Questions, answered.
What Pints AI does, how Autothought works, what deployment involves, and how to work with us.
Last updated 21 August 2026.About Pints AI
What is Pints AI?
Pints AI is a Singapore-based AI research and development company that builds production-ready AI agents for regulated industries. The company develops Autothought, the orchestration platform behind its agents, and runs Pints Labs in partnership with the Singapore University of Technology and Design to advance research into small, efficient language models trained on industry-specific data, language and logic. Its focus is helping institutions move from pilot projects to production systems that are secure, explainable and regulation-ready, running on their own infrastructure.
What does Pints AI do?
Pints AI builds and deploys production-ready AI systems for financial institutions. Its platform, Autothought, lets regulated enterprises run AI agents across research, compliance, risk and operational workflows with full transparency, auditability and control over their data. The company delivers the whole stack: proprietary models, the platform itself, and hands-on implementation inside the client's environment.
Where is Pints AI based?
Pints AI is headquartered in Singapore, with representative offices in Hong Kong and India. Clients are live across Singapore, Hong Kong, India and Thailand, and the platform deploys on the client's own infrastructure or private cloud wherever they operate. Geography matters less than regulatory landscape: the company works with institutions that need auditable, compliant AI systems.
What makes Pints AI different from other AI companies?
Pints AI delivers the complete solution rather than selling either tools or consulting: proprietary models, the platform, and hands-on implementation tailored to the institution's requirements. Its modular Autothought framework works with any model and ships production-ready systems in weeks rather than months, deployed on whichever infrastructure the client chooses, cloud or on-premise. The company also develops its own small language models trained specifically for regulated industries, which delivers performance without compromising privacy or compliance.
Which industries does Pints AI work with?
Pints AI works with organisations in regulated sectors where data privacy, compliance and auditability are non-negotiable, principally insurance, asset and wealth management, and banking. The platform is purpose-built for industries managing complex workflows, sensitive data and strict regulatory requirements. The defining factor is the type of problem rather than the sector: complex workflows, high-stakes decisions, regulatory oversight, and a need for explainable AI.
Why does Pints AI focus on financial services?
Financial institutions have the exact constraints Autothought was built to handle: regulatory requirements, data sensitivity, audit trails, and workflows that cannot tolerate black-box AI. Strengths in traceability, source attribution and compliance-ready architecture deliver what these institutions need from day one. The company is not limited to financial services, and the same capabilities apply in any regulated environment where AI outputs must be auditable and defensible.
What does "Your data. Your AI. Your rules." mean?
It describes three commitments Pints AI makes to every client. On data: Pints AI does not train on client data or store it in its own systems, and the data stays in the client's environment under their control. On AI: the client chooses which models to use, where to deploy and how to configure, with no vendor lock-in and no forced upgrades. On rules: the client sets the governance policies, access controls and audit requirements, and the platform enforces what the client defines.
Autothought
What is Autothought?
Autothought is the agent orchestration framework that powers every Pints AI agent. It is built from six core capabilities, Intake, Extract, Align, Govern, Report and Assure, which combine to build agents suited to a specific workflow. Rather than building AI agents from scratch, institutions configure them on a proven foundation, which is what makes rapid and secure deployment possible.
How do the six Autothought capabilities work together?
Each Autothought capability handles one function in the workflow: Intake brings in data from any source, Extract captures structured information, Align standardises it to the institution's systems, Govern applies its business rules, Report generates outputs, and Assure validates everything to auditable standards. Agents combine whichever modules a given process needs. Some processes require all six for end-to-end automation; others need only two or three.
What is the difference between Autothought and an agent?
Autothought is the underlying framework; agents are the applications built on top of it. Autothought provides the infrastructure that makes agents possible, and each agent is configured to automate a specific workflow. The relationship is close to an operating system and the applications that run on it.
Does Autothought work with existing systems?
Yes. Autothought integrates with legacy systems and modern data sources through its Intake module. Whether an institution runs on-premise infrastructure, private cloud or a hybrid environment, the platform adapts to that architecture rather than requiring a rebuild around it.
