Plan, make and supply medicines on one model. Agents find what changed, every step on the record.
Fifteen agents work with a pharmaceutical company's finance, supply chain, quality and commercial teams for Meridian Therapeutics, a fictional company in Australia and India, using sample data. IBM Planning Analytics holds the plan and calculates every number, from demand and plant capacity to expiring stock and the group P&L. Fixed rules decide what needs a person. Agents explain what moved and model what to do about it, in scenarios only. People approve every change to the plan.
Read, flagged, explained, modelled. Then approved.
Every step is labelled Agent, Planning Analytics, Rules or Person. Agents write only to three scenario slots, never to the actuals, the budget or the forecast.
Read
IBM Planning AnalyticsThe month's actuals land in the model: sales, production, stock, quality and cash, for Australia in dollars and India in rupees.
Flag
RulesFixed rules raise a case when something crosses a line: a plant short of capacity, stock close to expiry, one supplier carrying too much, an overdue fix.
Explain
AgentsWhat moved, by how much and why, in three plain lines: timing or structural, who owns it and when to check again.
Model
Agents and Planning AnalyticsAn agent sets the drivers of a scenario. Planning Analytics recalculates the whole plan, plants and P&L, next to the forecast.
Approve
A named personThe CFO, COO or the right lead approves or rejects, with a note. Nothing changes the forecast without them.
Promote
PlannerA person moves the approved scenario into the forecast. Next month starts from what was decided and recorded.
Short shelf lives, tight plants. One plan.
Medicines expire, plants are regulated and prices are set by governments. The desk is built around those facts.
Expiry and cold chain
Stock by batch and expiry date at every distribution centre, with the moves that save the most before a write-off.
Plants and quality
Capacity, utilisation and output by plant and line, next to deviations, CAPA and batch release.
Price pressure
Rebates, tender discounts and government price cuts, from list price down to net sales and margin.
Two countries, one group
Australia in dollars, India in rupees, and the group consolidated in Planning Analytics, with FX and tax.
A month at Meridian. Nine decisions.
Each story runs in the Pharma Desk with the real agents and Planning Analytics, on sample data.
A flu season comes early
Respiratory demand rises 20%. Can the plants cope, what would outsourcing cost, and what happens to profit and cash?
Stock about to expire
Vaccines and cancer medicines near expiry in two warehouses. The agent models the best moves and the write-off they save.
A supplier goes down
The main API supplier stops for six weeks. Do nothing, or re-source most of it at a higher price: both modelled side by side.
Margins wearing thin
Older medicines lose margin to tender discounts and a government price cut. The agent finds them and models a price recovery.
Too many warehouses
Two distribution centres run half empty. Close them, and see the saving, the cost to close and the risk to on-time delivery.
A trial runs late
A phase III medicine slips a year and its odds fall. The agent shows what that does to the pipeline's value.
An inspection is coming
Deviations spike at one plant before a re-inspection. The quality agent shows the open fixes and who owns them.
The rupee weakens
India's results translate into fewer dollars. Finance sees the effect on the group, line by line.
The monthly planning cycle
Demand review, supply review, the numbers reconciled, then an executive decision between three options, recorded with the approver.
Eleven specialists. A finance lead and three planners.
Each specialist covers one part of the value chain and reads the same Planning Analytics model, so the numbers always agree.
Specialists
Develop, plan, source, make, deliver, sell, comply, governR&D Agent
Explains the pipeline: phases, probability of success, launch timing and risk-adjusted value, and what a delay would cost.
Demand Agent
Explains demand and revenue by product, therapy area, market and channel, against budget and last year.
Procurement Agent
Shows API, excipient and packaging spend by supplier, and how much rides on a single source.
Production Agent
Checks plant capacity and utilisation, finds the months a plant runs short, and what outsourcing would cost.
Quality Agent
Explains batch release, deviations, CAPA and inspection readiness, plant by plant.
Inventory Agent
Finds stock expiring within 90 days and suggests where to move it before it has to be written off.
Warehouse Agent
Explains distribution centre throughput, utilisation, labour and cost per pick.
Logistics Agent
Explains freight, cold chain and on-time delivery, and what closing a distribution centre would do.
Commercial Agent
Explains government and hospital tenders, win odds and margin, in Australia and India.
Regulatory Agent
Tracks submissions, approvals and price controls, and the next deadlines.
Risk Agent
Explains the risk register: likelihood, impact and exposure, and what has changed.
Finance lead and planners
Model scenarios; a person approvesFinance Agent
The CFO's front door: group and country P&L, cash, tax, FX and variances. Models a scenario and asks a person to approve it.
Demand Planning Orchestrator
Runs the demand review: actuals, the plan and the statistical forecast side by side, with every override evidenced.
S&OP Orchestrator
Checks demand against plant capacity with the specialists, then builds three supply options as scenarios.
IBP Orchestrator
Reconciles the options to the P&L and prepares the executive decision. A named person approves.
The company, the cases and the plan.
Follow a month all the way through.
Company view
What needs attention for your role, eight key numbers against last month and last year, and the lists behind them, on one screen.
Variance commentary
Why a number moved, timing or structural, and what is being done about it. Recorded, so next month starts from it.
Operations Centre and cases
Every case raised by the rules, with the agent's brief, the evidence, the scenario and the approval trail.
Planning cycle
The four weeks of integrated business planning, each with its orchestrator, its output and its approver.
Live planning and what-ifs
Change a driver in a scenario and watch Planning Analytics recalculate the plan. Ask a what-if in plain English.
AI forecasting
Statistical forecasts tested against recent months, launch curves, loss of exclusivity and pipeline risk, published only with an approver.
Operational reports
Batches and expiry, service levels, cold chain, plant performance, forecast accuracy, quality events and gross-to-net.
Value case
The business case for the desk itself: bear, base and bull, the value of automation and the cost of a delay.
Planning Analytics
The PharmaFPA model behind every number, and Planning Analytics Workspace books for executives and operations.
The plan calculated. People in charge.
Fixed rules, not a prompt, so no agent can talk its way around them. Every agent step, scenario change and approval is in the audit trail.
Calculated by the engine. Not by AI.
Planning Analytics calculates every number. Agents read it, explain it and set scenario drivers; they don't do the maths.
Scenarios only. Never the plan.
Agents can write to three scenario slots and nothing else: never the actuals, the budget or the forecast, and never a past month.
Raised by rules. Decided by people.
Cases come from fixed thresholds, not an agent's opinion. Approvals need a named person.
Every step on the record.
Who asked, which agent answered, which tools it used, what changed and who approved it.
End to end, in the Pharma Desk.
See what needs attention, explain a variance, work a case from the rules to an approval, run the monthly planning cycle, change the plan in a scenario, forecast with AI and switch between Australia, India and the group.
- 01Actuals landAustralia, India, the groupPA
- 02Cases raisedFixed thresholdsRules
- 03Specialist agentsWhat moved and whyAgents
- 04Scenario modelledPlan recalculatedPA
- 05Leader approvesWith a notePerson
- 06Forecast updatedBy a plannerPerson