Feels a lot like home. I actually want to sit with somebody to see and learn their multi agent multi session workflow on actual critical engineering work where they know the stuff that's getting created. In fact a lot of times I need to ask the agent to explain the flow of tool calls to understand the workflow rationale.
Call me old-school, but I don't feel working on multiple things at once to make up for agent thinking time is the way to go. Unless it's trivial work, the context switching tax is brutal, even just for two sessions in parallel.
2 more points I would like to add
- understanding the boundary between the product and the services offered by the FDE to understand where the value is getting accrued
- most people talk about moving fast but the moment a FDE is in the org there is a lot of noise in terms of people/process/product to navigate to reach the signal and a bigger effort to take it to production. How is this being priced?
In actual mission critical software after the deal is signed and the SOW is agreed upon, the FDE actually has to swim through an incredulous magnitude of noise in terms of people, process, products in the customer's organisation to actually find that "solvable crux" that can produce an attributable outcome. Often solving that is 20% and taking it to production via the daisy chain of command is the 80%.
If you look at the India -> France corridor powered by the NPCI-NIPL-Lyra partnership a resident Indian can purchase goods/services in France real time which indicates the banks offer a real time LRS check for the outflow of such funds. Why is the same service not extended to brokers like Vested when somebody wants to buy a US stock? Purpose code attribution and LRS check cannot take 3-4 days. What am I missing?
@blinkitcares You have been breached RBI's harmonisation TAT as well. I have highlighted it multiple times but seems like you folks are side stepping that conveniently. DO NOT do that.
@blinkitcares My mother has been following up for a refund against an order cancellation for the last 12 days. Your customer support team is highly incompetent and hasn't be able to solve the issue for "12 days"!!!!!. You've also breached @RBI's refund harmonisation TAT of 5 days. cc @GooglePay was used as the payment method. @NPCI_NPCI
Understanding lean really gave a new perspective in terms of programming. The idea, that the association of data to a particular data type can be modelled as an association of a proof to a proposition such that a compiler can assert this fact across the program is quite powerful.
In fact it becomes a powerful modality for verifying agentic workflows in environments that need proof of correctness. I am imagining systems where the universe of laws in a domain can be represented as lean code and the output(s) of an agent loop can be adjudicated with that lean code for correctness with the exceptions being elevated for humans.
One does not bind time. You slice time and allocate a slice for an unit of work. Whenever I hear this phrase "let's take a time bound approach to this", there is this aforementioned internal commentary in my head :)
@hyojun_at@browsercompany Solved the problem with environment variable NODE_ENV=production. Had to figure it out by searching the CLI executable strings representation. Request you folks to improve the cli experience. For now I have added it manually to the mcp.json as well.
@hyojun_at@browsercompany Nice product. I was trying to use the aside cli but it fails with "fetch failed" error. Is the current cli stable and ready for use?
Is there anybody that's building something that can answer these set of questions for a company in 60 seconds.
How many customers do we have?
What are they buying from us?
What are the terms of purchase in terms of pricing, SLO?
How much are we supposed to make from them?
How much are we actually making from them? Is there a delta? If yes, why?
What is the opportunity to solve more problems and up-sell/cross-sell?
@harimenon_bb A different question. Do you think it is a challenge for D2C businesses today to understand their contribution margins on a sku or the promotions on a sku they run across all the disparate platforms and data sources?