Feature Roadmap

[workflow] Image output

Ability to generate images as output.

[workflow] Audio output

Ability to generate audio files as output.

[memory] Memory API

Persisted memory definitions for workflow to update and retrieve from, primarily for user informations like profile, preferences etc

[tools] customised tools builder

Build customised tools easily with AI, to fit to your workflow needs. Real customer examples: - Send SMS via twilio - Check private listed data via api - Read latest emails

[playground] Free playground for LP

Build, free AI demo playground integrated in landing page.

[RAG] Website scraping

Add support to scrape websites to build AI knowledge base.

[bugfix] Multimodal (video) input error

Error processing videos as input for gemini 2.5 models.

[RAG] AI vision knowledge chunks (multimodal)

Add AI vision to asset processor, to support videos, images and audios for rag database.

[playground]Custom tool builder. (AI tools v0)

Build customised tools for your AI to use. v0 - AI make any API call.

[workflow] Internet search

Add fire crawl API support. Ability to search the internet for your AI.

[workflow] Streaming response

Build web socket connection for streaming response.

[workflow] GPT 5 models

Add support for gpt 5, 5-mini, 5-nano models

[Infrastructure] EU data center

We’re moving our servers and database to EU - Berlin for faster latency across west Europe and East coast US and Canada.

[FEATURE] Conversations API

We’re releasing conversations API, your go to for building gpt style chatbots without writing any code or setting up any database. Checkout Conversation API documentation at https://amarsia.com/documentation Or test it live in our workflow builder.

[workflow] Claude 4.5 models

gpt 5.1 models

[workflow] gemini 2.5 models support

Add support for gemini 2.5 pro, flash and flash lite

[documentation] Update "/docs"

Add documentations for rag and variables.

[RAG] OCR support to build knowledge chunks

Add OCR support to knowledge chunks building, required for PDFs and pptx with images.