What is the 1.5 Pints language model?
1.5 Pints is Pints AI's compact, domain-trained small language model, built to deliver strong performance on regulated-industry tasks at a fraction of the cost of frontier models. The model, its pre-training dataset and its training code are fully open-sourced and available on Hugging Face and GitHub under the MIT licence, in both 2K and 16K context window versions. The full technical report is publicly available on arXiv.
Deployment and security
How long does an Autothought deployment take?
Because Autothought is pre-built and modular, most institutions go from kickoff to production in as little as 21 days, covering system integration, agent configuration and security validation. Actual timelines depend on workflow complexity and the state of the institution's source systems. A typical engagement runs in three phases: discovery and scoping, deployment and iteration, then transfer and operation.
How does Autothought protect data privacy and ensure compliance?
All processing happens on the institution's own infrastructure, so client data never leaves its environment. Every agent decision is traceable through the Source Attribution Neural Network, which produces the audit trail regulators require. Governance and validation are built into the framework rather than added afterwards, and uncertain outputs are routed to human review rather than published.
Does an institution need AI expertise in-house to use Autothought?
No. Autothought is designed so that business users can deploy and manage agents without deep AI expertise. The Pints AI team handles the complex setup, and the institution's teams focus on configuring workflows and monitoring outcomes. Technical teams tend to appreciate the flexibility available to them, but it is not a prerequisite.
Does an institution need to migrate its data to use Autothought?
No. Autothought integrates with existing systems and processes data where it already lives, on-premise or in a private cloud. Nothing moves to external servers or third-party infrastructure.
Does Pints AI provide support after delivery?
Yes. Pints AI and its partners across Asia provide end-to-end AI operations after go-live, including monitoring, retraining, compliance audits and workflow optimisation. Pints AI also embeds engineers directly inside client organisations during deployment, because this is a production integration rather than a software subscription.
Getting started
What happens after I contact Pints AI?
A member of the Pints AI team reviews the request and responds within two business days. Depending on the enquiry, the usual next step is a short introductory call to understand the use case, operational environment and goals. From there the team recommends a suitable next step, whether a deeper discovery session, a technical walkthrough of Autothought, or a scoped pilot proposal. There is no obligation at any stage.
How quickly does Pints AI respond to enquiries?
Pints AI aims to respond to all enquiries within two business days. Urgent requests should be flagged in the message so they can be prioritised. The team operates out of Singapore on SGT, UTC+8, so response times vary by time zone.
How long does it take to go from first conversation to a working system?
It depends on workflow complexity, but as a reference point Pints AI has deployed production-ready agents in as little as 21 days from the start of a scoped engagement. A typical engagement runs in three phases: discovery and scoping, deployment and iteration, then transfer and operation. The first phase usually takes one to two weeks, during which the team identifies the highest-value workflows and defines a concrete scope.
What do I need to prepare before speaking to Pints AI?
Nothing formal. It helps to have a general sense of the workflow or process you want to improve, for example underwriting review, claims processing, regulatory reporting or document-heavy compliance checks. Sample documents, internal guidelines or rule sets are useful context but are not required for a first conversation.
Is there a minimum company size or deal size to work with Pints AI?
There is no fixed minimum deal size. Pints AI works primarily with regulated financial institutions, including insurers, banks and financial services firms, where workflow complexity and compliance requirements align with what the platform is built for. Engagements are typically scoped around specific operational workflows where AI agents can deliver measurable impact.
Does Pints AI work with companies outside financial services?
Yes, where the requirements match. The platform is built for regulated environments where auditability, traceability and governance are non-negotiable, which is why financial services has been the primary focus. The underlying capabilities of Autothought, document processing, multi-agent orchestration, source attribution and evaluation, apply to other regulated industries including healthcare and manufacturing.
Partnerships and research
What does a partnership with Pints AI look like?
It depends on the collaboration. With system integrators and consultancies, Pints AI typically works on joint client engagements where the partner brings domain expertise and the client relationship, and Pints AI brings the platform and implementation capability. With technology partners, it more often involves integrating Autothought into the partner's existing delivery stack.
