proposed My Second Brain / Personal RAG locally. This could include: • Documents and PDFs • Research and captured webpages • Personal notes and knowledge • Writing and project history • Embeddings/vector database • RAG indexes • Prompts and workflows • API credentials • MCP configurations • Private business information The user should be able to choose where this data is stored. For highly sensitive work, there should be a Local Only mode where the knowledge base never leaves the device unless the user explicitly sends selected context to an external AI provider. Clear privacy controls I would like Remalt to distinguish clearly between: Local data → remains on my device Synced data → encrypted and synchronised between my devices External AI data → explicitly sent to a selected model/provider This would make it much easier to understand exactly where information is going. Privacy should not depend on users simply trusting that their data is safe. The product should make data movement visible and controllable. Bring Your Own API Keys The desktop app could also support BYOK for providers such as OpenAI, Anthropic, Google and others. My API keys could be stored securely on the device using the operating system’s secure credential storage. This would allow Remalt to act as the interface and knowledge layer while users retain control over which AI provider receives each request. Local AI models Longer term, it would be valuable to support local models as well. For example, users could connect Remalt to locally running models through tools such as Ollama or other compatible local inference systems. That could enable workflows where: Documents → embeddings → retrieval → model → output all remain on the user’s computer. Cloud AI could then become optional rather than mandatory. Local MCP The desktop application could also act as a secure MCP host/client. My local Second Brain could expose selected tools and knowledge to authorised applications without making the entire knowledge base publicly accessible. Permissions should be granular: • Read knowledge • Search/retrieve • Add knowledge • Modify knowledge • Access specific collections • Access specific projects Each MCP connection should be visible, revocable and auditable. Chrome Extension + Desktop This would also work extremely well with the Remalt Chrome extension. The browser becomes the capture layer: Browse → Capture → Send to Remalt Desktop → Process local
New feature: My Second Brain 🧠 Personal RAG for Remalt
I would love to see Remalt introduce My Second Brain: a persistent personal RAG knowledge layer that users can build once and use across Remalt, external AI tools, agents and workflows. The idea is bigger than normal AI memory. Instead of repeatedly uploading the same documents, explaining context or rebuilding prompts, Remalt could become the central knowledge layer containing what I know, research, save, write and approve. 1. Personal RAG Knowledge Base Users could add: • Documents and PDFs • Notes and research • URLs and web content • YouTube/video transcripts • Previous projects and conversations • Writing samples and approved outputs • Personal knowledge and experience • Opinions, principles and established POV • Decisions and reasoning • SOPs, prompts and frameworks • Examples and anti-examples Remalt would index this content for semantic/hybrid retrieval and bring only the most relevant context into each task. It should preserve source provenance so retrieved knowledge can show where it came from, rather than becoming an unexplained AI memory. 2. Chrome Extension for Knowledge Capture The Remalt Chrome extension could become the capture layer for the Second Brain. While browsing, I should be able to save: • Full webpages • Selected text • Articles • Research • PDFs • YouTube pages/transcripts • Screenshots • Quotes • Notes added during capture Ideally I could select: Save to Second Brain → choose workspace/project → add tags/note → save Remalt could then process, chunk, index and make that source retrievable through RAG. This would turn everyday browsing and research into a continuous knowledge-building workflow rather than something I have to manually organise later. 