Products
OpenClaw AI Agent
Self-hosted OpenClaw AI agent on hardened Ubuntu. Connect Telegram, Slack or Discord with your own keys.
Hermes AI Agent
Nous Research's Hermes AI agent on hardened Ubuntu, chatting on Telegram with your own model key.
Hermes AI Agent + n8n
Hermes AI agent and n8n workflow automation, preconnected on hardened Ubuntu. Hermes builds and runs your workflows.
Sushi OpenShell
NVIDIA OpenShell on Graviton: policy-controlled Docker sandboxes behind a private mTLS gateway.
OpenHands Core
Self-hosted OpenHands Agent Canvas with Python, terminal and Git tools, plus optional private Telegram.
Nanobot AI Agent
Self-hosted nanobot agent with a password-protected web UI, workspace file tools and private Telegram.
Agent Zero Core
Focused Agent Zero workbench with file, Bash and Python tools, your own cloud model and private Telegram.
Aider AI Pair Programmer
Terminal AI pair programmer that edits and commits to your git repo, with the model provider you choose.
Goose CLI Agent
Open-source CLI agent that plans tasks, edits files and runs commands on a private instance.
Open Interpreter
Let a model write and run code from plain-English requests, with confirmation before every run.
Tabby Coding Assistant
Self-hosted AI code completion and chat server on loopback; attach the model you choose after launch.
OpenCode Coding Assistant
OpenCode 1.19.4 terminal coding assistant with a sample workspace, a local file-edit check, and your own model provider after launch.
CrewAI Agent Starter
Two-agent CrewAI 1.15.28 starter that turns a sample sales CSV into checked JSON, using Amazon Bedrock Nova Lite or an offline fixture.
Sushi ML Studio
No-code machine learning in your browser: upload a spreadsheet, train scikit-learn models, predict.
DuckDB Analytics
DuckDB CLI for fast local CSV and Parquet queries, with pinned extensions for reading your S3 data.
Open WebUI
Private ChatGPT-style chat UI for any OpenAI-compatible endpoint, reached through Session Manager.
AnythingLLM RAG Workbench
Self-hosted AnythingLLM workspace to chat with your documents (RAG) using the models you choose.
Flowise
Drag-and-drop Flowise builder for LLM chatbots, RAG pipelines and agents, on loopback-only HTTPS.
LiteLLM Proxy
LiteLLM gateway: one OpenAI-compatible HTTPS endpoint in front of all your LLM providers.
Qdrant Vector Database
Private single-node Qdrant vector database for RAG and semantic search, with a per-instance API key.
Docling document conversion
Convert born-digital PDF and DOCX files into Markdown and structured JSON on CPU. Local layout and table models are included; OCR and remote enrichment are disabled.
Typesense search server
Index a product catalog and query typo-tolerant text search, filters and facets on one instance. The source-built single-node server uses local persistent storage behind a private authenticated HTTPS gateway.
Whisper CPU transcription
Transcribe English and Spanish recordings on CPU with faster-whisper and a bundled multilingual small model. Responses include text, segments and word timestamps. No model-provider account or GPU is required.
Presidio text redaction
Detect and replace selected sensitive entities in English and Spanish text using local Presidio analysis and bundled spaCy models. Responses include redacted text and input spans.
Langflow
Langflow 1.13.0 visual editor, package 0.2.1, with an administrator login, SQLite storage, a local demo, and your own model provider.
Gradio AI Interface
Authenticated Gradio 6.30.0 page for a local sentiment model, with SQLite results you can read back on Ubuntu 24.04.
Chroma Vector Database
Persistent Chroma 1.5.9 vector database on loopback for embeddings you supply, with a demo that checks the saved vectors.
MLflow Experiment Tracker
MLflow 3.17.2 tracking server for parameters, metrics and artifacts, with a per-instance sign-in and local SQLite storage.
LibreChat
LibreChat 0.8.8 chat workspace on Ubuntu 24.04. Bring your own TLS MongoDB and Amazon Bedrock Nova Lite. No database server is included.
Sentiment Studio
Sentiment Studio 0.1.0 local Streamlit page for one text field or a small CSV. Sign in through a Session Manager tunnel. No model provider or external database.
Vane Search
Vane 1.12.2 answer engine with SearxNG citations on Ubuntu 24.04. You bring the chat and embedding models. Package 0.2.2.
