TGIS/AI
← Technical LandscapeRapid Research

AI Tools & Chatbots

Existing AI assistants in the transport sector and the integration gap TGIS fills

AI Bots and Assistants in the Transport Data Sector

Last reviewed: 2026-02-13 Confidence level: Mixed -- some entries well-documented, others based on limited public information

Purpose

Several organisations in the global transport and development sector have built or are building AI-powered tools for querying transport data. This document catalogues what exists, what is planned, and what the limitations are. The framing question: if a Transport Global Intelligence System (TGIS) were to sit above these as an integrative layer, what would it connect, and where are the gaps?

1. ADB Asian Transport Outlook / Observatory

Operator: Asian Development Bank (ADB), in partnership with SLOCAT and AIIB URL: https://asiantransportoutlook.com/ Status: Active platform. No publicly confirmed AI assistant.

The Asian Transport Outlook (ATO), now rebranded as the Asian Transport Observatory, is ADB's flagship transport data initiative for Asia-Pacific. Launched in March 2021, it covers 51 economies with 450+ transport indicators across nine categories, plus ~1,100 policy documents, 260+ urban-level indicators, and 1,300+ transport project records. Phase 3 (2023--2025) focused on translating data into policy-relevant insights.

"Transport Genie"

The name "Transport Genie" appears in some internal and conference discussions as a potential AI query layer for the ATO database. I could not verify its public existence. Multiple searches across ADB sites, the ATO platform, and conference proceedings returned no public-facing chatbot, no documentation of an AI assistant, and no product page. It is possible that:

  • It exists as an internal prototype not yet publicly launched
  • It was discussed at ADB Transport Forum events (2024 or 2025) as a concept or demo
  • The name is informal or has changed

What the ATO data could support: A chatbot querying 450+ indicators across 51 economies would be a strong single-source tool for Asia-Pacific transport data. The structured, tabular nature of ATO data (national indicators, time series) is well-suited to natural language query interfaces.

Limitations even if it exists: Asia-Pacific only. National-level indicators only (no subnational, no geospatial). Single-source -- cannot cross-reference with climate risk, financial flows, or infrastructure condition data from other platforms.

Verification status

Unverified. If someone at ADB or SLOCAT can confirm the existence and status of an ATO AI assistant, that would fill a gap in this analysis.

2. GIZ TDCI AI Assistant

Operator: GIZ (Deutsche Gesellschaft fur Internationale Zusammenarbeit) Related platforms: Changing Transport (https://changing-transport.org/), MobiliseYourCity (https://www.mobiliseyourcity.net/), TUMI Data Hub (https://hub.tumidata.org/) Status: Referenced in project documents but not independently verified as a public product.

GIZ is the most active bilateral agency in transport data and digital tools. Their portfolio includes:

  • TUMI Mobility Data Hub -- a CKAN-based platform with data from 49 cities, 1,149 datasets, covering public transport, urban planning, and air quality. Operated with BMZ funding.
  • Changing Transport -- knowledge platform for sustainable transport, including the Transport and Climate Change Week.
  • MobiliseYourCity -- partnership supporting sustainable urban mobility planning in developing countries, including a data collection toolkit for local governments.
  • Moverity -- a digital platform for real-world fuel consumption data with a chatbot, developed for Thailand's Ministry of Energy and handed over in September 2020.

The "TDCI AI assistant"

The project's CONTEXT.md references "GIZ's TDCI AI assistant" as an existing tool. The TDCI (Transport Data Commons Initiative) newsletter is managed through GIZ's Changing Transport infrastructure. However, I could not find a publicly accessible AI assistant product with that name. Possibilities:

  • It may refer to a prototype or internal tool connected to the TUMI Data Hub or the TDC portal
  • "TDCI" in this context may refer to the Transport Data Commons Initiative broadly, with the AI assistant being a planned or early-stage component
  • It may be a tool demonstrated at GIZ transport events but not yet publicly deployed

GIZ does have demonstrated chatbot capability (the Moverity chatbot) and active AI exploration (Telekom-GIZ AI partnership for mobility), so building such a tool would be within their operational envelope.

Limitations if it exists: Likely focused on GIZ's own data holdings or the TUMI Hub's 49-city dataset. Urban mobility focus rather than full transport sector coverage. No evidence of cross-source querying.

Verification status

Not independently confirmed as a public product. The reference in project documentation suggests someone on the team has direct knowledge of it -- worth following up through GIZ transport contacts.

3. Transport Data Commons (TDC) / CKAN AI Chatbot

Operator: Transport Data Commons Initiative (30+ organisations), with platform built by Datopian on CKAN + PortalJS Portal URL: https://portal.transport-data.org/ Docs: https://docs.transport-data.org/ Status: Platform active. AI chatbot capability available via CKAN ecosystem, not yet confirmed as deployed on TDC specifically.

