Pints AI
Compliance first AI agent platform for regulated banks and insurers, deploying auditable agents across underwriting, claims, risk, and compliance, with model routing and on premise or private cloud deployment.
Pints AI is a Singapore based AI research and development company that helps regulated financial institutions put AI into production for critical functions like underwriting, claims, credit decisioning, risk, and compliance. Its platform, Autothought, is a workflow layer for production ready AI agents, paired with Pints Labs, a research effort with the Singapore University of Technology and Design focused on small, efficient language models trained to understand industry specific data and logic. Founded around 2024 by Partha Rao and Calvin Tan, the company raised a five point six million dollar pre Series A round in June 2026 led by Tin Men Capital and co led by SBI Ven Capital, bringing total funding to roughly seven million dollars, and reports twelve institutions across four countries saving a combined ten million dollars using it.
What distinguishes Pints AI is a compliance first design for a domain where most AI pilots never reach production because decisions must be traceable and defensible to regulators. Autothought uses an agent orchestration framework that routes each task to the most suitable model, from its own purpose built small language models to frontier models like Claude to open source options, aiming for enterprise accuracy at lower cost. Every AI assisted decision can be traced back to its source data, uncertain outputs are flagged for human review, and the platform is built to meet the requirements of regulators including the Monetary Authority of Singapore, the Reserve Bank of India, and the Hong Kong Monetary Authority. Crucially, it deploys on a client's own infrastructure or private cloud rather than a shared service.
Delivery is hands on: Pints AI embeds engineers inside client organizations to integrate Autothought with existing banking and insurance systems, and it is building Autothought Studio so institutions can build and manage AI applications in house. As a pre Series A company, some of that self serve tooling is still maturing, and its reach is deliberately narrow, focused on regulated finance rather than general purpose use. For a bank or insurer that needs auditable, model flexible AI running on its own infrastructure and standing up to regulators, Pints AI is a strong, production minded fit; a team outside regulated financial services will find it more specialized than they need.
Vendor details
Canonical URL
https://pints.ai
Category
Enterprise operations agent
Subcategory
Regulated financial services AI agents
Funding status
Independent, headquartered in Singapore with representative offices in Hong Kong and India, founded around 2024 by Partha Rao and Calvin Tan. Raised a five point six million dollar pre Series A round in June 2026 led by Tin Men Capital and co led by SBI Ven Capital, with SEEDS, NTUitive, the SUTD Venture Fund, and Tenity participating, bringing total funding to roughly seven million dollars. Runs Pints Labs with the Singapore University of Technology and Design, is recognized by regulators and institutions including the Monetary Authority of Singapore, and reports twelve financial institutions across four countries saving a combined ten million dollars using its platform.
Company status
independent
Use cases & customers
Primary use cases
Target customers
Deployment options
Integrations
The Autothought platform deploys AI agents across research, compliance, risk, underwriting, claims, and operational workflows, integrating with a client's existing banking and insurance systems through embedded engineers. It routes each task to the most suitable model, from purpose built small language models to frontier models like Claude to open source, runs on the client's own infrastructure or private cloud, and keeps every decision traceable to source data. Autothought Studio, a toolset for institutions to build their own AI applications, is in development.
In practice
Your AI underwriting or claims pilot cannot go live because regulators need every decision traced and defensible. Pints AI's Autothought makes each AI assisted decision traceable to its source data with a full audit trail.
You cannot send sensitive financial data to a shared AI service. Pints AI deploys Autothought on your own infrastructure or private cloud, so data stays under your control.
Frontier models are accurate but expensive for every task. Pints AI routes each task to the most suitable model, from its own small language models to frontier models, for enterprise accuracy at lower cost.
Sources & related URLs
Research sources
Research notes
Added via Crunchbase agentic discovery CSV, enriched full fidelity 2026-07-07. NOTE: CSV blurb described an SRE/DevOps production-monitoring tool, which is stale/incorrect; the actual company is a regulated financial services AI agent platform (Autothought). Recategorized to Enterprise operations agent accordingly. Pre Series A June 2026; confidence medium.
