Meta completes first open-source AI incubator in Singapore

Source: Meta. The inaugural Llama Incubator Program Demo Day in Singapore. 

Meta completed its inaugural Llama Incubator Program in mid-October with a Demo Day naming teams CREX, MyRepublic Broadband, and Straits Interactive as winners. Three runners-up were also selected: AgriG8, IntentAI, and i-Sprint Innovations. Winning teams receive US$30,000 each from Meta, and the runners-up, US$10,000 each.

The Llama Incubator Program is designed to build capabilities and drive innovation on open-source AI among startups, enterprises, and public sector teams in Singapore. Participants are provided with expert mentoring, technical resources, and funding to help them develop practical AI solutions built on Meta's open-source model, Llama.

More than 100 organisations participated in the foundational workshop held in March 2025. Forty organisations were selected for six months of dedicated business and technical mentoring, as well as training on responsible AI practices including Llama’s protection tools and the IMDA Starter kit for Safety Testing of LLM-based applications, Project Moonshot and the AI Verify Testing Framework. 

Meta also partnered with AWS Singapore to offer eligible companies AWS credits, empowering them to accelerate their solutions to market. As part of this collaboration, AWS hosted an enablement session for startups on utilising Llama models with Amazon SageMaker AI and Amazon Bedrock, and will continue to provide tailored technical advisory and support.

Collectively, participants developed over 30 innovative Llama-powered solutions spanning the finance, healthcare, education, and public sectors.   

“At Meta, we see ourselves as an ecosystem partner, working to advance AI capabilities and make their benefits widely accessible. The Llama Incubator Program exemplifies this drive, and we appreciate the opportunity to collaborate with the Singapore government on this initiative. We are proud to support the Smart Nation 2.0 vision—empowering local small businesses and attracting startups to build and innovate in Singapore,” said Sandhya Devanathan, VP - India & Southeast Asia at Meta.

“We are excited by the participation of global AI leaders like Meta in Singapore. The Llama Incubator Program is an example of how collaboration can rapidly translate cutting-edge technology into real-world solutions. By giving startups, SMEs and public agencies hands-on technical guidance, this opportunity will spur innovation, and provide hands-on experience, strengthening Singapore’s role as an AI hub in the region,” said Philbert Gomez, Senior VP & Executive Director, Digital Industry Singapore. 

“The digital economy is driving Singapore's growth, and at the heart of it is the digital transformation of ur local enterprises. IMDA is committed to helping our enterprises build digital capabilities, particularly in AI, to address business needs. Through the GenAI x Digital Leaders initiative, IMDA partners with global tech leaders like Meta to give our enterprises access to expertise and tools to harness AI for productivity and growth. 

"Programmes like Meta’s Llama Incubator provide hands-on experience to help companies navigate AI’s complexities, empowering them to innovate and sharpen their competitiveness,” said Johnson Poh, Assistant Chief Executive, Sectoral Transformation Group, IMDA. 

The solutions from the winning teams included: 

- CREX 

CREX empowers hotels to deliver a premium guest experience by transforming sustainability actions into AI-ready trust signals that boost bookings and brand value. With the largest emission factor database in Southeast Asia, CREX makes carbon accounting fast, precise, and actionable, the company said.

Although 76% of travellers want to book sustainable hotels, fewer than 0.5% of hotels on major hotel booking websites are certified as sustainable. CREX explains that sustainability reporting is complex, manual, and resource-heavy, leaving most hotels unprepared. The company uses Llama-powered AI to automate document extraction, emissions reporting, and guest-facing narratives, making sustainability simple, trustworthy, and revenue-driving. 

CREX uses Llama 4 multimodal models for document understanding, emissions reasoning, and narrative generation across multiple languages. Its agentic workflows combine retrieval-augmented generation (RAG), safety guardrails, and emissions factor databases (10,000+ AI-ready ground truths) to ensure auditable, compliant outputs. 

