Agent 01 · Analyst
Comments on variances
Analyzes data and writes variance comments that go straight into the form.
Analysis
Results commentary
ProductFincorex AI
- Analyst
- Operator
- Architect
- Human-in-the-Loop
Three AI agents.One decision – yours.
Analyst, Operator and Architect work inside Fincorex® only within your permissions. They propose changes to data, metadata and settings, and the decision is always yours.
Book a free demo▦ Income Statement with Comments
2023-12
LE11 IT Company Sp. z o.o.
Cost centers – All
Projects – All
Line item
Actual
Budget
Act - Bud
Results commentary
Revenue
21 960 059
18 865 999
3 094 059 ↑
+16,4% vs budget…
Cost of sales
−4 676 435
−3 383 077
−1 293 357 ↓
−38,2% vs budget…
Direct margin
17 283 624
15 482 922
1 800 702 ↑
+11,6% vs budget…
Departmental costs
−10 476 945
−12 900 008
2 423 062 ↑
18,8% savings…
Sales margin
6 806 679
2 582 915
4 223 764 ↑
+163,5% vs budget…
Context: FORM: Income Statement with Comments
Prepare results commentary: actuals vs. budget for December 2023.
› </> Tool calls
OK
⚠ Confirm write operation
Tool: fdx_set_cell_comment
Review the comments below – you can edit each one before approving.
A#PRZYCHODY · W#MTD
Revenue — December 2023, MTD, PLN. Actuals came to PLN 21,960,058.74 against a budget of PLN 18,865,999.35, a favorable variance of PLN 3,094,059.39 (+16.40%).
A#KOSZTY · W#MTD
Departmental costs — December 2023, MTD, PLN. Actuals came to PLN −10,476,945.46 against a budget of PLN −12,900,008, i.e. PLN 2,423,062 less than planned.
The agent wants to perform a write operation. Check the arguments and approve or reject.
openai · gpt-6.1-sol
Analyst
✎ write mode
● Online
▦ P&L 12M
Budget
2027
LE11 Company 11
BS16 Product line
Excl. ICO
Line item
2027-01
2027-02
2027-03
2027-04
PLM_SALE : Sales revenue
× 1,10
× 1,10
× 1,10
× 1,10
PLM_COS : Cost of goods sold
% rev.
% rev.
% rev.
% rev.
PLM_MRG : Gross sales margin
Σ
Σ
Σ
Σ
PLM_C : Operating costs
% rev.
% rev.
% rev.
% rev.
PLM_EBITDA_ADJ : EBITDA
Σ
Σ
Σ
Σ
Context: FORM: PnL 12M
Forecast revenue for the Budget (SV#Budget_Last) for 2027 based on the growth rate between Forecast 2026 and Actuals 2025. For cost lines, use the current cost-to-revenue ratio in the 2026 Forecast scenario. Save the data.
02.10.2026 12:45
› </> Tool calls (17)
OK
⚠ Confirm write operation
Tool: fdx_calc_from_base
Mode
Recalculation with an engine rule – overwrites target cells
Application
CONSO_PL
Rows
121
Source
SV#Forecast_Last
Factor
1,10
Target
P#202701 … P#202712
The agent wants to perform a write operation. Check the arguments and approve or reject.
openai · gpt-6.1-sol
Operator
✎ write mode
● Online
▦ Income Statement
2024-03
LE22 IT Company SK
Cost centers – All
Projects – All
Line item
Actual
Budget
Act - Bud
Act - Bud %
Revenue
419 564
345 289
74 275 ↑
21,5%
Cost of sales
−217 348
−108 536
−108 812 ↓
−100,3%
Direct margin
202 215
236 752
−34 537 ↓
−14,6%
Departmental costs
−176 950
−214 905
37 955 ↑
17,7%
Sales margin
25 265
21 847
3 418 ↑
15,6%
Context: FORM: Income Statement
Add a new company with code LE24 and name LE24 IT Company BG, and assign it the currency EUR.
