Skip to content
Case studies · Illustrative

Examples of what we build.

These are illustrative cases: composite examples of the kind of work we do, not named clients. We explain each system's problem, architecture and operation.

Showing 4 of 4 case studies

01
Calculation engine

Illustrative case

Expert knowledge as an API

Calculation rules available to other systems, with a record of every run.

Problem

Critical calculations were in legacy spreadsheets. They lacked controls to audit changes and versions. Maintenance depended on specific people.

Solution

We moved the rules to an API: an interface that other systems can query. We validate and version each rule. We record each run. We can expand the catalog without changing the core.

Current operation

  • Every result is traced: what was calculated, with which data and which version
  • The catalog grows without touching what is already validated
  • Where it fits: technical calculations, rate setting, quotes, scoring

Origin: standards-based calculation

View architecture
The challenge

Calculations based on technical standards were in spreadsheets. Auditing rules, controlling versions and reusing them across projects was difficult.

Architecture
Request with the dataValidated, versioned rulesA trace of every runResults for any system
Result

Every run is traced, and the calculation catalog grows without touching the core.

View case study
02
Document generation

Illustrative case

Automated technical documents

From your data to a document for review, with no manual transcription.

Problem

The team wrote reports, proposals and technical memos by hand. Copying data caused errors. Experts and writers needed several review rounds.

Solution

The system receives and validates data. It generates a document for review with a consistent format and stable formulas. It requires no manual transcription.

Current operation

  • The same data and rules generate the same document, without manual transcription
  • Validation against reference templates
  • Where it fits: proposals, recurring reports, contracts, technical documentation

Origin: technical memos

View architecture
The challenge

The team wrote technical memos in Word. Revisions could alter formulas and require another check.

Architecture
Request with the dataValidationCalculation engineWord document with editable equations
Result

The system generates a Word document with editable equations. The team reviews the document before delivery.

View case study
03
Opportunity monitor

Illustrative case

Automated market monitoring

Filtered and prioritized opportunities, delivered daily.

Problem

The team manually reviewed dozens of portals, bulletins and official websites. Some sources were missed. The team received false positives and late summaries.

Solution

The system checks sources daily and scores opportunities using the team's criteria. AI summarizes the selected opportunities. The team receives summaries by email or messaging.

Current operation

  • Coverage: official portals, bulletins and multilateral bodies, extendable by country
  • Quality filter: only the highest-scoring opportunities consume AI resources
  • Where the same technique fits: any daily flow of outside information that needs to be filtered and prioritized

Origin: international tenders · daily delivery

View architecture
The challenge

Manual tracking of dozens of public sources, with false positives and summaries that arrived too late.

Architecture
Daily source checksScoring against your criteriaAI summaryEmail + messaging
Result

The filter selects and prioritizes opportunities before calling the model. Only selected opportunities proceed to summarization and delivery.

View case study
04
AI assistants

Illustrative case

Private AI for your team

Built into the tools your team already uses, with the data in your own environment.

Problem

The team needed conversational AI in its own cloud, integrated with internal tools. The products evaluated did not offer the required data terms.

Solution

We deployed assistants with memory, versioned functions and standard connectors. In your own cloud and in the EU: your data stays in your environment.

Current operation

  • It remembers the context: sessions and persistent memory
  • Capabilities tested end to end, with a quality threshold before they are switched on
  • Where it fits: internal support, operations, company knowledge

Origin: an assistant in the team’s internal chat · the client’s cloud

View architecture
The challenge

Conversational AI without giving up data privacy or forcing changes on their internal tools.

Architecture
Persistent sessionsVersioned capabilitiesStandard connectorClient environment
Result

We measure and adjust the assistants based on real use. In your own cloud and in the EU: your data stays in your environment.

View case study

Related articles

Frequently asked questions

What people ask us.

Could this improve a process in my company?

We assess processes with repeated calculations, rule-based documents, information classification and assistants that use internal knowledge. On the first call, we check whether these solutions fit your case.

How long does it take to get something similar?

A focused tool takes about two weeks. A full-scope tool takes four to six. Document automation has a three-to-four-week pilot. In the AI integration service, we prepare the report first. The pilot then takes four to six weeks. We confirm the timeline for your scope in the proposal.

Your case?

Let us review your process.

If your problem looks like any of these, the first conversation comes with no commitment.