Cash and runway
Runway, burn rate, break-even, cash position, working capital, and seasonality.
Finance Assistant
Eigenn runs 34 specialist agents over your finance records. Each one prepares a specific piece of work and shows how it got there. Nothing an agent proposes changes a record or reaches a customer until a person approves it.
assistant.eigennawaiting review
assistant proposal
34 agents
1 approval | Agents prepare the work. A person decides whether it happens.
01
Agent drafted the follow-up
02
Reasoning shown beside the draft
03
Operator approves before it sends
34
Specialist agents, each scoped to one job
Proposes
An agent drafts; a person approves
Recorded
Every agent action lands in the audit log
Most finance automation is either trivial or unaccountable
Rules-based tools only handle work you can describe in advance. General AI tools produce output you cannot check. Finance teams need work prepared for them and evidence of how it was prepared.
Repetitive
Categorization and follow-up consume the week
Unverifiable
A suggestion without reasoning cannot be checked
Unbounded
Automation that acts alone is hard to trust with money
How assistant proposal works
Rules-based tools only handle work you can describe in advance. General AI tools produce output you cannot check. Finance teams need work prepared for them and evidence of how it was prepared.
01
Ask, or let it run
Ask a question in the assistant, or set a schedule and a watcher so the work runs without being asked.
02
A specialist does the work
A router picks the agent for the job — forecasting, collections, reconciliation, document questions — and it reads only the records it needs.
03
You approve or reject
The proposal arrives with its reasoning and the records behind it. Approve it, change it, or reject it. Auto-approved actions keep an undo window.
What the agents prepare
Each agent covers one area and reads from the same records your team works in.
Forecasts and stress tests
Forecasts, sensitivity analysis, and scenario comparison for hiring, revenue, and collection delays.
Collections
Which accounts to chase first, payment likelihood, the best time to send, and drafted messages.
Transactions and reconciliation
Categorization with the reason attached, suggested matches, and anomaly detection.
Documents
Answers drawn from your documents, with the source passage cited.
Watchers and schedules
Standing checks that run on a schedule and raise something only when a condition is met.
Decision memory
Decisions you record stay attached to the records that prompted them, so the reasoning survives the meeting.
Autopilot
Let named agents act inside limits you set. Off by default; you choose which actions qualify.
Where this shows up in the week
Monday morning briefing
A scheduled run summarises what changed in cash, AR, and transactions since the last review.
The chase list
Overdue accounts ranked by payment likelihood and amount, with a drafted message on each.
The question you cannot answer from the books
Ask what happens to runway if the largest customer pays 30 days late, and get the model with its assumptions.
Let the agents prepare the week
Connect your records and see what the assistant proposes before you approve any of it.