28.09.2026 13:20
› </> Tool calls (6)
OK
⚠ Confirm write operation
The agent wants to add a new company to the group structure. Check the arguments and approve or reject.
Code
LE24
Name
LE24 IT Company BG
Currency
EUR
Type your message…
openai-compat · deepseek-v4-flash
Architekt
● Online
▦ RZIS 12M
2025
MTD
LE01
Budget Working
MPK IT EN
Pozycja
202501
202502
202503
202504
Przychody
13 582 595
15 235 561
16 553 313
14 663 166
Selling costs
−2 991 606
−3 566 976
−4 515 898
−4 231 197
Direct margin
10 590 988
11 668 585
12 037 415
10 431 969
Departmental costs
−10 287 994
−10 692 584
−11 750 603
−10 684 396
Sales margin
302 994
976 001
286 811
−252 427
Kontekst: FORM: RZIS 12M
Add the user andrzej.nowak@mojafirma.com, assign them to the G_STANDARD group and give them read access to cost center 1301.
11.10.2026 20:50
› </> Tool calls (4)
OK
Zatwierdzono: fdx_user
⚠ Write operations to approve: 2
The agent wants to perform several write operations. Uncheck the ones you don’t want to run and approve the rest.
✓
Zaznacz wszystkie
✓
fdx_access · grant · CONTROLLING_DEMO
nieodwracalne
✓
fdx_access · grant · CONTROLLING_DEMO
nieodwracalne
The agent wants to perform a write operation. Check the arguments and approve or reject.
anthropic · claude-sonnet-5-5
Operator
✎ tryb zapisu
● Online
Work with specialized AI agents
In Fincorex®, working with AI is based on a clear division of roles and permissions. Three agents are at your disposal.
Your prompt
Agent 02 · Operator
Automates everyday tasks
Helps automate everyday controlling tasks: forms and rules.
Forms
Rules
Agent 03 · Architect
Manages the environment
Safely manages the controlling environment: applications, models and permissions.
Applications
Models
Your approval
AI doesn’t act on its own. The decision is always yours.
With the Human-in-the-Loop mechanism, every change to data, metadata or system settings requires human confirmation. The agents work strictly within the permissions of the user who uses them, so you get powerful support without the risk.
01
Prompt
02
AI proposal
03
Your approval
Tool: fdx_calc_from_base
The agent wants to perform a write operation. Check the arguments and approve or reject.
Impact preview
Mode
Recalculation with an engine rule – overwrites target cells
Application
CONSO_PL
Rows
121
Row sample
A#CTB5111, A#CTB5112, A#CTB5113…
Where
SV#Budget_Last; E#LE11; BS#BS16
Source
SV#Forecast_Last; E#LE11; BS#BS16
Factor
1,10
On rows
PLM_SALE, PLM_COS, PLM_C, PLM_DEPR, PLM_FIN, PLM_TAX
Target periods
P#202701 … P#202712
Less clicking. More time for decisions.
01 / 05 · Analyst
Generate variance analyses that land in your reports on their own
The system analyzes results in seconds and automatically places comments directly in the right fields of your template. One click and you’re done! A fully completed report goes straight to the Management Board, with no lost time and no extra formatting.
Line item
Act - Bud
%
Results commentary · AI
Revenue
3,094,059
16.4%
Actual PLN 21.96m vs. budget PLN 18.87m – favorable variance of +PLN 3.09m.
Selling costs
−1,293,357
−38.2%
Selling costs PLN 1.29m above plan.
Departmental costs
2,423,062
18.8%
Costs PLN 2.42m below budget, mainly thanks to personnel costs (+PLN 2.57m).
Sales margin
4,223,764
163.5%
Margin PLN 6.81m vs. PLN 2.58m in the budget.
