Building an AI-powered Cognitive Command Centre as a Decision Support System for DMRC's Operations Control Centre

SUMMARY

Details of implementation

This Request for Proposal signifies a collaborative initiative between JICA and Delhi Metro Rail Corporation (DMRC). It aims to identify a technology solution provider (Digital Partner) for a Proof of Concept (PoC) designed to develop and deploy the Cognitive Command Centre - an AI-powered decision support system for DMRC's Operations Control Centre (OCC).

DMRC's OCC represents one of the most demanding real-time operational environments in urban metro / rail - and consequently, one of the highest-impact deployment contexts for AI. At its core, the solution embeds a voice-enabled assistance agent directly into the Traffic Controller’s workstation - providing real-time, context-aware intelligence to resolve incidents faster and more accurately, eliminating manual effort through automated incident logging and building operational intelligence capability via analytics. Designed for network wide scale & functional evolution, this initiative represents one of the foundational layers of DMRC's comprehensive AI stack.

The objective of this PoC is to design, develop and implement solution on Line 2 of the network, to showcase the success of an AI powered decision support system for DMRC’s OCC and deliver the following objectives:

Drive faster guidance response by Traffic Controllers, translating into faster resolution of incidents – leveraging real time transcription of radio conversations and making collective knowledge instantly accessible
Enhanced accuracy in incident resolution guidance, driven by unified search across comprehensive knowledge base consisting of all DMRC manuals and historical resolution data, ensuring retrieval of most relevant solutions and specialist grade guidance
Optimize critical Traffic Management decisions which significantly impact service reliability and commuter experience by factoring in multiple decision factors & real-time state of the network, while lowering cognitive load on Traffic Controllers
Elimination of manual data entries by auto-populating incident logs, reducing the time and effort put in by OCC team to manually update entries for each incident
Unlocking new insights for DMRC team by undertaking analytics over incident database to identify incident / faults trends and surface patterns across lines, assets, locations, etc. – to drive data driven decisions & targeted interventions

Timeline

The implementation is expected to span ~6 months, beginning in September 2026 and concluding by mid-February 2027.

Status

PROJECT CONTEXT

DMRC operates one of the largest and most complex urban rail networks in the world, comprising approximately 400 kilometers of track across 10 lines and over 300 stations. DMRC is committed to the continuous improvement of its operational capabilities and the adoption of technology to enhance service reliability, safety, and efficiency for ~7 million passengers it serves every day.

JICA has been a long-standing development and financing partner of DMRC for over two decades, having supported the construction and expansion of the Delhi Metro network across multiple phases through ODA loans, one of the largest and most sustained bilateral infrastructure investments in India's urban mobility sector. Building on this enduring partnership, JICA collaborated with DMRC on its data, digital and AI transformation journey, a comprehensive blueprint defining DMRC's vision and roadmap for data-led transformation.

In extension of these efforts, this RFP is being initiated to identify a Digital Partner (DP) with the capability to deliver AI powered decision support system for DMRC’s OCC. Join us on this transformative journey to elevate DMRC’s vision of becoming a world-class operator in the digital and AI era and let’s work together to embed intelligence into DMRC’s core operations - augmenting the OCC with cognitive capabilities to enable faster, smarter, and more proactive decision-making.

PROJECT STAKEHOLDERS

Strategic Partner

JICA (Japan International Cooperation Agency)
Japan International Cooperation Agency (JICA) has a longstanding partnership with DMRC since more than two decades. With JICA DXLab, the partnership has continued to strengthen, with the collaboration on DMRC’s data, digital and AI transformation journey.
With the AI powered Cognitive Command Centre, JICA and DMRC aim to accelerate DMRC’s transformation efforts by embedding intelligence across core operations. In addition to providing strategic guidance, technical expertise, and financial support, JICA plays a pivotal role in keeping these projects aligned with global best practices in digital innovation. This initiative underscores JICA’s wider mission of fostering innovation and sustainable development in partner regions.

Project Owner

Dr. Amit Kumar Jain – Director (Operations & Services), DMRC
DMRC operates one of the largest and most advanced urban rail networks in the world, spanning approximately 400 kilometers across 10 lines and 300+ stations serving Delhi NCR. With over 7 Mn passenger journeys every day, it is the transportation backbone of one of the world's most populous metropolitan areas.
With the ambition to become a world-class operator in the digital and AI era, DMRC aims to harness AI to drive innovation, enhance operational resilience, and unlock new value across its entire value chain.
DMRC is excited to work closely with JICA and the selected Digital Partner in this PoC to build an AI powered Cognitive Command Centre. This partnership reflects our collective vision to build a smarter, more resilient, and future-ready urban transit system, and jointly delivering impact for Delhi’s citizens.