The TDC is a coalition of 30+ organisations (including UNECE, JRC, IIASA, GIZ, SLOCAT, ADB) working to create a shared, public transport data catalogue. The platform is built on CKAN, the open-source data management system used by many government open data portals worldwide.

AI chatbot via datHere / CKAN ecosystem

A significant development in the CKAN ecosystem is relevant here. datHere, a CKAN services company, unveiled an AI Chatbot in September 2025 called the "People's API" -- described as an "Answering People/Policy-Maker Interface." Key facts:

  • Uses a "zero-copy" approach: sends metadata context to the LLM rather than raw data, aiming for fast, cheap responses with lower hallucination risk
  • Can automatically identify, merge, and analyse data across an entire CKAN portal
  • Handles queries like "Which counties have young populations but high unemployment?" by merging datasets at query time
  • In 2026, being deployed across government portals, enterprise catalogues, and community data initiatives
  • Part of a broader "Data Librarian/Concierge" strategy
  • Built through a datHere-OKFN partnership announced December 2025

The connection to TDC: Since the TDC portal runs on CKAN (built by Datopian), the datHere AI Chatbot could in principle be deployed on it. Whether this has happened, or is planned, is not confirmed in public sources. The technical fit is there -- the question is whether TDC has adopted or will adopt this capability.

Limitations: Even with an AI chatbot layer, TDC queries only the datasets catalogued in TDC's CKAN instance. As of now, TDC aggregates data from UNECE, JRC, and IIASA primarily. It does not query external APIs in real time (World Bank, OPSIS, PortWatch, etc.). It is a catalogue, not a cross-source query engine.

Verification status

The CKAN AI Chatbot (datHere) is real and documented. Its deployment on TDC specifically is not confirmed.

4. World Bank AI Tools

The World Bank has several AI-adjacent tools, none specifically for transport data but relevant as precedents.

4.1 WITS Chatbot (alpha-bot)

URL: https://wits.worldbank.org/ Status: Active. Launched ~early 2023, one-year anniversary noted March 2024.

The World Integrated Trade Solution (WITS) features a chatbot called alpha-bot in the top-right corner of the page. Over 6,500 questions asked in its first year, with 75%+ answer rate.

  • Data search: Finds links to relevant data pages among millions of pre-aggregated trade statistics pages
  • Narrow search: Walks users through filter selection
  • Knowledge search: Answers methodology and definition questions
  • Tech stack: Microsoft LUIS + QNA Maker, Azure AI Language services

Transport relevance: WITS covers trade data, which intersects with transport (trade routes, port volumes, logistics performance). It is not a transport tool per se, but trade and transport data are closely linked. The Logistics Performance Index (LPI), hosted alongside World Bank data, is directly transport-relevant.

Limitations: Single-source (WITS trade data only). Rule-based / traditional NLP rather than LLM-powered. Cannot synthesise across World Bank datasets, let alone external sources.

4.2 PPP Resource Centre AI Chatbot

URL: https://ppp.worldbank.org/ppprc/ai-chatbot Status: Active.

An AI-powered chatbot for navigating the Public-Private Partnership Resource Centre's curated content on infrastructure finance. Searches across PPPRC content and other vetted World Bank Group sources.

Transport relevance: Infrastructure PPPs include transport (roads, ports, airports, rail). The chatbot helps users find guidance documents, case studies, and legal frameworks for transport infrastructure financing.

Limitations: Searches documents, not structured data. Content is about how to structure PPPs, not transport performance metrics. Single domain.

4.3 World Bank Data API -- MCP Server (community-built)

URL: https://github.com/anshumax/world_bank_mcp_server Status: Published June 2025. Community-built, not official World Bank product.

An open-source Model Context Protocol (MCP) server wrapping the World Bank Open Data API. Lets AI agents list countries, list indicators, and query indicator data. No API key required.

Transport relevance: The World Bank API contains hundreds of transport-relevant indicators (infrastructure investment, road density, logistics performance, safety statistics). This MCP server makes them directly accessible to AI agents.

Significance for TGIS: This is exactly the kind of thin wrapper that the TGIS concept describes -- making an existing API agent-accessible via MCP. It already exists for World Bank data. The TGIS approach would extend this pattern to OPSIS, PortWatch, OECD/ITF, and other sources.

Limitations: Covers World Bank indicators only. No geospatial capability. Built by an independent developer, not maintained by the World Bank.

5. ITF / OECD

URL: https://www.itf-oecd.org/ Status: No AI chatbot or assistant identified.

The International Transport Forum (ITF) at OECD publishes substantial transport policy research and maintains transport statistics via SDMX. They have published guidance on AI in transport policy ("AI for Transport Authorities: Principles and Practical Guidance") but do not appear to operate an AI-powered data query tool.

