FlexPulse-EU
Research methodology, built into an agentic workflow.
FlexPulse-EU turns a behavioural research framework into a multilingual survey-to-analysis platform. It connects questionnaire preparation, human review, response collection, deterministic scoring and interactive exploration in one traceable workflow.
- Study participants
- 1,000
- Countries
- 3
- Survey languages
- 3
The challenge
Energy-flexibility research brings together behavioural concepts, practical household conditions and different forms of declared preference. Moving from those concepts to a questionnaire, and from answers to usable analysis, requires a consistent connection between research design and software implementation.
What we built
We built an integrated platform around the FlexPulse Behavioural Schema. An agentic workflow assists questionnaire preparation and multilingual refinement, while researchers retain control through review and editing.
Published surveys preserve their measurement and mapping contracts. Responses are then processed through deterministic rules into scores, device-specific capability indicators and detailed facets, available through the analytics dashboard and interoperability interfaces.
Beyond the state of the art
-
Generative where it helps, deterministic where it counts
Language models plan, write and translate the survey instruments; scoring runs on explicit, versioned rules. Results are reproducible and auditable, something generative pipelines alone cannot guarantee.
-
Schema-grounded agents
Agents work against a formal behavioural schema, with automated validation and expert review at every step, so each generated question stays tied to the concept it measures.
-
End to end, open and interoperable
Research framework, published survey, scoring, analytics and API in one traceable system, released as open source and listed on the O-CEI marketplace.
Architecture
- 01Behavioural schemaResearch framework
- 02Assisted survey preparationAI-assisted
- 03Human reviewResearchers
- 04Published measurement contractsVersioned
- 05Response collectionMultilingual
- 06Deterministic scoringExplicit rules
- 07Analytics and APIDashboard · Interfaces
The architecture keeps behavioural concepts, survey questions, scoring rules and analytical outputs connected throughout the workflow.
In use
The platform was used to conduct a multilingual study with 1,000 participants across Spain, France and Ireland. The study exercised the complete survey-to-analysis workflow and provided evidence for interpreting the framework’s measures and descriptive profiles.
The project completed its O-CEI Stage 3 evaluation successfully, with all four technical performance indicators met.
Resources
Source code licensed under Apache-2.0.
Funding acknowledgement
Developed within the O-CEI Open Call, Pilot 1 / Challenge P1C2.
This work was framed in the context of the project OCEI, which receives funding from the European Union’s Horizon Europe research and innovation programme under grant agreement 101189589.
Views and opinions expressed are, however, those of the author(s) only and do not necessarily reflect those of the European Union. Neither the European Union nor the granting authority can be held responsible for them.