PoC IMPLEMENTATION REQUIREMENTS

Eligibility (excerpts)

Candidate DPs must fulfill all the following criteria to participate in the RFP (please refer to the RFP document for full details).

Legal entity in India: Bidder must be a legal entity in India (firm / company / partnership / proprietorship firm registered under the Companies Act, 2013)
Financial stability: Minimum average turnover in India in the last 2 years to be USD 1.5 Mn / INR 15 Cr
Relevant experience: Has undertaken projects of a minimum size which involved deployment of fine-tuned LLM models, RAG, ASR / STT models deployed in a live production environment
Conflict of interest: Bidder must only participate in one bid. None of the Bidder's subsidiary, associate, holding company, or companies having common director(s), or companies in the same group of promoters / management shall participate in the bid process
Legal history: Bidder's management and proposed team members do not include members who have a history of corruption, arrest records, or involvement with criminal organizations
Debarment: Bidder not debarred / blacklisted by JICA in the past are only allowed to bid

Only eligible bidders will be considered for technical evaluation and invited for the technical presentation session. Proposals will be evaluated on technical and commercial criteria, as detailed in the RFP document. The winning candidate will be selected based on combined technical and commercial evaluation scores.

Primary scope of work

The PoC will involve the E2E development, testing and deployment of the AI powered Cognitive Command Centre for DMRC’s OCC. The selected Digital Partner will be responsible for the following scope of work (please refer to RFP document for full details):

Voice Capture & Transcription: Deploy a real time voice capture & transcription function with multilingual capabilities, to capture live radio conversations between Train Operators (TOs), Traffic Controllers (TCs), Station Controllers (SCs), etc. and convert them into structured, usable text for LLM model in real time.
Incident Resolution guidance: Automatically identify reported faults in the radio conversation and surface relevant, ranked guidance on TC's screen in a timely manner, leveraging knowledge base comprising of DMRC Manuals, historical incident data, etc.
Traffic Management Support: Showcase optimal plan for ‘Traffic Management’ decisions such as running Single Line operations, Short Looping, etc. to minimize impact on service for commuters in case of high-severity incidents that affect normal train operations across the line.
Feedback loop & Automated Incident logging: Capture TC feedback for every recommendation offered by system to enhance accuracy, while also automatically recording each incident in DMRC’s incident management system (PRA) to eliminate manual data entry.
Front end dashboard with Analytics: Role-appropriate views for TCs, Chief Controllers, OCC Managers, and DMRC Management for real-time incident tracking and network analytics.

Budget Limit

The maximum budget for the services is USD 300,000 including all taxes and expenses. Proposals that exceed this budget ceiling will not be considered.

Submission deadline

Electronic submission must be received at JICADxLab-Metro-AI@bcg.com latest by 23:59 Indian Standard Time on 12 August, 2026.

RFP

The RFP includes further information, such as the project background, location of work, and detailed Scope of Work. download

REQUIRED TECHNOLOGIES

Knowledge Base preparation:

Document ingestion & OCR processing to produce machine-readable text from DMRC manuals, including version control to manage document updates
Vector database to be set up for knowledge base – a structured, AI-queryable repository built from DMRC's Manuals, historical incident

Voice Capture & Transcription:

Use of Automatic Speech Recognition (ASR) / Speech to Text (STT) models for real-time transcription engine capable of processing live multilingual radio conversations, with domain adaptation for OCC-specific vocabulary

Incident Resolution guidance & Traffic Management:

Combination of Natural Language Processing (NLP), fine-tuned Large Language Model (LLM), Retrieval-Augmented Generation (RAG), Traffic Management AI / optimization models to showcase relevant, ranked guidance on TC's screen for identified incidents & optimal Traffic Management decisions with defined input / output guardrails

System Integration Layer:

APIs and integration connectors for all relevant data sources & systems for successful solution deployment

INFO SESSION

We plan to host an information session for prospective bidders tentatively on 27 July, 2026. If you wish to participate, please contact us at JICADxLab-Metro-AI@bcg.com. Confirmation of participation by email is mandatory to attend the information session.

INQUIRIES AND CONTACT

We are happy to address any inquiries from Digital Partners via email. Please reach out to us at JICADxLab-Metro-AI@bcg.com latest by 23:59 Indian Standard Time on 31 July, 2026. The email subject line must follow the format: “Inquiry – DMRC AI PoC - [Your Organization Name]”.

Other projects from JICA DXLab

JICA DXLab Project