Whether you are running a team, a small business, or an individual brand, are there any manual tasks that are wasting your time or requiring headcount to handle? If so, you can consider using CraftBot to automate them for you.
In our team, CraftBot has already taken over 30% of the manual workload, and we are still working to improve it further to the point where it can handle at least 70% of computer-based tasks.
Here are some of our use cases:
Team directory management: CraftBot helps us organize our team folder on GDrive. All we have to do is upload files into a temporary folder, where they are transferred, categorized, and sorted. CraftBot manages an MD file as an index when needed to retrieve data for us.
PM work: CraftBot helps us crawl issues in our GitHub repo and put them into a requirements page in Notion. Then, it keeps track of progress every day at 7 AM and 5 PM JST. No more copying and pasting one by one for me.
Lead sourcing: It helps us research leads that fit given requirements and compile a report of their contact information, reasons why they qualify, and more. Then, I reach out to the leads personally. (as people can sniff out AI and sometimes reject it) But of course, outreach can easily be automated using CraftBot with app integrations and channels.
These are just some of our own use-cases. I would love to hear yours and discuss how CraftBot can help you save time and cost, and create value for you and your team. Please leave a comment below!
Looks awesome! Can local LLM serving be done via ollama?
CraftBot
@rohan_ghosh5 Yes you can! You can set it to point to an Ollama end point, or you can just do one-click installation to setup Ollama directly.
ConnectMachine
How does CraftBot compare performance-wise when running locally vs cloud-heavy agents?
CraftBot
@syed_shayanur_rahman CraftBot is faster and cheaper for desktop automation, personal workflows, and background monitoring. Tool calls (filesystem, shell, GUI automation via the browser/CLI/GUI adapters) execute in-process through the local action executor. No serialization, no HTTPS round-trip, no sandboxed VM spin-up per action.
Cloud agents pay that cost on every tool invocation, plus egress for any file or screen context they need to inspect. They also monitor compute when idle (file watchers, schedulers, long-running monitors) and so burn budget even while waiting; CraftBot's event loop is effectively free at idle since it's just a local Python process.
Documentation.AI
Congratulations on shipping. What’s the most common use case users are adopting right now with CraftBot?
CraftBot
@roopreddy Thank you! Have not conduct any survey yet, but from a small group of testers we know, they are using it to generate reports. Basically moving data from one place to another + compile them.
Onboarding might be overwhelming for non-technical users. Is there a guided template or tutorial to follow? Thanks in advance.
CraftBot
@zerotox Thanks for the feedback! We are working on user playbook and documentation for that.
This look super exciting, but since it’s local-first, how are you handling security and permissions when the agent executes tasks on behalf of the user?
"Lives inside your machine and works 24/7" — what are the actual resource requirements when it's running idle? If this is sitting in the background continuously, knowing whether it's sipping 200MB of RAM or hammering the CPU matters a lot for whether it's practical on a normal laptop.
Local AI agent that plans ahead and doesn't send your data anywhere — solid foundation. 150+ default skills sounds impressive until you start wondering how many actually work reliably. Worth testing either way. 🤔