Many users expressed excitement about Mercury's Command AI Agents for agentic banking because they can handle repetitive time-sensitive tasks, while some prefer basic features like Zelle or distrust AI for sensitive actions.
No Digg Deeper questions have been answered for this story yet.
Most Activity
2/ In Command you can:
a) Run agentic workflows for invoicing, categorization etc b) Send payments and issue cards c) Get insights on your business d) Connect to human support when you need it
1/ Today we are launching @Mercury Command - the future of agentic banking.*
Command lets you kick off AI agents to do work, while staying in full control, all with human support integrated.
@immad @mercury Would love to give you a tour of @llmgateway if interested! (I'm an early Mercury customer, since 2020)
We're SOC 2 type II certified fyi https://llmgateway.io/enterprise
6/ Mercury can work with AI wherever you are.
For those that live in Claude Code/ Cowork / Codex we have our CLI + MCP.
For those that live in Mercury Dashboards, now we have Command.
5/ Command is fully integrated to our human and AI support agents.
We will be continuously improving what the AI is capable of but when it can’t solve your problem we have humans (24/7 for business users) to help answer any questions you have.
3/ Command keeps you in control.
The AI can never trigger actions on your account, it can only prompt users with actions that are then deterministically fired off by the user.
2/ In Command you can:
a) Run agentic workflows for invoicing, categorization etc b) Send payments and issue cards c) Get insights on your business d) Connect to human support when you need it
4/ Command brings complex UX into AI interfaces.
Command triggers graphs and other complex UX components from the Mercury app.
This can give you the best of UX components and text.
3/ Command keeps you in control.
The AI can never trigger actions on your account, it can only prompt users with actions that are then deterministically fired off by the user.
5/ Command is fully integrated to our human and AI support agents.
We will be continuously improving what the AI is capable of but when it can’t solve your problem we have humans (24/7 for business users) to help answer any questions you have.
4/ Command brings complex UX into AI interfaces.
Command triggers graphs and other complex UX components from the Mercury app.
This can give you the best of UX components and text.
7/
Available at your dashboard here https://app.mercury.com/command
We are just getting started here! Let us know what else we should do here.
* While we’ve received conditional approval to become a bank, Mercury today is a financial technology company, not a bank. Banking services provided through Choice Financial Group and Column N.A., Members FDIC.
6/ Mercury can work with AI wherever you are.
For those that live in Claude Code/ Cowork / Codex we have our CLI + MCP.
For those that live in Mercury Dashboards, now we have Command.
As much as I love @mercury I would never use this to invite users, issue credit cards, make payments, etc. Those are highly sensitive and deterministic actions that I would never entrust to a probabilistic system.
Getting insights on my spending or getting help, sure. But that's it!
@immad @mercury 10x my account balance this month, make no mistakes
@immad @mercury Dope!
@immad @mercury I would rather Zelle work than any of this.
@immad @mercury Incredible!
@immad @mercury congrats, you're already killing it, but this is huge!
@immad Looking forward to playing with it!
@immad @mercury Congrats! One thing you're missing on the left menu is Compliance. That's a very very important part of any bank operating system!
@AbhiGutgutia @immad Command doesn't take actions itself! It configures them and surfaces them for your confirmation. Only once you approve, the action runs through an API, same as any other page in Mercury. the LLM never touches the execution. 🤝
@immad @mercury This is exciting — banking is exactly the kind of space that benefits from agentic AI handling the repetitive, time-sensitive work. Curious to see how it handles edge cases at scale. Good luck with the launch!
@immad @mercury Feels like the hard part is less “can the agent act?” and more “can every handoff be reversible and auditable?” Banking is the right stress test for that.