100% TypeScript Stack for Building Production-Ready AI Agents
Good stuff, thanks for sharing Andrew 🙏 any updates since posting? What use cases have you developed these agents for?
Thank you, Alex!
Few updates:
I use Langfuse and AI SDK Telemetry for observability now mostly.
For memory I switched fully to Mastra memory + building my own memory framework
My recent agentic projects:
AI Analyst for luxury real estate (client project)
Club members concierge agent (client project)
Private markets (secondaries) analyst (client project)
Agent that helps me to operate my cohort (Advanced Knowledge Systems) in Slack
[all built on this exact stack]
Andrew, where can I find info about your next cohort?
https://aks.systems/
Coming mid-February
Thank you for sharing!
Excellent work, Andrew. Do you have a working demo and a rough estimate of the implementation cost? What are the annual running costs? Thanks, Lee.
Thank you, Lee! Demo is coming!
In terms of running costs - it really depends on usage.
Vercel has a very generous free tier
AI SDK & Mastra are open source
Mem0 also have a free tier
So if it's your pilot project - you pay only for the LLM tokens
Good stuff, thanks for sharing Andrew 🙏 any updates since posting? What use cases have you developed these agents for?
Thank you, Alex!
Few updates:
I use Langfuse and AI SDK Telemetry for observability now mostly.
For memory I switched fully to Mastra memory + building my own memory framework
My recent agentic projects:
AI Analyst for luxury real estate (client project)
Club members concierge agent (client project)
Private markets (secondaries) analyst (client project)
Agent that helps me to operate my cohort (Advanced Knowledge Systems) in Slack
[all built on this exact stack]
Andrew, where can I find info about your next cohort?
https://aks.systems/
Coming mid-February
Thank you for sharing!
Excellent work, Andrew. Do you have a working demo and a rough estimate of the implementation cost? What are the annual running costs? Thanks, Lee.
Thank you, Lee! Demo is coming!
In terms of running costs - it really depends on usage.
Vercel has a very generous free tier
AI SDK & Mastra are open source
Mem0 also have a free tier
So if it's your pilot project - you pay only for the LLM tokens