Case study · Automated fund reporting
From 20+ hours of manual work to a hands‑off monthly pipeline.
How Amula AI rebuilt bilingual factsheet production for IST Investmentstiftung as one governed Power BI pipeline in IST's own design: 39 investment groups in scope, German and French, automated PDF output. The system is live in production, and the rollout across the remaining investment groups is in progress.
At a glance
20+ hrs → 3–4 hrs
Total monthly effort to produce the factsheets, before and after: an internal estimate based on first measurements
39
Investment groups in scope across the mandate, rolled out group by group; rollout ongoing
DE + FR
Every factsheet in German and French, generated automatically
~2 months
From strategy to the first live production pipeline
Live and in production since the end of October 2025; the rollout across the remaining investment groups is ongoing.
One shared data model instead of one dataset per report: a single source of truth.
Near-zero manual steps in monthly PDF generation.
Client & context
A Swiss pension investment foundation with no tolerance for wrong numbers.
IST InvestmentstiftungIST Investmentstiftung is a Swiss investment foundation that manages assets on behalf of pension institutions. Its mandate with Amula AI covers 39 investment groups (Anlagegruppen), each of which publishes a monthly factsheet in German and French for professional and institutional investors.
In a regulated pension context, these documents have to be accurate, consistent, on-brand, and on time, month after month. The risk that matters most is a wrong figure reaching investors without anyone noticing, and there is no tolerance for it.
The challenge
A critical monthly deliverable running on manual labour and tribal knowledge.
Before the rebuild, the process was manual, fragmented, and high-risk:
Heavy manual effort
Producing and checking the factsheet PDFs took an estimated 20+ hours of work in total every month: repetitive assembly, formatting, and review work that grew with every additional investment group.
The bilingual multiplier
Every factsheet exists in German and French, which doubled the production and review surface.
Fragmentation and drift
Reporting logic lived across many separate files, so formatting and calculation rules drifted apart over time. Keeping them consistent had become a maintenance liability.
Details that cost credibility
Swiss number formatting, the client's mandated corporate font, and the language handling in charts and figures: small details that, when wrong, undermine credibility with investors.
Wrong figures that nobody notices
The dangerous errors in financial reporting do not announce themselves: a value scaled wrongly, a number treated as text, a missing data row that leaves a performance figure blank. Each can reach investors without anyone noticing, and the process had no systematic defence against it.
Manual, brittle PDF output
Final delivery depended on manual export steps that were slow and easy to get wrong.
The approach
Reporting rebuilt as governed infrastructure.
Amula AI treated reporting as infrastructure: a governed pipeline, designed once, that now runs every month with minimal human intervention. Five pillars:
- 01
A governed data foundation
Volatile, multi-source workbooks were turned into reliable inputs for the reporting layer. Scaling, formatting, and display rules were encoded once, centrally, instead of being repeated in every file.
- 02
One data model, many thin reports
Instead of one dataset per report, Amula built a single shared Power BI data model that feeds a lightweight report per investment group and language. German and French ship as separate, correctly localised reports. Adding an investment group is a configuration task, not a rebuild, which is how the rollout across the remaining groups proceeds.
- 03
IST's own design, Swiss conventions built in
The factsheets keep IST's corporate identity: layout, typography, and the client's mandated corporate font. The pipeline enforces Swiss number formatting and the correct German and French conventions in every figure and chart, so each factsheet is consistent and on-brand without manual fixing.
- 04
Hands-off PDF production
Monthly output is automated end to end. Data refreshes through an authenticated, automated process inside IST's own Microsoft tenant, and the final PDFs come out of Power BI's native export, so the corporate font renders exactly as mandated. Which reports run, in which languages, and on what schedule is configuration, not code.
- 05
Engineering for trust
The build ran in numbered phases with hard validation gates and a separate quality-assurance layer: the builder never signs off its own work. Every deliverable is checked for the errors that would otherwise reach investors unnoticed: wrong scaling, numbers treated as text, incomplete data rows.
The objection
“But our reports are designed, not generated.”
Many asset managers have their annual report, and often their factsheets, built in InDesign by a designer, where the visual result depends on precise typographic control. It is a fair objection to any automation offer. Here is how it is handled, and how you can verify it before committing.
The design stays
Your design is not replaced. The existing layout is rebuilt as a template, and the data flows into it. The output keeps your corporate identity, including typography and fonts.
What changes is the copy-paste
What changes is not who owns the design. It is that figures no longer have to be copied and pasted into it by hand.
Designer-grade pages stay with the designer
Where specific pages need full designer-grade control, those pages stay with your designer, and the automated system feeds them clean, final data instead of a spreadsheet to retype.
Amula AI does not promise pixel-identical reproduction of arbitrary InDesign layouts. It proves the result instead: in a proof of concept, up to 10 key pages of your report are rebuilt in your design, with dummy data, so you can judge the output before committing to a full build.
Architecture
One governed pipeline, end to end.
Source data
Excel workbooks: key data, performance, structure
Governed file store
inside IST's own Microsoft tenant
One shared Power BI model
scaling, formatting and language rules, defined once
Reports per investment group
German and French
Automated monthly run
authenticated refresh and PDF export in the client's tenant
Bilingual monthly factsheets
DE + FR, on schedule
Results & impact
What the live system has changed so far.
Headline outcomes
- Monthly reporting effort down from 20+ hours to 3–4 hours, an internal estimate based on first measurements.
- One reporting system for all 39 investment groups, rolled out group by group; rollout ongoing.
- German and French factsheets generated automatically for every live investment group, removing the duplicate manual workload.
- First production pipeline live roughly two months after the strategy phase began.
What it means
- One source of truth
- A single shared data model replaced one dataset per report, ending the drift between files.
- Scales by configuration
- Adding an investment group or a report is a configuration change, not a rebuild. That is how the remaining groups are being rolled out.
- Defence against silent failure
- The failure mode that matters most in financial reporting, a wrong figure reaching investors without anyone noticing, is checked for explicitly rather than assumed away.
- Owned by IST
- The pipeline runs inside IST's own Microsoft environment and belongs to IST. The system is owned, not rented from a platform.
Why it holds up
The engineering principles behind the build.
Architecture before tooling
A shared model plus thin reports eliminates drift across many files.
Silent failure is the primary risk
Mis-scaled values, numbers treated as text, and incomplete rows are flagged explicitly, not assumed away.
Data normalisation belongs at the source
Raw values plus format rules, not pre-formatted strings, keep the model clean and auditable.
Configuration, not hardcoding
Which reports run, in which languages, and on what schedule lives in configuration. A change is a setting, not a code change.
Phased delivery with QA gates
The builder never signs off its own work.
Tech stack
Built on what the client already runs.
IST already ran on Microsoft, so that is what we built on. The same approach works on any stack: another BI tool such as Tableau, open-source tooling or fully custom code, in the cloud or on your own servers.
What it runs on
See it in your own design before you commit.
Proof of concept: Amula AI rebuilds up to 10 key pages of your report in your design, with dummy data, so you can judge the output before committing to a full build.