TevaPath ESIA – S0 prototype

Step 1 – Screening mode and scenario context

This step defines the overall context for the screening: which project you are working on, how the screening will be used, and under which “mode” it should run. The fields include:

Screening mode – prototype setting indicating whether the user intends a quick indicative check, a full IFI-style screening, or another mode. At this stage all modes use the same questions and AI logic. The value is stored with the dataset and will be used in future versions to adjust the depth of analysis and target standards.
Project reference – short internal identifier that links the screening run to a specific project (e.g. project code or internal ID).
Scenario reference – optional code describing a particular variant or scenario (e.g. “Option 1 – greenfield”, “Option 2 – expansion”). This is reserved mainly for future iterations where several scenarios will be compared across screening and scoping modules.

Implementation note: all fields on this step are currently stored and passed to AI as part of the screening dataset, but they do not yet change the screen flow or AI model selection. They are kept as a stable context for future multi-run / multi-scenario features.

Step 2 – Project passport

This step collects the core “passport” data for the project. These fields are used directly by the AI engine to reconstruct a clear project profile and to ground the screening conclusions.

Project name – official or working title of the project.
Country / region / city – basic geographic location sufficient to understand the context (urban / rural, national jurisdiction). No GIS analysis is performed at this stage.
Project owner / sponsor type / contact person – who is responsible for the project and what type of institution is behind it (public, private, PPP, FI, etc.). Used mainly for context.
Sector and subsector – classification of the project (e.g. waste, energy, transport, water, industrial, social infrastructure). This is a key input for AI categorisation and topic selection.
Project stage – where the project currently is in the cycle (concept, design, construction, operation, expansion, decommissioning).
Investment volume – indicative order of magnitude of investment (can be approximate, but helps understand scale).
IFI involvement flag – indicates whether an international financial institution is (or may be) involved.
IFI profile – simple dropdown describing the type of IFI (e.g. EBRD, IFC, World Bank, other IFI). This field is currently used only as context for the AI and may later drive more specific references to standards.

Implementation note: all fields on this step are fully implemented and included in the AI input. No dummy fields are present here; however, IFI profile is currently descriptive only and does not yet switch between detailed IFI-specific rule sets.

Step 3 – Project type and scale

This step defines what kind of activity is being screened and how large it is. It also captures some basic regulatory context (e.g. Annex-type classification) and the presence of transboundary aspects. These inputs are central for risk profiling and indicative category selection.

Project type – dropdown listing typical project types (e.g. landfill, MBT plant, transfer station, industrial facility, energy project, etc.). This helps the AI align with typical IFI sector categories.
Annex class – simplified representation of national/EU-style Annex I / Annex II / “no Annex” classification. Currently used as a qualitative signal of likely EIA requirements; in future it may drive more detailed regulatory logic.
New site or modernisation – indicates whether this is a greenfield project, expansion or modernisation. When “new site” is chosen, the “additional capacity” field is hidden, as it is not applicable.
Primary capacity – main indicative capacity parameter for the project (e.g. tonnes per year, MW, m³/day). A short hint explains what to enter.
Additional capacity – incremental capacity for expansions/modernisation. Hidden for pure greenfield projects. This field is implemented, but treated as secondary by the AI.
Transboundary flag – simple yes/no flag indicating whether potential transboundary impacts are expected. Used by the AI as a specific IFI-style trigger.

Implementation note: all fields on this step are implemented and used in AI analysis. The Annex class and transboundary flag are currently simple qualitative triggers, but may later be linked more tightly to specific IFI and Espoo-type requirements.

Step 4 – Territory and context

This step describes where the project is located and how it relates to settlements, land use and sensitive areas. It replaces detailed GIS analysis at this stage and provides the AI with a high-level spatial and social context.

Location summary – free-text overview of the location (e.g. “greenfield site 10 km from a regional town, surrounded by agricultural land”). This is the primary narrative for context.
Settlement setting – dropdown or classification of the setting (urban, peri-urban, rural, remote). Used by the AI to understand receptor sensitivity.
Nearest settlement name / distance – basic information on the closest settlement and its distance from the project site.
Protected areas flag / description – whether any protected areas or critical habitats are present near the site, and a short description if yes.
Sensitive receptors flag / description – presence of schools, hospitals, dense residential zones, culturally important sites, etc., with a short description.
Current land use – main land use type at and around the project site.
Land take flag – whether the project requires new land acquisition.
Physical displacement / economic displacement – qualitative indication of expected resettlement or loss of livelihoods. These fields are important triggers for IFI resettlement and livelihood restoration requirements.

Implementation note: all fields on this step are implemented and included in AI analysis. No GIS files or maps are loaded at this stage; the tool relies solely on the textual and categorical information entered here. Future versions may add optional map-based features, but they are not active now.

Step 5 – Key topics and risk profile

This step captures the user’s expert judgement on which environmental and social topics are most relevant, and how strong the overall risk level appears to be. It is a key input for prioritising issues in the AI-generated analysis.

