The answer is by household and step
Counts split across harvest, account placement, rebalance, and cash. You see who moved, not a single overnight total.
Athena AI
Clients ask about their money and get answers in your firm’s voice. Advisors use the same chat on every Mission Control screen. You can ask what traded last night, or why an item is waiting, instead of piecing it together from a trade list.
Book a demoPopular questions to get you started
Demo of the same Financial Advisor AI clients open. Not a live household.
Mission Control
Mission Control is not a dashboard you only stare at. Trading, exceptions, email, and conversations each have Athena AI on that screen. Ask what ran overnight, why something is open, whether an email series landed, or what a client already heard. The answer uses the record on that page. You stay on the household you already have open, instead of opening a second chatbot.
Advisors ask in four places. The client chat is above. These are the advisor screens, with the same household attached.
Trading
The investing engine harvests losses, places assets in the right accounts, rebalances, and reinvests overnight. In Mission Control you do not rebuild that story from a trade list. You ask. Athena reads the overnight run: how many households traded, which accounts moved, which harvests stayed in the same asset class, and which items need review.
Typical questions include what ran last night, whether Chen harvested a loss, which households sold to raise cash, and whether the book stayed inside policy. The answer names households and the step that ran, not a generic “trades posted.”
| Household | Stage | Status |
|---|---|---|
| Chen | Harvest · US equity | Wash-sale hold |
| Hale | Cash raise | Waiting for you |
| Rest of book | On policy | No trade needed |
Counts split across harvest, account placement, rebalance, and cash. You see who moved, not a single overnight total.
Chen’s lot was sold and a peer fund in the same asset class was bought. Wash-sale rules were checked across taxable, Traditional, and Roth accounts. The household’s market exposure did not change.
Thirteen households raised cash. Available cash was used first, then tax-aware sales. Hale is the one still open, because the request is over the firm limit.
Most of the book did not need a trade. That is a result, not a missing report. The only rows that need you are the exceptions.
Hale household
Cash request · above the firm limit $7,400 cash used first. Tax-aware sells staged. Roth untouched. Waiting on you.Exception review
Routine harvests do not need you. The items that do need you, like a wash-sale hold, a cash request over the firm threshold, or a lot that cannot be sold, open with context. Chat on the exception page tells you why the engine stopped, what it already tried, and what is staged for your approval.
You see the investment policy note, the cash already used, the tax-aware sales queued, and the last client conversation on the same household. Approve it, send it back, or ask a follow-up. You are not rebuilding the case from three systems.
This is not a ticket number. Hale went over the firm limit. That is the only reason this is in the queue. The engine already used available cash and staged the tax-aware sales.
Sales are queued in tax order: losses first, then gains, with restricted lots skipped. Roth accounts are protected. Nothing has gone to the custodian yet.
Last night’s AI conversation sits beside the decision. You see what Hale already heard before you approve the cash request.
Email effectiveness
Outbound email lives on the household, not in a separate email tool you check after the meeting. On the email screen, ask how a series performed: who it was sent to, who opened it, who replied, who booked a meeting, and who asked the client AI the same night. Ask who was held back because of an exception.
The point is not a vanity open rate. It is whether the review email started a next conversation, and whether a household that should have received it was sitting in the exception queue instead.
Clients on the Q2 review series
Opened the review in the first two days
Wrote back on the household thread
Booked from the same send
Chat review
Client AI history lives on the household. On the conversation page, ask what they asked this week, whether the answers stayed on plan, what you rated, and what to pick up on the next call. You are not hunting a transcript in another product.
If a client asked to talk to a person, that request is on the same record. If an answer used the retirement projection, you can see it. If you left a comment, the next person who opens the household sees it.
Tuesday was the rollover. Then they asked about retiring at 62. Both answers used the plan on file. Nothing was invented. The numbers came from the planning engines.
You rated the retire-at-62 answer as on plan. On the rollover conversation you left a note for whoever opens it next.
Start with the rollover they asked about after the review email. They already heard that 62 still works if the 6.8% mortgage is refinanced first.
Client AI
The client app is the other side of the same layer. Clients ask about retirement, Social Security, contributions and the employer match, allocation and risk, opening a managed account, cash, Roth versus pretax, debt, credit, term life, healthcare documents, housing and mortgage, 529 college accounts, budget, estate basics, net worth, and what Athena did this week. They can email a recommendation or ask to talk to a person.
The voice is the firm’s, shown under your brand, not a generic model. Numbers come from the household and the planning engines. The conversation is stored on the household record you open in Mission Control.
How it stays safe
Core recommendations come from planning and investing engines, not from the model making up numbers. The model writes the answer in the firm’s voice. It is not allowed to invent balances, accounts, allocations, or ages. Advisors review, rate, and comment in Mission Control. People take the edge cases.
AI questions buyers ask
What clients can ask, what you ask in Mission Control, and what the model is not allowed to invent.
Core recommendations come from deterministic engines and simulations. The model explains them in the firm’s voice. It does not invent balances, accounts, allocations, or ages.
Chat is available on every Mission Control screen. On trading, ask what the overnight run did. On exceptions, ask why a household is open and what is waiting for you. On email, ask whether a series landed and who followed up. On conversations, ask what the client already heard and what to pick up.
Yes. The assistant is on the page you are already on, whether that is trading, exception review, email, or conversation review, so the answer uses that record instead of sending you to a second tool.
Yes. Client AI history lives on the household. Ask the conversation page what they asked this week, then pick up the same story on the call.
Yes. Conversations are stored. Mission Control can review, rate, and comment so supervision stays on the household record.
They can talk to a professional or schedule a meeting. That request sits on the household with the thread, so the advisor is not starting from a blank page.