How we used Anaplan AI to model sustainable banking – and finished second at Planathon Sydney

The Planathon: A practitioner’s favourite

If you’ve spent time in the Anaplan ecosystem, you’ll know Planathon holds a special place in the event calendar. Planathon is where Anaplan practitioners actually build – live, under time pressure, with real judging criteria.

What is CoModeler?

Anaplan CoModeler is an AI-assisted model generation tool built directly into the Anaplan platform. Rather than manually configuring lists, modules, line items, and formulas one by one, practitioners submit a structured natural language prompt and CoModeler builds those objects for you – lists populated with relevant items, modules dimensioned correctly, and formulas written to Anaplan best practice standards.

This year’s Sydney Planathon raised the stakes considerably. The theme: Anaplan’s emerging AI capabilities, specifically CoModeler and Agent Studio – two tools that are reshaping how Anaplan connected planning models are built, managed and interacted with.

CoModeler understands the difference between classic and Polaris environments, generating code that respects the nuances of each. For experienced model builders, it doesn’t replace expertise, but dramatically cuts time.

What is Agent Studio?

Agent Studio is Anaplan’s new natural language interface for interacting with live planning models. Where CoModeler is a build-time tool, Agent Studio operates at run-time – allowing planners and executives to query their models, explore scenarios, and surface insights using conversational language rather than navigating module grids. It changes how non-technical users engage with planning data, lowering the barrier between the model and the decision-maker.

Our use case: Sustainable banking

Being a recently certified B Corp ourselves, the Zooss team was immediately drawn to a sustainability-aligned use case. We built a fictional organisation – SMAC Bank (Sustainable Modern Accountable Bank) – and designed a Business Banking Lending Revenue Forecast model centred on a values-based lending program.

The core idea: SMAC Bank offers a Green Loan product with a discounted interest rate for B Corp certified business customers. The discount is tiered by B Corp certification score; the higher the score, the better the discount. This structure incentivises companies to recertify and genuinely improve their sustainability performance year-on-year, not just achieve certification once and coast.

Scenario modelling allowed us to flex the discount rate assumptions and project the loan book’s growth trajectory, giving a CFO a real-time view of both the revenue cost and the strategic value of the green lending program.

Sustainability-linked finance is a real and growing part of business banking strategy, and having it modelled so cleanly inside Anaplan made the demo land well.

What we built

In just a few hours, we delivered a complete Anaplan model including:

  • Data loads – 20 Australian SME customers with B Corp scores, band classifications, and product assignments
  • Input modules – origination volume assumptions, tiered NIM rate assumptions, and green portfolio targets
  • Calculation chain – balance roll-forward, adjusted NIM by B Corp band, and monthly lending revenue (NII + fees)
  • Reporting outputs – revenue summary, green portfolio monitor with stock and flow KPIs, and a band analysis showing the revenue cost of each discount tier
  • UX dashboard – a demo-ready page surfacing the key metrics a Business Banking planning team would actually need

CoModeler: The wins

The tool delivered on its promise. Key observations from building with CoModeler:

Speed and accuracy. Lists and modules that would take 20-30 minutes to configure manually were generated in under 5 minutes. The output wasn’t just fast – it was structurally correct, with dimensions, line item formats, and summaries set appropriately.

Best practice by default. CoModeler outputs follow DISCO layering (Data, Inputs, System, Calculations, Outputs) and respects Hyperblock performance principles – summaries off in calc modules, correct dimension ordering. It builds the way a senior model builder would.

Environment awareness. The tool correctly distinguishes between classic and Polaris environments, which matters for formula syntax and module behaviour. This is non-trivial and reflects genuine platform depth in the underlying model.

It was exciting to see ideation turn into a working prototype so quickly – but don’t forget the ‘Co’ in CoModeler. It proved to be quite the accelerator, but doesn’t replace the thinking.”

Timothy Kirby
Principal Solution Architect and Master Anaplanner

The speed surprised even experienced builders. As Timothy noted, it was exciting to see ideation turn into a working prototype so quickly – but the bigger takeaway is that strong architecture and clear problem definition still matter going in.

CoModeler: The learnings

Preparation is everything. The teams who got the most from CoModeler came in with a clear plan. Our approach:

  1. Start with the end result. Define your output reports and UX wireframes before writing a single prompt. If you know what the CFO dashboard needs to show, you can work backwards to the exact modules and line items required.
  2. Segment the build. Rather than one large prompt, we wrote 15 individual scoped prompts – one per list or module – submitted in DISCO layer order. This minimised dependency errors and made it easy to resubmit a single prompt if CoModeler’s output needed adjustment.
  3. Save your prompts. Prompt documentation became our audit trail. It’s also reusable – our prompt library is something we can build on for future engagements.

On the day

Attendance was strong, with Anaplan’s Sydney customer ecosystem well represented across seven competitive teams. Zooss teamed up with colleagues from Macquarie Bank – a natural collaboration given the banking use case – and the partnership worked well under time pressure.

We finished second overall and are proud of what the team built.


Zooss is an Anaplan partner specialising in connected planning implementations. If you’re exploring what CoModeler or Agent Studio could mean for your planning model build, we’d love to talk.

Find out more

Better Planning. Better Planet.