Chatterbox Text-to-Speech
Chatterbox Nano text-to-speech on CPU. One English sentence at a time, in the voice of a short sample you provide. Package 0.1.7.
PDF Translator
Translate digital English PDFs into Spanish and keep the page layout. BabelDOC 0.6.2 with your own translation model. Package 0.1.4.
Crawl4AI Web Extractor
Crawl4AI 0.9.5 turns pages into Markdown and JSON with a headless browser. Token-protected HTTPS on the instance. Package 0.2.7.
PaddleOCR Text Recognition
PaddleOCR 3.7.0 reads English text from PNG and JPEG images on CPU and returns words, scores, and boxes. Package 0.1.3.
BentoML Prediction Server
BentoML 1.4.39 prediction API on CPU with a bundled scikit-learn sample model and a Swagger page. Package 0.1.3.
Caddy HTTPS Gateway
Caddy reverse proxy, private by default, with a per-instance local CA and automatic HTTPS.
HAProxy Community LTS
HAProxy LTS HTTP load balancer with a password-protected stats page.
WireGuard Private Access
WireGuard endpoint with keys generated on first boot and UDP 51820 closed until you add peers.
Unbound DNS Resolver
Validating, caching DNS resolver that answers on loopback only until you open it up.
OpenSSH SFTP Drop Box
OpenSSH SFTP drop box: key-only users locked to their own folder, with no shell.
Uptime Kuma Monitoring
Uptime monitoring with HTTP, TCP, DNS and ping checks behind a private HTTPS dashboard.
Gatus Status Page
Status page with HTTP, TCP and DNS checks, SQLite history and per-instance TLS.
Fluent Bit Log Gateway
Fluent Bit log gateway with HTTPS ingest, a bounded disk buffer and a local file output.
Vector Log Pipeline
Vector log pipeline with authenticated HTTPS and file ingest, VRL redaction and a disk buffer.
OpenTelemetry Collector Gateway
OpenTelemetry Collector OTLP gateway for traces, metrics and logs, with TLS and file export.
Gotify Notification Server
Private Gotify notification server with a REST API, web UI, live WebSocket stream and history.
Langfuse LLM Observability
Self-hosted Langfuse to trace, evaluate and debug LLM apps, with the full single-node stack.
VictoriaMetrics single-node
Store time-series samples and query them with MetricsQL on one instance. The community single-node server uses local disk and one-month retention behind a private authenticated HTTPS gateway.
Promptfoo Evaluation Workbench
Promptfoo 0.124.1 workbench to compare prompt variants and inspect JSON and SQLite results, including a model-free example.
Valkey Single-Node Cache
Valkey single-node cache with loopback TLS, an ACL user, an LRU memory cap and no persistence.
PgBouncer Connection Pooler
PgBouncer PostgreSQL connection pooler with required client TLS and SCRAM authentication.
NATS Core Broker
NATS Core broker with loopback TLS, per-user subject permissions and the nats CLI.
Mosquitto MQTT Broker
Eclipse Mosquitto MQTT broker with loopback TLS, users, topic ACLs and persistence.
OpenTofu Runner
OpenTofu with a pinned offline AWS provider mirror for validate and plan. Nothing auto-applies.
Ansible Core Runner
ansible-core runner with pinned collections, a localhost inventory and a vault password.
Restic Backup Worker
restic backup worker: daily encrypted backups, a safe restore script and an optional S3 repository.
Rclone Transfer Worker
rclone transfer worker: a daily checksum copy that never deletes, with an optional S3 target.
Gitea Git hosting
Host private Git repositories, issues and pull requests on one instance with SQLite storage and native HTTPS authentication. The first boot creates a private administrator account.
No product matches that search. Try a broader term or .
How every product works¶
-
1. Subscribe
Pick the product on our AWS Marketplace seller profile. Most products include a free trial.
-
2. Launch
Start an EC2 instance from the listing with your own key pair, VPC and security group. Secrets are generated on first boot.
-
3. Follow the guide
Each product's getting-started page takes you from launch to a working service, then configuration and day-two operations.
Billing runs through your AWS account: the product's hourly software fee plus the usual EC2, EBS and data-transfer charges. Stop any time by terminating the instance.
Need help?¶
Write to info@softwaresushi.com with the product name, its version and your AWS Region. Software Sushi is an AWS Partner; every product is supported by the engineers who build it.