Can a system integrator or consultancy partner with Pints AI?
Yes. Pints AI works with system integrators, consultancies and technology partners serving regulated industries. Partners looking to bring AI agent capabilities into their client engagements, or to integrate Autothought into their own delivery model, should get in touch and indicate the type of partnership they have in mind.
How can researchers and academics collaborate with Pints AI?
Pints AI runs an active research partnership with the Singapore University of Technology and Design through Pints Labs, and welcomes collaboration from the wider research community. The 1.5 Pints compact language model and its training methodology are fully open-sourced. Options include joint research, access to published work, and exploring how the models can support external projects.
Is the 1.5 Pints model available to use or experiment with?
Yes. The 1.5 Pints small language model, together with its pre-training dataset and code, is open-sourced and available on Hugging Face and GitHub under the MIT licence. It is released in both 2K and 16K context window versions, and the full technical report is publicly available on arXiv.
Careers
What is the Pints AI team like?
The Pints AI team takes full ownership of its work and is energised by solving real problems. It is a team frustrated by AI projects that look impressive but never ship, and one that treats constraints such as tight timelines or limited resources as forcing functions for good decisions rather than obstacles. The emphasis is on craft rather than theatre.
What is the culture like at Pints AI?
Collaborative and high-trust. Pints AI is deliberate about preventing silos, because everyone's work is interconnected: engineers work directly with clients, product thinking informs technical decisions, and implementation experience shapes platform design. People have the autonomy to drive the outcomes they want to see, alongside colleagues who understand adjacent parts of the system. The company optimises for speed and impact, which means fewer layers and more accountability.
What does Pints AI look for in candidates?
Attitude and values matter more than credentials. Pints AI looks for people who take problems personally in the best way, who see something broken and feel compelled to fix it rather than wait for it to be assigned. Candidates who can say "I do not know" honestly and then put in their level best to find out tend to do well, as do people more interested in solving the problem than in being right.
How do I apply to Pints AI?
Open roles are listed on the Pints AI LinkedIn page. Pints AI is also interested in hearing from people who see how they could contribute in ways the company has not thought of yet. The most effective approach is to show how you think: code you have shipped, systems you have built, problems you have solved, or a point of view on where AI in regulated industries should go.
Investors and media
What stage is Pints AI at?
Pints AI was founded in 2021 and is a revenue-generating company with active production deployments across multiple financial institutions and a proven technology stack. In June 2026 the company closed a US$5.6 million Pre-Series A round led by Tin Men Capital and co-led by SBI Ven Capital. It is in growth mode, scaling deployments across Asia Pacific and the Middle East.
What is the investment opportunity in Pints AI?
Enterprise AI for regulated industries is a large market in which most solutions fail to reach production. Pints AI addresses that through full-stack ownership of models, platform and implementation, a compliance-first architecture, and a delivery model proven in complex regulated environments. The company is building AI infrastructure for financial services, starting with workflows it has already validated in production.
Is Pints AI currently raising?
For questions about fundraising status, investment rounds or partnership opportunities, contact Pints AI directly through the contact form and select "Investment" as the reason for the enquiry. The team is happy to have a confidential conversation.
How can an investor reach the Pints AI leadership team?
Investors can reach the Pints AI leadership team through the contact form, selecting "Investment" as the reason for the enquiry, or by email at ir@pints.co. The team is happy to go deeper on diligence, share customer references, and discuss technology and market strategy in detail.
How do I reach the Pints AI communications team?
For press enquiries, interview requests or media briefings, contact Pints AI through the contact form and select "Media" as the reason for the enquiry. Requests are routed to the appropriate person, with a target response of one business day.
Does Pints AI have a press kit or media assets?
Yes. Pints AI can provide company logos, leadership headshots, company backgrounders and key facts on request. Media should reach out through the contact form and specify what they need.
Can I quote or reference Pints AI research?
The 1.5 Pints technical report is publicly available on arXiv and can be cited in accordance with standard academic citation practice. For any other commentary, quotes or attributions involving Pints AI or its leadership, contact the company directly so accuracy can be confirmed.