3. MCP Access The Second Brain should expose an authenticated MCP server. This would allow MCP-compatible AI agents and tools to query my Remalt knowledge directly. For example: “Search my Second Brain for everything I have collected about AI-assisted writing.” “Find my previous thinking about this product decision.” “Retrieve the research and sources behind this argument.” “Find relevant examples from my approved writing.” MCP should ideally support granular permissions such as: Read only Search and retrieve knowledge. Read/write Allow authorised agents to create or update knowledge. Sensitive actions should remain permission-controlled. 4. API Access Please also expose the knowledge system through an API with user-generated API keys. Useful endpoints could support: • Add/update/delete sources • Search • Semantic retrieval • Metadata filtering • Retrieve source content • Create knowledge records • Manage collections/projects • Retrieve citations/provenance This would allow developers to use Remalt as the RAG layer behind their own applications, automations and AI agents. API keys should support scopes, expiration/revocation and usage visibility. 5. Knowledge Governance The Second Brain should not become an unc
Chrom plugin, update
Please add the option to add custom category and tags to the Chrome plugin, when capturing different urls/ sources for better structure and re-use
Improve Brand Architecture, AI Onboarding, Projects & Knowledge/RAG
The current Brand setup needs a clearer hierarchy and onboarding flow. Right now, users are asked to choose a writing style before AIValance has enough information about the brand. The following Brand Guidance screen then repeats similar information across multiple fields. I suggest restructuring Brand around an AI-first analysis and inheritance model. CORE HIERARCHY Workspace → Brand → Project → Whiteboard → Output Each level should inherit relevant context from the level above while allowing controlled overrides. For example, a Project inherits its Brand voice, audience, knowledge and assets, but can override the tone or add project-specific instructions without changing the master Brand. START BRAND CREATION WITH SOURCE ANALYSIS Instead of starting with “Pick the closest version of you”, first ask the user to provide sources AIValance can analyse: • Website URL • LinkedIn profile or company page • Other public URLs • Google Drive • Uploaded documents • Pasted text AIValance should crawl and analyse the available sources and automatically generate a proposed Brand Profile. The user then reviews and edits the results instead of defining everything manually before AI has any context. GENERATE A STRUCTURED BRAND PROFILE AI should extract and propose separate fields such as: • Brand Summary • Business • Positioning • Offers • Target Audiences • Audience problems / Jobs-to-be-Done • Brand personality • Tone of voice • Writing style • Speaking style • Content pillars • Messaging themes • Preferred vocabulary • Language to avoid • Signature phrases Where possible, AI-generated conclusions should show their source and confidence level so users can understand why something was inferred. MOVE VOICE CALIBRATION AFTER BRAND ANALYSIS Direct, Conversational, Educational, Storyteller, High Energy, etc. should not be mutually exclusive choices. AI should first infer the communication style and then allow users to adjust dimensions such as: • Directness • Energy • Authority • Warmth • Playfulness • Formality • Educational depth • Storytelling The existing radar visualisation could be used for this. ADD PROJECTS UNDER BRANDS A Brand should support multiple Projects. Example: Workspace → AIValance Brand → LinkedIn Campaign → Product Launch → Commercial Film Each Project should contain: • Brief and goals • Project instructions • Selected audiences • Selected offers • Project-specific knowledge • Assets • Whiteboards • Outputs • Versions and history Projects inherit Brand defaults but allow specific overrides. ADD WHITEBOARDS INSIDE PROJECTS Whiteboards should be project-level thinking environments for research, strategy, brainstorming, customer journeys, campaigns, storyboards and content planning. Users should be able to: • Create multiple Whiteboards • Link existing Whiteboards • Add documents, images, URLs and AI outputs • Ask AI about the entire board or a selected frame • Choose whether a board is available as AI context • Promote validated information into Project or Brand Knowledge Whiteboard brainstorming should not automat
Bugs?
Remalt chat keeps messaging after adding my prompt instruction, from extracting information from uploaded pdf’s: “I couldn't finish this response: The chat provider truncated the response. No partial editor arguments were applied. Completed drafts and app changes are retained?”
