Each planning round follows the same pattern. Distributing templates, collecting values, reconciling top-down and bottom-up approaches, and manually reworking many aspects after each cycle change. This is exactly where AI changes the picture in integrated corporate planning. Not as a number-cruncher, but as a preparer who takes the drudgery off the human’s hands.
This third part of the series presents a vision and honestly distinguishes what is already possible today and what lies ahead. Two ideas are at the forefront: planning through natural language and a forecast that continuously prepares itself.
What does planning look like today, and where does it get stuck?
Most planning processes rely on templates. Each cost center receives its table, fills it out, and sends it back. After that, the reconciliation between the directives from above and the reports from below begins. That costs rounds and time.
The most expensive part is the cycle change. When switching from plan to forecast or to the next version, a lot of work starts all over again. Transfer values, adjust templates, reconcile statuses. Exactly this manual work is the lever for AI in controlling.
What does planning by natural language mean?
It’s not about chatting for the sake of chatting, but about real planning in plain German. Instead of opening masks, searching for fields, and filling out templates, you say or write what needs to be planned. The AI understands the instruction, converts it into specific values per account and month, and enters them into the system upon your approval.
The difference is the hurdle. Whoever has the numbers in their head but doesn’t like the planning interface can still plan in minutes. No clicking through menus, no searching for the right line.
“Plan travel expenses for the Sales cost center ten percent above last year for the second half.”
“Add an additional full-time position in Development starting in March, at the standard salary.”
“Reduce material costs by five percent and spread the effect evenly across the months.”
“Show me how the result changes if revenue increases by eight percent in the fourth quarter.”
Planning in natural language doesn’t just mean entering values. You can just as easily adjust, distribute over months, run through scenarios, compare versions, or apply the same change to multiple cost centers. And you can ask back, for example, which cost center is the most over budget. The AI responds with the number and suggests the next step.
Before anything is posted, the AI displays the derived values for review. If an instruction is unclear, it asks for clarification. Only after your approval do the numbers go into the system. This way, the planning stays with you, while the AI takes care of the hard work.
How does AI continuously prepare the forecast?
The second idea works in the background. A process continuously inputs new actual data and updates the forecast accordingly. He calculates a forecast based on the trend and takes into account the top-down guidelines as a framework, ensuring that the preparation remains within the scope of the company’s objectives.
The effect is noticeable during the cycle change. The user opens the system and finds the specifications already prepared. Instead of starting from scratch, he reviews a well-founded proposal and makes only minor adjustments. From a planning round, it becomes an approval round.
Where does most of the manual work hide in your planning process?
The first step is rarely the big vision – it’s the one recurring task that eats up time every cycle. In a free skill check, we’ll find the biggest lever together.
Which of these is already possible today, and what is the target vision?
A target image is only as valuable as the honest assessment behind it. This table distinguishes what is already in progress today and what lies ahead of us.
We are already implementing the writing back of an AI forecast into the cube. The planning in discussion is a target image that we are working on. The path to that is not a leap, but a chain of individual steps, of which the first ones are already in place.
Why does the human remain the control instance?
An AI that silently changes planned figures has no place in controlling. Therefore, at the beginning of every meaningful structure, there is a clear principle. The AI prepares, the human decides.
Three things ensure that. The top-down guidelines act as a firm framework that keeps the preparation within bounds. Each prepared value is accompanied by a justification that can be verified. And nothing goes into production before it is released. So the responsibility stays where it belongs.
How do you get there step by step?
The path to the target image follows a logical sequence. Each step brings its own benefit, even if the next one is still pending.
- Get the data foundation in order. A clean cube is the foundation for everything else.
- Start with reading skills. Commentary and deviation analysis quickly provide benefits without risk.
- Introduce writing skills with approval. The forecast write-back is the first step where AI becomes active.
- Define guidelines and governance. What guidelines apply, who gives approval, what is documented.
- Pilot the planning using natural language in a clearly defined area, such as a cost center group.
Glossary
Frequently Asked Questions about AI in Business Planning
What does planning in natural language specifically mean?
You say in plain German what needs to be planned, such as travel expenses ten percent above last year or a new position starting in March. The AI translates this into values per account and month, displays them for review, and only enters them after your approval. You can also adjust, distribute, simulate, and query.
Does the AI independently enter planned figures into the system?
No. The AI prepares values and justifies them. Nothing becomes productive without approval. The decision remains with the human.
Which of these is already possible today?
The connection to the Cube and the writing back of an AI forecast are already in progress. The ongoing integration of new data is in progress. The planning in discussion is a target image.
How are the top-down directives maintained?
They act as a firm guideline. The AI preparation operates within these guidelines, ensuring that departmental planning aligns with the company’s goals.
Where is the best place to start?
With the data foundation and a reading skill. Whoever sees value there, will gradually move toward the target image through writing skills with approval.
Why Blue Bee Intelligence is the right partner
The path to AI-supported planning goes through the systems where your planning is currently running. We connect Claude with IBM Planning Analytics and the reporting platform BlueHive and build the skills that gradually relieve the planning process.
We have already implemented the first step, writing back an AI forecast into the cube, ourselves. So you don’t start with an idea, but with a building block that is already in place.