Overall risk level – user’s qualitative assessment (e.g. low / medium / high). This is not a formal IFI classification, but it guides the AI when interpreting the rest of the data.
Environmental topics flag + list – indicates whether there are significant environmental topics and allows the user to list them in free text (e.g. air, water, waste, noise, odour, etc.).
Social topics flag + list – same for social issues (communities, land, livelihoods, vulnerable groups).
Labour & OHS topics flag – whether there are notable issues around labour conditions or worker safety.
Biodiversity topics flag – whether biodiversity/critical habitat is an important theme.
Climate topics flag – whether there are relevant climate change or climate-risk aspects (mitigation and/or adaptation).
Cultural topics flag – presence of cultural heritage or related sensitivities.
Other topics – free-text space for any other cross-cutting or context-specific issues.

Implementation note: all fields on this step are implemented and used. The flags and lists are treated as user-driven scoping hints: the AI uses them to build the list of main issues and to calibrate the overall severity profile. Future versions may use these flags to dynamically adjust which follow-up modules and tools are offered after screening.

Step 6 – Supporting information

This step allows the user to add any extra narrative context and, in future, to attach supporting documents. For the current prototype the focus is on textual comments, while file uploads are technically prepared but not used in the AI analysis.

Additional narrative information – free-text field (up to approximately 1,000 characters) where the user can record any clarifications, caveats, or special circumstances that were not captured by previous steps. This text is fully included in the AI input and can significantly influence the analysis.
Upload slots (GIS / environmental / social documents) – reserved fields for future use, intended to store key supporting files (maps, baseline reports, EIAs, etc.). At this stage they are either hidden in the interface or stored only as technical metadata and are not read or analysed by the AI.

Implementation note: “Additional narrative information” is implemented and actively used in the AI analysis. Upload-related fields are placeholders for future document-ingestion features; they may be visible or partially visible in the UI but do not yet influence the automated screening result.

Step 7 – Screening result and recommendations

This step presents the consolidated screening result. It contains a compiled snapshot of all input data, fields for the AI-generated summary and recommendations, and navigation buttons for further processing.

Compiled input summary – an automatically generated text field that lists the collected inputs (by field name and value). It is used as a transparent “check protocol” and is also passed to the AI engine as part of the screening dataset.
Summary of screening result – long text field populated by the AI after you press Perform analysis. It provides a narrative conclusion and indicative category. The user is free to review and edit this text.
Main issues – multiline field where the AI lists and explains the priority topics (usually as bullet points). This can be adjusted by the user before sharing with clients or IFIs.
Next steps – multiline field with recommended follow-up actions (scoping, studies, plans, etc.). One of the recommendations deliberately points to proceeding to ESIA scoping and a structured E&S gap analysis against IFI and national requirements. The user can edit this list as needed.

Buttons:

Perform analysis – sends the structured dataset to the AI engine, shows a short “Processing…” indicator, and then fills the result fields. This is fully implemented.
Start over – clears the current in-browser dataset and returns the user to the beginning. This is implemented as a client-side reset and does not affect any external storage.
Proceed to Scoping – moves user to the dedicated scoping module. All information collected on screening phase passed to the next module.
Prepare the Report – reserved for future automated report generation (screening note / ESIA summary). Currently implemented as a simple placeholder alert, without producing files.

Implementation note: all result fields and the “Perform analysis” logic are fully functional. The scoping and reporting buttons are dummy placeholders reserved for next phases of TevaPath, when automated scoping sheets and structured screening reports will be added.

Use screening results to fill the fields (recommended): Yes No

Scoping step 1 – Scenario selection & start

This step defines the high-level scoping context: indicative ESIA/EIA requirement, IFI profile, associated facilities and cumulative impact focus. These fields will later guide the structure and depth of the scoping note.

There in an option to fill some fields with a narrative summary coming from screening phase output. If you decide to use this option, please adjust the text to get more focused scoping results.

Scoping entry summary – short narrative introducing how this scoping exercise will be used (e.g. “EBRD concept review”, “internal Go/No-Go”);
Indicative ESIA/EIA requirement – early view on whether a full ESIA, focused ESIA or national EIA is expected;
IFI / standard profile – which IFI or ESG framework should be used as reference;
Associated facilities and cumulative impacts – flags and brief descriptions supporting later scoping decisions.

Scoping step 4 – Social topics

This step mirrors the environmental scoping but for social, labour and community issues. The narrative summary comes from screening output "as is". Please adjust the text to get more focused scoping results.

In S0 it is captured as narrative fields:

Social scoping summary – overview of key social and labour issues driving the scope;
Priority social topics – ordered list (e.g. land acquisition, livelihoods, community health, vulnerable groups, security) with rationale;
Indicative social baseline / engagement programme – planned surveys, consultations and stakeholder engagement elements.

TevaPath ESIA – S0 Screening & Scoping prototype