I work with hundreds of startups each year. These are some of the common tools I see. These are in no particular order and I am not endorsing any of these vendors.
- https://github.com/aws-samples/amazon-bedrock-client-for-mac
- https://github.com/danny-avila/LibreChat
- https://github.com/open-webui/open-webui
- Claude Computer Use
- StageHand
- https://generalagents.com/ (ACE) - new foundation model
- BrowserBase
- NovaAct
- Stitch
- FiveTran
- AWS Zero ETL
- Debezium
- Bucado
- https://github.com/e2b-dev/E2B
- https://github.com/huggingface/smolagents
- https://github.com/open-webui/open-webui
- https://github.com/ColeMurray/ai-code-agent-environment
- Cursor
- Devin
- Continue
- Cody - https://sourcegraph.com/cody
- Aider
- AmazonQ
- Windsurf
- Codebuff
- Cline
- OpenDevin
- Pythagora
- Roocode
- ClaudeCode
- Augment - https://www.augmentcode.com/
- https://www.factory.ai/
- https://duplocloud.com/home-r2
- https://mcp.so/
- https://cursor.directory/mcp
- https://playbooks.com/mcp
- https://github.com/cline/mcp-marketplace
- https://glama.ai/mcp/servers
- https://github.com/punkpeye/awesome-mcp-servers
- https://playbooks.com/
- https://cursor.directory/
- https://docs.cline.bot/improving-your-prompting-skills/cline-memory-bank
- CrewAi
- AWS Multi Agent Orch
- AWS Bedrock Multi Agent (preview)
- LangChain
- LlamaIndex
- Dynamiq
- Flowise
- Gumloop
- n8n
- Xpander
- Owl
- Autogen
- Maestra
-
New Relic
-
Data Dog
-
Post Hog
-
Sentry
-
AWS Cloud Watch
- Sentry
- Airbrake
- PostHog
- PropelAuth
- Auth0
- Clerk
- https://supertokens.com/
- Keycloak
- AWS Cognito
- https://github.com/camel-ai/owl
- https://github.com/agno-agi/agno
- https://github.com/mastra-ai/mastra
- CrewAI
- AWS Bedrock Agents
- AG2 (aka Autogen)
- BoundryML (BAML)
- DotTxt Outlines / https://github.com/dottxt-ai/outlines
- https://browser-use.com/
- https://www.skyvern.com/
- https://github.com/aws/nova-act (AWS Nova Act)
- https://github.com/qodo-ai/pr-agent (AI PR reviews)
- https://www.ellipsis.dev/
- Langchain
- Langfuse
- BAML
- Zenbase - https://github.com/zenbase-ai
- Vellum
- Composio
- HumanLayer
- LangGraph
- https://www.llamaindex.ai/
- Retell
- vocode
- Reducto
- Mindee
- LayoutLLM
- AWS Textract & Comprehend
- Multi-modal LLMs direct (OpenAI & Claude)
- Mistral (https://mistral.ai/news/mistral-ocr)
- Twilio
- Sendgrid
- Resend
- Supabase
- Firebase
- Eleven Labs
- Play.ht
- OpenAI Whisper
- Cartesia
- https://github.com/gpt-omni/mini-omni
- https://github.com/ictnlp/LLaMA-Omni
- https://github.com/livekit/agents
- https://github.com/kyutai-labs/moshi
- https://emova-ollm.github.io/
- Modal
- BaseTen
- Beam
- Fireworks
- Groq
- Runpod
- AWS Bedrock
- LiteLLM
- OpenRouter
- llms.txt - https://llmstxt.org/
- agents.json - https://github.com/wild-card-ai/agents-json
- Llama3 Qwen
- OpenAI GPT
- https://huggingface.co/nvidia/NVLM-D-72B
- Mistral
- Anthropic Claude
- Luma
- Poolside
- AWS
- Lambda Labs
- SF Compute
- Voltage Park
- CoreWeave
- Fullstory
- Pendo
- Posthog
- Posthog
- Amplitude
- NextJS
- Figma
- https://moqups.com/
- Sketch
- Framer
- Retool
- Stripe
- Vercel
- AWS Amplify
- Bolt
- Lovable
- Spawn
- Marbelism
- Smithery
- MCP.so
- Drata
- SecureFrame
- Tugboat
- DashSDK
- Vanta
- FastAPI
- Flask
- OpenRouter
- LiteLLM
- Temporal
- AWS Step Functions
- SAM
- CDK - https://aws.amazon.com/cdk/
- SST - https://sst.dev/
- Porter
- FlightControl
- CDK
- Flight Control
- Port7777
- Metabase
- Tailscale
- https://github.com/sigoden/aichat
- ClaudeCode
- Amazon Q CLI
- https://paperswithcode.com/method/sepformer
- https://github.com/JusperLee/Dual-Path-RNN-Pytorch
- https://github.com/JusperLee/Conv-TasNet
- https://github.com/asteroid-team/asteroid
- https://docs.nvidia.com/nemo-framework/user-guide/24.09/nemotoolkit/asr/speaker_diarization/intro.html
- https://github.com/speechbrain/speechbrain
https://github.com/matterport/Mask_RCNN https://paperswithcode.com/method/retinanet
- Flowise
- Gumloop
- Lindy
- n8n
- VectorShift
New models are coming out every day. By the time you figure how how to find tune a new model will be released with better performance. Focus on prompt engineering and inference time fine tuning.
No, don't use Kubernetes unless you have a business requirement for it.
- https://arxiv.org/pdf/2410.24190 (Bias in LLM models)
- https://www.instantdb.com/ (build apps with data fragments instead of data objects)