CREX's solution: 

▪ Cuts sustainability reporting time from 18 months to 18 hours 

▪ Provides 20+ trust signals to boost guest confidence and conversions 

▪ Helps hotels cut up to 30% in operational costs through sustainability adoption

- MyRepublic Broadband

MyRepublic delivers large language model (LLM)-agnostic, AI-powered automation that integrates across data sources, business systems, and collaboration platforms. 

Sales leaders at small and medium sized enterprises (SMEs) juggle multiple tasks with lean teams, such as managing customers, chasing leads, updating customer relationship management (CRM) systems, and handling enquiries across channels. Traditional CRM tools can help, but are often expensive, manual, and complex.

The AI Sales Coworker uses AI agents powered by Meta’s Llama for lead research, prioritisation, assignment, product recommendations, and outreach generation. Integrated with RAG for product-specific knowledge, it delivers trustworthy, personalised sales automation while ensuring oversight, safety, and governance.

The solution:

▪ Reduces manual effort by automating the sales outreach process to support sales closure in a single management platform

▪ Provides 24x7 virtual AI coworkers via WhatsApp, Teams, Telegram, and Google Chat for seamless sales support

▪ Potentially disrupting a traditional SME CRM market worth US$46.4B growing at 8.6% CAGR

- Straits Interactive 

Straits Interactive, through its Capabara platform, offers data protection, governance, and responsible AI adoption. The company said that 80% of non-tech professionals are disadvantaged in AI transformation - facing low adoption, high costs, and fears of job displacement. 

Straits Interactive bridges this gap by combining competency and governance: delivering hands-on AI training, a software-as-a-service (SaaS) platform for custom AI tutors and workflows, and consulting services that ensure responsible deployment.

Capabara integrates with Meta's Llama 4 to power explainable and responsible AI. Its patent-pending Reasoning-Aware AI refines outputs for accuracy, fairness, and explainability, while built-in risk analysis, model cards, and prompt shields ensure compliance and safety. 

The platform supports local, secure deployments and is tailored for non-technical professionals. To date:  

▪ Five thousand apps built, 1,100 National AI prompt competition participants, 2,300+ AI Blended course participants, and 450 AI-certified professionals trained 

▪ Sixteen times more cost-effective AI deployment compared to alternatives 

▪ Democratises AI creation for ASEAN’s 300 M+ professionals, unlocking productivity and reducing workforce displacement risks

The solutions from the runners-up included:

- AgriG8

AgriG8 is an agri-fintech platform that bridges the gap between financial institutions and smallholder rice farmers. Through its CropPal app and partner dashboard, AgriG8 digitises farmer data, provides real-time crop insights, and enables alternative credit assessments. This empowers financial institutions to lend confidently to an underserved yet high-potential market, while helping farmers break free from costly informal financing. According to AgriG8, AgriG8:

▪ Unlocks access to a US$12 B financing market across 12 M hectares of farmland in Southeast Asia

▪ Reduces farmers’ borrowing costs by replacing 40% informal interest rates with affordable formal credit

▪ Enables financial institutions to expand portfolios into an underserved but high-growth sector with scalable, data-backed lending

- IntentAI

IntentAI helps enterprises deploy AI agents in production safely and easily. The team has delivered scalable solutions across the Asia-Pacific region for leading banks, telcos, and healthcare providers.

According to IntentAI, 95% of AI projects fail to deliver results due to poor tooling, steep learning curves, and expensive frameworks. IntentAI solves this with its AI Agent Orchestrator, an intuitive platform that makes enterprise deployment open, easy, and production-ready.

The Intent Orchestrator integrates with Meta’s Llama-powered Code Generator to guide flow creation, embed guardrails, and enable local or on-premise deployments. By combining agent orchestration, flow optimisation, and open-source flexibility, IntentAI delivers scalable, safe, and enterprise-ready AI. 