ITF data is available through OECD's statistical systems, which use SDMX -- a format that is machine-readable but not AI-friendly without translation (the query syntax is complex and the response format is XML-heavy). The TGIS concept document identifies OECD/ITF SDMX as a source needing an MCP server to make it agent-accessible.

6. Google Mobility AI

URL: https://research.google/blog/introducing-mobility-ai-advancing-urban-transportation/ Status: Active research programme, commercial offerings via Google Public Sector and Maps Platform.

Google Research's Mobility AI programme translates transport research into tools for agencies and planners. Key capabilities:

  • Multi-source data fusion (sensors, cameras, floating car data, user feedback)
  • City-scale traffic simulation for infrastructure planning
  • Demand prediction and route optimisation
  • Claims ~15% higher accuracy than traditional methods for urban traffic scenarios

Transport relevance: High, but focused on operational traffic management and urban planning rather than the development-sector policy questions TGIS targets. Works at city/corridor scale, not global comparative analysis.

Limitations: Commercial product, not open data. Focused on urban traffic, not cross-modal transport investment, safety, or climate risk. Not designed for the "how does Kenya compare to Tanzania on road safety investment?" type of question.

7. Other Notable Mentions

BESSER DataBot (University of Luxembourg)

An open-source platform for creating chatbots that answer questions about specific data sources. Can auto-generate a "bot swarm" to serve all data sources in an open data portal. Academic/research project. Could theoretically be deployed on transport data portals but has not been, as far as I can find.

DHL Gen AI / Commercial Logistics

DHL and other logistics companies are deploying generative AI for supply chain operations, shipment tracking, and customer service. These are commercial, proprietary tools for logistics operators -- not relevant to the development-sector policy use case except as evidence that the transport industry is adopting AI chatbots.

Summary Table

ToolOperatorData SourceScopeStatusCross-source?
ATO "Transport Genie"ADB/SLOCATATO database (450+ indicators, 51 economies)Asia-PacificUnverifiedNo
TDCI AI assistantGIZLikely TUMI Hub / TDCUrban mobility, Global SouthUnverified as public productNo
TDC + CKAN AI ChatbotTDC coalition / datHereTDC CKAN catalogueMulti-modal, emergingCKAN chatbot exists; TDC deployment unconfirmedWithin CKAN only
WITS alpha-botWorld BankWITS trade dataGlobal tradeActiveNo
PPP AI ChatbotWorld BankPPPRC documentsInfrastructure financeActiveWithin WBG docs
WB Data MCP ServerCommunityWorld Bank Open Data APIGlobal indicatorsActive (community)No
Google Mobility AIGoogleMulti-source urban dataUrban trafficActive (commercial)Within Google data
ITF/OECDOECDSDMX transport statsOECD countriesNo AI toolN/A

The Integration Gap

Every tool in this list queries a single data source or a single organisation's holdings. None of them can:

  • Cross-reference ADB transport indicators with World Bank financial data
  • Overlay OPSIS infrastructure condition data with IATI aid spending
  • Compare GIZ urban mobility data with ITF national statistics
  • Combine PortWatch trade disruption alerts with UNECE road network data

This is the gap TGIS would fill. The individual tools are useful within their domains. The value of an integrative layer is in the connections between domains -- the questions that require data from sources that have never been combined.

The pattern is consistent: each organisation builds a chatbot for its own data. Nobody builds the bridge.

What TGIS adds

The existing tools confirm that the individual components work -- AI querying of structured transport data is technically proven. What does not exist is:

  1. A shared discovery layer -- an agent that knows about all these sources and can route queries to the right one(s)
  2. Cross-source synthesis -- combining results from World Bank indicators + OPSIS geospatial + PortWatch trade data in a single analytical flow
  3. Geographic and temporal alignment -- translating between the different spatial and temporal reference frames each source uses
  4. Mode-bridging -- connecting road, rail, maritime, and air data that currently sit in separate institutional silos

The MCP server pattern (as demonstrated by the community-built World Bank MCP server) shows the technical approach: thin wrappers that make existing APIs agent-accessible. TGIS would apply this pattern systematically across the sector, with the Skills and registry layers providing the discovery and comprehension that no single wrapper provides.

Research Gaps and Next Steps

  • Confirm ATO AI tool status: Contact ADB transport team or SLOCAT to verify whether a "Transport Genie" or similar exists
  • Confirm GIZ TDCI assistant: Follow up through GIZ Changing Transport / TUMI contacts
  • Monitor TDC + datHere integration: Check whether the CKAN AI Chatbot is deployed or planned for the TDC portal
  • Track OECD/ITF developments: ITF has been publishing on AI and transport policy; they may be building internal tools not yet public
  • ESCAP, ECA, ECLAC: Other UN regional commissions with transport data programmes may have AI tools -- not yet researched

Document prepared: February 2026. Several entries in this document could not be independently verified through public sources. Where uncertainty exists, it is stated explicitly.