✓ APPROVED · FDX_SET_CELL_COMMENT
Analyst
02 / 05 · Analyst
Choose the tone of the automatic commentary
With one click you set the tone and level of detail of the automatic commentary. The Management Board gets the essentials and a clear business takeaway for decisions, while operational managers get a precise variance analysis of their budgets.
Essentials · for decisions
December 2023: revenue PLN 3.09m (+16.4%) above budget, departmental costs PLN 2.42m below plan. Sales margin PLN 6.81m vs. PLN 2.58m in the budget.
Detailed analysis · accounts
Departmental costs: −PLN 10.48m vs. −PLN 12.90m in the budget (+18.8%). The largest savings are in personnel costs. Overruns in taxes and fees and in internal transactions.
Personnel costs
+2,568,556 · 25.2%
+2,568,556 · 25.2%
Taxes and fees
−59,702 · −141.7%
−59,702 · −141.7%
Internal transactions
−204,550 · −86.6%
−204,550 · −86.6%
03 / 05 · Operator
Modify the system without any coding
Forget about technical barriers. Want to add a new column, move rows or adjust a budget template? Describe it in a plain sentence. The Operator Agent lets you manage templates and rules using simple text instructions alone.
04 / 05 · Operator
Paste data from messy Excel files
Need to load data from an external file in a completely different format into a template? Just attach it to the conversation. Fincorex AI analyzes the file, catches any errors and, if needed, asks you a few short clarifying questions. Only checked, complete and verified numbers make it into your system.
XLSX
costs_march_branches.xlsx
3 sheets · 1,284 rows
File validation
Columns matched to the template
OK
12 rows without a cost center
NO COST CENTER
2 duplicate entries
DUPLICATE
Which cost center should the 12 rows from the “Kraków Branch” sheet be assigned to?
05 / 05 · Session history
Reuse previous chat conversations for recurring analyses
Don’t waste time typing the same questions again. Session history in Fincorex AI automatically preserves your analytical processes. When a new period comes, you only change the date in the question, so month-end close can run more smoothly, without stress or worry about missing key variances.
Saved sessions
P&L variances · Board
2026-01
P&L variances · Board
2026-02
P&L variances · Board
2026-03 · new
Comment on the variances in the IT GROUP income statement for 2026-03. Tone: for the Management Board.
Your AI. Your rules.
AI models
Integrate any AI model
Fincorex® can be integrated with ChatGPT, Claude or your own AI infrastructure. You use the solutions your company already has, in full compliance with security requirements.
ChatGPT
Claude
Your own model
Other
MCP standard
Connect your own AI agents
Already have your own AI agents? With the open MCP standard they connect securely to Fincorex®, pull current data and trigger actions, without involving developers.
Company
Your agents
MCP
EPM
Fincorex®
FAQ
Frequently asked questions.
How do the Analyst, Operator and Architect agents differ?
The Analyst analyzes data and writes variance comments that go straight into the template. The Operator helps automate everyday controlling tasks, i.e. templates and rules. The Architect manages the controlling environment: applications, models and permissions.
Can an agent change data in the system on its own?
No. Thanks to the Human-in-the-Loop mechanism, every change to data, metadata or system settings requires human confirmation. The agent proposes an operation and shows its effects, and you approve or reject it.
What data can the agents access?
Agents work strictly within the permissions of the user who is using them.
Do I need programming skills to change templates?
No. The Operator Agent lets you manage templates and rules with simple text instructions, e.g. add a column, move rows or adjust a budget template.
Which AI models can Fincorex® be integrated with?
Fincorex® can be integrated with ChatGPT, Claude or your own AI infrastructure. You use the solutions your company already has.
Can I connect my own AI agents?
Yes. With the open MCP standard, your agents connect securely to Fincorex®, pull current data and trigger actions, without involving developers.
See your model
before you decide.
As a proof of concept, we’ll rebuild your reporting or planning model in Fincorex® and show it to you at our next meeting.
Get in touch