The solution: 

▪ Reduced lead assignment time from 5 days to 5 minutes with automated agent workflows 

▪ Successfully scaled to handle 370,000+ appointments over WhatsApp for healthcare clients 

▪ Achieved 77% query resolution for medical support bots while ensuring safety and accuracy 

- I-Sprint Innovations

Identity and access management (IAM) provider i-Sprint Innovations secures digital identities for over 150 banks across Asia Pacific, including two of Singapore’s three local banks. Their flagship product, AccessMatrix, delivers multifactor authentication and single sign-on for trusted and secure access to digital services. 

The company notes that cybercriminals are shifting from “breaking in” to “logging in” using stolen or misused credentials. Traditional defences like firewalls and security information and event management (SIEM) tools cannot detect valid but compromised accounts, leaving organisations blind to identity-based attacks. 

Llama-powered identity threat detection & response (ITDR) delivers visibility, context, and automated responses. The solution integrates Meta Llama 3.1-8B with Llama Guard 3 into the AccessMatrix IAM platform. The system analyses audit trails, assigns severity levels, and generates explainable incident reports with recommended actions such as invalidating sessions or blocking IP addresses. 

The solution:

▪ Detects anomalous identity activities that bypass legacy rule-based systems 

▪ Cuts incident investigation time, reducing security operations centre (SOC) workload and response delays 

▪ Helps enterprises prevent account takeovers, fraud, and compliance penalties

Mrs Josephine Teo, Singapore Minister for Digital Development and Information, delivered opening remarks at Meta Singapore’s Llama Incubator Demo Day. She said that incubators like the Llama Incubator Program are key enablers of purposeful AI use across planning, prediction, personalisation, automation, and anomaly detection, and also that partnerships and community will remain vital for Singapore to remain competitive and drive continued digital and economic growth.

Minister Teo noted that AI is increasingly mainstream in Singapore, and is likely to benefit the whole of society. When innovation only benefits a segment of society, “the economy continues to chug along, but it does not realise its full potential; it does not uplift all companies the same way as the rising tide lifts all boats,” she explained.

“Experimentation takes courage, but very importantly, it takes resources. The Llama Incubator Program makes available not just access to an advanced model, but also the engineering resources, knowledge and insight that Meta, together with your partners, have brought to the table,” she said.

Minister Teo also touched on distinguished organisations talking about how AI experimentation is failing, and that AI is not producing desired results. “These are fair comments, but they also reflect just how difficult experiments could be,” she said.

“Yet at the same time, we have to learn. We have to acquire capability. Even for a general-purpose technology, it takes time for its benefit to be realised.”

“There are also important lessons to be taken away when studies show that AI adoption seen as a fool's errand. We need to find the sweet spot in the way we experiment, to build long term capabilities, and a sense of trust that the experimentations are serious and should not be given up on easily,” she added, introducing the “three P’s and two A’s” method for looking at use cases.

“The three Ps where AI can be put to very good use are personalisation, planning and prediction. The two A's are automation and anomaly detection, that is the ability for AI to trawl through tons of data to find items that seem suspicious,” she said.

“If we could build a case to accept that the experimentation may not always produce the results that we want, but we will get better at understanding the problem and build capabilities that enable the organisations to achieve success in AI in the future, we will still have gained something.” 

Even though studies suggest that AI adoption will not yield results, it is not about giving up, but doubling down, she emphasised. She added that it is important to engage people who are in the community as there is a lot that can be learned from them, highlighting the Lorong AI community as one example.

In 2025, Meta partnered with the Singapore government to launch the Llama Incubator Program – a first for Meta in the Asia Pacific region. Collaborators include the Ministry of Digital Development & Information (MDDI), InfoComm Media Development Authority (IMDA), Government Technology Agency of Singapore (GovTech), Digital Industry Singapore (DISG), Enterprise Singapore (EnterpriseSG), AI Singapore, and SGInnovate, along with programme and technical partners.

*All results are self-reported and not identically repeatable. Generally, expected individual results will necessarily differ.