I was testing a bunch of token saving techniques and tools for my coding agent, and surprisingly it can save 20-60% cost.
I bet most people do not want to pay attention on cost saving tricks every time while building, so I'm curious if people will pay for someone else to reduce their Ai bills.
and what would you need to see before trusting to pay and how much?
Providers can cut any savings you might shave off at any time...
the way for true savings is local open models, most are good enough, that's where the true value lies... just make it easy to implement and the bean counters will love your products
Yea, but running local models means high investment at front and less capable models. What I mean by saving cost is more like you can bring your own AI, and we help to make your coding agent work efficiently, like it aware when to verbose and not, which may also helpful to run your local agent faster. In that case, would you think it's helpful?