Add Source Categories/Folders to Chrome Capture + Source Library Node
Please improve the Chrome Capture extension and Source Library so saved sources can be organised into categories/folders and then reused directly inside Remalt flows. In the current Chrome Capture popup, I can paste or capture a link. I would like an additional field such as: Category / Folder For example: Link: https://example.com/article Category / Folder: AI Research When the source is captured, Remalt should automatically save it into that folder inside the Source Library. The Chrome Capture workflow could become: Capture Paste a link here... Category / Folder [ AI Research ▼ ] [ + Create new folder ] [ Capture ] The folder field should ideally support: selecting an existing folder creating a new folder directly from Chrome remembering recently used folders optional tags in addition to folders a default "Uncategorised" folder if none is selected Then, inside the Remalt flow builder, I would like a dedicated Source Library Node. That node should show my saved folders and allow me to check which folders should be used as sources for the workflow. For example: SOURCE LIBRARY Select source folders: ☑ AI Research ☑ Ecommerce ☐ SaaS ☑ UX & CRO ☐ Competitor Research ☐ Content Inspiration [ Include all sources from selected folders ] The node should then pass the selected source material into the next workflow node. Example flow: Chrome Capture ↓ Saved to: AI Research Ecommerce UX & CRO ↓ Source Library ↓ ☑ AI Research ☑ Ecommerce ☑ UX & CRO ↓ Research / Extract / Summarise ↓ AI Model ↓ Article / Report / Strategy / Content Ideally the Source Library node should also support: selecting individual sources inside a folder selecting multiple folders "Use all sources in folder" search within saved sources filter by date filter by tags preview sources before using them show source count per folder preserve title, URL, author and publication date pass citations/source URLs downstream choose between full-page content and extracted content optionally refresh outdated sources from the original URL The important part is that Chrome Capture, Source Library and workflow nodes become one connected research system. At the moment, capturing a source and using it later in a flow are separate actions. Folder-based organisation would make saved research reusable and much easier to manage, especially when building research-heavy workflows.
Add SiliconFlow as an AI Connector
Please add SiliconFlow as a supported AI Connector in Remalt. SiliconFlow provides one OpenAI-compatible API for 200+ AI models across LLMs, vision, image, video and audio. It supports models from providers such as DeepSeek, Qwen, Kimi, MiniMax, GLM, OpenAI open-weight models and others. This would give Remalt users a much broader and more cost-efficient choice of models without needing separate integrations for every provider. Useful capabilities for the connector would include: API key authentication OpenAI-compatible API support Dynamic model selection Chat and reasoning models Vision models Image generation Video generation Audio models Streaming Function/tool calling where supported Embeddings and reranking Usage and cost tracking Secure API-key storage SiliconFlow also offers pay-as-you-go serverless inference with no minimum spend and very competitive pricing across many models. Because the API is OpenAI-compatible, the implementation may also be relatively straightforward compared with building a completely new provider interface. SiliconFlow explicitly describes its service as providing one API across models with OpenAI compatibility. Official documentation: SiliconFlow Documentation Pricing: SiliconFlow Pricing
Full Document Preview, Resizable Nodes + More Export Formats
When I use the Website tool to extract information from one or several URLs and connect those sources to a Prompt node, I would like a better way to read, review, and export the complete generated document. Currently, longer outputs are displayed inside a relatively small node. For research documents, implementation guides, SOPs, reports, and other long-form outputs, this makes reviewing the complete result difficult. Please consider adding: ⦿ Full document preview: Open the complete output in a larger reading view, modal, or dedicated document viewer. ⦿ Resizable Prompt nodes: Allow users to manually increase both the width and height of Prompt nodes. ⦿ Expand / collapse view: Quickly switch between the compact Brainboard node and a full reading view. ⦿ Internal scrolling: Allow the complete document to be read without leaving the Brainboard. ⦿ Preserved document formatting: Maintain headings, paragraphs, lists, tables, links, code blocks, and Markdown formatting. ⦿ Clickable source references: When Website nodes are used as sources, retain links or citations back to the original webpages. ⦿ More export formats: In addition to the existing PDF and image export, please add: ⦿ DOCX: For editing and sharing in Microsoft Word or Google Docs ⦿ PDF: For final formatted documents ⦿ Markdown (.md): For knowledge bases, documentation, GitHub, AI workflows, and portability ⦿ JSON: For structured outputs, APIs, automations, AI agents, and downstream processing ⦿ TXT: Useful as a simple universal fallback Ideally, export could be available directly from the node: Export → DOCX | PDF | Markdown | JSON | TXT | Image For JSON export, it would be useful to preserve structured metadata such as: ⦿ Node title ⦿ Generated content ⦿ Source URLs ⦿ Model used ⦿ Creation date ⦿ Connected source nodes ⦿ Citations or references where available This would make Remalt much stronger not only as a content workspace, but also for research, documentation, knowledge management, AI agent workflows, and moving outputs into other tools. Remalt already supports PDF and image exports from several node types, so DOCX, Markdown, and JSON would be a natural extension of the current export system.
