Cobalt aggregates your bank, brokerage, transaction, budget, and subscription data in one place, and this integration lets an AI agent in Neotask read that aggregated picture and answer questions about it directly. Ask how much you spent on subscriptions last month and it checks your Cobalt data and tells you, broken down by service if you want. Ask whether your spending in a category is trending up and it pulls the transaction history and gives you a straight answer instead of a spreadsheet you'd have to build yourself from statements pulled across several separate apps. Because Cobalt already pulls together accounts that would otherwise sit in separate places, the agent can talk about your whole financial picture in one conversation, checking your budget against actual spending, flagging a subscription you may have forgotten about, or summarizing brokerage activity, without you switching between banking and investment apps to piece any of it together on your own during a busy week. Over time, that also means the same questions get easier to ask again the following month, since the agent is always working from the same aggregated view rather than a fresh export you'd have to prepare each time. It reads more like a running conversation about your finances than a one-time report you generate and then set aside.
| Review accounts | Pull a summary of connected bank, brokerage, and other accounts |
| Analyze transactions | Search and summarize transaction history by category, merchant, or date range |
| Check budget status | Compare actual spending against a set budget for a category or period |
| List subscriptions | Surface recurring subscription charges detected across accounts |
| Summarize brokerage activity | Pull recent brokerage account activity and balances |
| Spot spending trends | Compare spending in a category across two or more time periods |
At the start of the month, you ask the agent how your spending compared to your budget in the previous one. It pulls your Cobalt transaction data, breaks it down by category, and points out where you went over, such as dining or subscriptions, so you can adjust before the new month gets underway. It also notes any category that's been trending up for three months in a row, which is easy to miss when you only look one month at a time, and it suggests which categories are worth setting a tighter limit on given the pattern. The whole check-in takes a few minutes instead of an evening spent cross-referencing statements. You ask it to remember the adjusted limits, and the next month it opens the conversation by telling you whether you actually held to them.
You ask the agent to list every recurring subscription charge it can see across your accounts. It pulls the aggregated data from Cobalt, groups the charges by service and amount, and flags a couple you don't recognize using regularly, giving you a concrete list to cancel from rather than scrolling through months of statements. It also totals what those subscriptions cost annually, which tends to be a bigger number than expected, and it notes which ones have increased in price since you first signed up, based on the transaction history it can see. A few weeks later you ask it to check whether the canceled ones actually stopped billing, and it confirms two did while a third still shows a charge, so you know exactly which cancellation to follow up on.
No, it reads and summarizes your aggregated financial data; any transfer, payment, or cancellation is still something you do yourself once you know what needs attention.
Whatever accounts you've connected to Cobalt, including bank, brokerage, and other linked sources, are what the agent can read and summarize when you ask a question.
The agent reads your Cobalt data to answer your questions within the conversation; it doesn't use that data for any purpose beyond what you specifically ask about.
Yes, you can ask for a comparison across any date ranges covered by your connected transaction history, including looking at trends over several months at once.
It can summarize brokerage account activity and balances as part of your aggregated Cobalt data, alongside banking and budget information, in the same conversation.
You can ask it to review recent transactions for anything that looks out of pattern compared to your usual spending in that category, which is a quick way to catch a surprise charge.