AI agents for finance teams, every memo and report sourced before its meeting.
Get back one document for each deal, close or review, with the points left for your call. Duale AI works through the data room, the ledger and the press in the background.Due diligence, every finding sourced for the committee
For the CFO, corporate development and the deal team, when a deal has a committee date.Your deal team gets one committee memo to check and decide on. Hand over the target, the data room and the committee date. One Task reads the press and each document family in focused branches. It extracts each contract’s key terms on the way. Follow the due diligence Task step by step.
Month-end close commentary, each variance tied to its entries
For controllers and financial planning teams, before the monthly review.You hand over: “Explain every variance to forecast above €50,000 in the September close, for Tuesday’s review.”
How the Task splits it: one branch per line of the income statement reads the ledger extract, the forecast and last month’s commentary. Code calculates each variance from your data through Tools. Your threshold and past commentary come from a Library. A review branch then checks each explanation against its source entries.
You get back: one row per variance, with its amount, main driver, source entries and status.
Illustrative example: the September close of Heron Supply, a fictional distributor of professional kitchen equipment.
Variance to forecast
| Revenue | Freight costs | Payroll | Marketing |
|---|---|---|---|
| −€70,000 | +€62,000 | +€58,000 | −€61,000 |
Main driver
| Revenue | Freight costs | Payroll | Marketing |
|---|---|---|---|
| A €70,000 hotel order in Germany moved to October | Express deliveries for the Spanish orders placed before the October 1 price increase | A bonus accrual booked in September; the forecast expected it in December | No invoice received yet for the Lyon trade fair |
Source entries
| Revenue | Freight costs | Payroll | Marketing |
|---|---|---|---|
| 297 German order lines | 46 carrier invoices | Journal entry 0917 | Purchase order 2291 |
Status
| Revenue | Freight costs | Payroll | Marketing |
|---|---|---|---|
| Explained | Explained | To confirm | To confirm |
Your controller decides: whether to confirm the two open lines, whether to accrue the trade fair, and what goes to the CFO.
Data analysis and reporting, every figure traced to its source
For finance and controlling teams that produce the same review every month.You hand over: “Prepare the September sales review in our usual format.”
How the Task splits it: one branch per region calculates every figure with code, from your sales and finance data. Your report template sets the layout, and last month’s report sets the comparison. Both come from a Library.
You get back: every figure linked to its source rows, the commentary for your leaders and the points for your call.
Illustrative example: the September sales review of Heron Supply.
- Region
- France
- Revenue
- €2.41M
- Change from last month
- +6.2%
- Main driver
- Two new restaurant-chain accounts in Lyon
- Source rows
- 412 order lines
- Region
- Germany
- Revenue
- €1.87M
- Change from last month
- −3.4%
- Main driver
- A €70,000 hotel order moved to October
- Source rows
- 297 order lines
- Region
- Benelux
- Revenue
- €0.96M
- Change from last month
- +1.2%
- Main driver
- Steady repeat orders
- Source rows
- 158 order lines
- Region
- Spain
- Revenue
- €0.74M
- Change from last month
- +12.8%
- Main driver
- Orders placed before the October 1 price increase
- Source rows
- 131 order lines
- Region
- Total
- Revenue
- €5.98M
- Change from last month
- +2.9%
- Main driver
- Source rows
- 998 order lines
Commentary for your leaders. September revenue reached €5.98M, up 2.9% from August. France and Spain drove the growth. Germany fell 3.4% because a €70,000 hotel order moved to October. For your call: does the October forecast for Germany include that order?
Your leaders decide: the conclusions to draw and the forecast to set.
The capabilities behind these jobs
Each one is set up once on the Platform and serves your other use cases too.Tools
Code execution calculates every figure from your ledger and sales data. Your own code and MCP servers reach those systems. Web research reads the press.
Libraries
Every new close and review follows your templates and thresholds, and compares with the previous one. Store them once, with your past reports.
Before your first finance job
The facts your security and procurement colleagues check.One fixed price per Task
A standard Task costs €0.10 and a fast Task €0.20, excluding VAT, whatever the number of branches or the duration. See pricing
Your providers, your contracts
Your model providers bill you directly for inference.
Your choice of hosting
Use the managed service, hosted in the European Union, or run the Platform in your own infrastructure.
Encrypted storage
All stored data is encrypted, from Library files to database backups.
Unlimited agents
Add agents as your teams and use cases grow.
Your content remains yours
Duale AI does not reuse your content to train a model. See subprocessors
Finance questions
Yes. Each line comes back with its budget and the gap. See how Libraries keep your reference files.
Yes. Every commentary uses your terms, such as your own definition of gross margin or working capital. See what teams keep in Libraries.
Yes. Each entity’s figures come back beside the group total, so one review serves the group and each subsidiary. See how Tools reach your finance systems.
No. Your own code reads them from inside your network. See how Tools reach your private systems.