Capability coverage
10.5 / 14 capabilities · 75%
| Integrations & Tool CallingIntegrates with a client's existing banking and insurance systems through embedded engineers and operates across research, compliance, risk, underwriting, and claims workflows, Pints AI docs 2026-07-07 | Full |
|---|---|
| Workflow OrchestrationAutothought is a workflow platform with an agent orchestration framework that coordinates agents across underwriting, claims, risk, and compliance processes, Pints AI docs 2026-07-07 | Full |
| Knowledge Grounding & RAGUses small language models trained to understand industry specific data and logic and grounds decisions in the institution's source data, moving from fragmented data to auditable decisions, Pints AI docs 2026-07-07 | Full |
| Human Oversight & GuardrailsUncertain outputs are flagged for human review, with a compliance first design that keeps decisions defensible to regulators, Pints AI docs 2026-07-07 | Full |
| Security, Identity & GovernanceBuilt for regulated finance with private, on premise deployment, data privacy, and compliance with regulators including the Monetary Authority of Singapore, the Reserve Bank of India, and the Hong Kong Monetary Authority, Pints AI docs 2026-07-07 | Full |
| Observability & AuditabilityEvery AI assisted decision can be traced back to its source data with a full audit trail built for regulatory scrutiny, Pints AI docs 2026-07-07 | Full |
| Memory & State PersistenceRetains institutional data and context, but a distinct agent memory that learns per user or evolves over time is not documented, Pints AI docs 2026-07-07 | Partial |
| Deployment & Data ResidencyDeploys on the client's own infrastructure or private cloud rather than a shared service, giving strong deployment control and data residency, Pints AI docs 2026-07-07 | Full |
| Prebuilt Agents, Templates & PacksProvides agents for financial workflows like underwriting, claims, and onboarding, but delivery is heavily customized per client rather than a broad self serve library, Pints AI docs 2026-07-07 | Partial |
| Triggers & Channel CoverageRuns agents within financial back office workflows, but coverage is process driven rather than a broad set of real time event triggers or channels, Pints AI docs 2026-07-07 | Partial |
| Model Flexibility & RoutingAn orchestration framework routes each task to the most suitable model, from its own purpose built small language models to frontier models like Claude to open source options, Pints AI docs 2026-07-07 | Full |
| APIs, SDKs & MCP ExtensibilityAutothought Studio is being built to let institutions build and manage their own AI applications, but this developer tooling is still in development and a public SDK or MCP is not documented, Pints AI docs 2026-07-07 | Partial |
| Testing, Debugging & OptimizationFocuses on moving from concept to production with enterprise accuracy and flags uncertain outputs, a form of validation, but a dedicated agent testing and debugging surface is not documented, Pints AI docs 2026-07-07 | Partial |
| Browser & Computer UseActs through system integrations within financial workflows rather than driving a browser or operating a computer interface, Pints AI docs 2026-07-07 | Unable to verify |
Pricing
Not public; enterprise engagement with embedded engineers, scoped to workflows and deployment
enterprise subscription scoped to workflows, deployment, and institution
What is public
No list pricing. Pints AI sells through enterprise engagement with forward deployed engineers, with no self serve tier.
Billing mechanics
Presumed enterprise subscription scoped to workflows, deployment, and institution, delivered with embedded engineering, though not disclosed.
Cost watchouts
Forward deployed engineering and on premise deployment can add implementation and infrastructure cost, even if per task model cost is lower.
Variable cost rationale
Runs on the client's own infrastructure and leans on cost efficient small language models, so once deployed the running cost is more controllable than a pure frontier model usage model, though scope grows with the number of workflows and institutions.
Additional watchouts
Delivery includes embedded engineering, so expect a services component, and confirm how pricing scales as you add workflows and whether on premise deployment changes the model.
Sales call required
Yes — required for paid access
Free / trial
Engagement and scoping on request; no public free tier
Key ambiguities
No public rate is disclosed, and the split between platform and embedded delivery services is not clear.
Related vendors
- Algebra AI — Delivery led provider that builds and runs human governed AI agents…
- Apprentice.io — AI native manufacturing platform for pharma, biotech, and cell and…
- Arkestro — Predictive procurement orchestration combining AI, game theory, and…
- Ashby — All-in-one recruiting platform combining ATS, CRM, analytics, and…
- Atomicwork — AI native ITSM and ESM platform that lets enterprises deploy and…
- Auditoria — Agentic AI for the Office of the CFO that runs accounts payable,…