Add Pinterest & Reddit to Social Discovery
Please expand Social Discovery with these two high-value research sources: ⦿ Pinterest: Creators, boards, pins, video pins and visual trend discovery ⦿ Reddit: Subreddits, posts, comments, community insights and pain-point research These would add capabilities that are different from the current social platforms. For Pinterest, it would be valuable to search and import relevant Pins, video Pins, Boards and creator content into a Board for visual inspiration, competitor research and trend analysis. Pinterest’s official API supports image/video Pins and Boards. For Reddit, it would be valuable to discover relevant subreddits, posts and comments, then analyse recurring questions, complaints, customer language, pain points, sentiment and emerging topics. Reddit’s developer platform supports access to Reddit posts and comments, subject to its permissions and platform rules. Ideally, both integrations should preserve: ⦿ Original source URL ⦿ Creator/community ⦿ Publication date ⦿ Engagement metadata where available ⦿ Images/video/content ⦿ Board/subreddit context ⦿ Search keyword/topic used The goal would be to make Pinterest and Reddit proper research and discovery sources, not just additional publishing channels.
Expand BYOK with fal.ai, Replicate + Dynamic Model Sync
Remalt already has a strong BYOK foundation with OpenRouter, WaveSpeed, Gemini, Anthropic and other providers. I would love to see the provider layer expanded further rather than adding individual AI models one by one. Please consider adding native BYOK support for: ⦿ fal.ai: Strong production API ecosystem for image, video, audio and multimodal generation ⦿ Replicate, broad catalogue of image, video and open-source models ⦿ Hugging Face Inference Providers - only if this has not already been added More importantly, it would be valuable if Remalt could dynamically discover and expose compatible models available through connected providers instead of requiring every new model to be manually added by Remalt. For example, WaveSpeed already provides access to many leading image and video models. If a model is already available through a connected provider, Remalt should ideally expose it automatically rather than requiring another dedicated integration. The same architecture could support fal.ai, Replicate and Hugging Face as their catalogues evolve. Ideally the system would: ⦿ Automatically sync newly available provider models ⦿ Clearly show which provider powers each model ⦿ Avoid duplicate model entries across providers ⦿ Let users choose the preferred provider when the same model exists on several APIs ⦿ Show estimated cost/credits before generation ⦿ Support provider-specific image/video parameters rather than only generic prompts ⦿ Fall back to another compatible provider when enabled This would make Remalt much more future-proof and allow new models to become available much faster without waiting for individual Remalt releases
Option to add several urls
Hi, I would like to be able to add several urls in the website node, in order to scrape the fully sitemap (website including the different sub pages, like knowledge document site for different SaaS application I use) in order to connect to my AI Agents.
Editor tool panel disappears when using browser zoom
Hi team, I've noticed since first onboarding, and mention it again; an annoying “bug” in the editor workspace. When I zoom in or out using the browser's zoom controls (or touchpad gestures that trigger browser zoom), the side panel and main tool options completely disappear from the screen. The UI doesn't seem to scale correctly, and it forces me to constantly hit reset (Ctrl+0) or refresh the page to get my tools back. It would be great if the tool panels could stay locked or adapt responsively without breaking the workspace layout. Thanks!
Canvas navigation responsiveness
I dont know if it is me only, but on browser the canvas navigation through keypad is slower, however introduction of minimap has improved it but still on screen navigation needs to be more smoother
Deck
Having a deck that consists different elements like canvas, tasks, notes/documents, calender, boards, table This can be in one canvas as well, however it will still require the tasks, boards, kanban, and table