Thought waves on seed venture

Refold AI is an agentic integration platform that transforms enterprise connectivity from a services bottleneck into a seamless agent-driven superpower. It combines autonomous agents, memory-based orchestration (which retains and applies context over time), and secure edge deployment to take teams from integration request to live workflow in hours instead of months.

We’re proud to be leading Refold’s Seed round, backing experienced founders who are building the connective tissue for modern enterprise software. Refold replaces the manual work of systems integrators with embedded AI-native infrastructure. It empowers software vendors to ship enterprise-grade integrations in weeks rather than months, turning connectivity into a product advantage instead of a delivery cost.

Markets with tailwinds

Enterprise software is at an inflection point. AI is being embedded into every layer, but one major problem is holding back progress: integration. Every AI tool, legacy system, and data lake still needs to connect and communicate. Today, 80% of that work is done manually and holding back the real benefits of AI being deployed.

This is a $200 billion global challenge hiding in plain sight.

With the shift to agentic software, multi-cloud environments, and real-time orchestration, and automation, the right data has become essential. Refold is entering the market at the right time, when demand is high, legacy tools are failing, and enterprises are actively seeking a new solution.

Refold’s platform is purpose-built for complex environments, including hybrid and on-premise deployments. It solves the “last-mile” complexity that older iPaaS (Integration-platform-as-a-service) tools and consulting models cannot, making integration a feature rather than a burden.

Products that change the game

Refold doesn’t just automate steps. It makes decisions. At its core, Refold combines reasoning and reinforcement learning, enabling agents that learn and adapt over time.

The platform is built on three powerful layers:

• Workflow code agents: These write, test, and maintain custom integration logic
• MCP Chains: A natural language interface, based on the Model Context Protocol, that lets users describe workflows and launch them instantly
• Embedded integrations toolkit: Prebuilt UI components that let software vendors deliver native integrations to customers

This architecture supports teams from engineers to end users. Refold turns ad-hoc service requests into repeatable software in days. Unlike legacy systems that rely on rigid templates or billable consultants, Refold is built to eliminate complexity, not profit from it.

It enables rapid time-to-value, unlocks self-service opportunities, and codifies reusable workflows at scale. For customers like Incorta, Refold cut integration timelines by 90% and unlocked millions in revenue.

This is not a minor improvement. It is a complete rethinking of how enterprise systems connect.

Founders that hustle

Refold is led by Jugal Anchalia (CEO) and Abhishek Kumar (CTO), who previously built and exited JustDoc. They experienced the integration problem firsthand inside SAP-heavy enterprise environments, where a single schema change could result in days of downtime and six-figure escalation costs.

That experience has shaped a platform designed for speed, resilience, and developer control. The team has moved fast, developed deep product conviction, and already secured early enterprise traction. With 20 team members across San Mateo and Bangalore and growing, Refold is focused on expanding its integration catalogue and delivering zero-friction deployments.

A compelling business model

Refold has two distinct go-to-market motions:

1. Embedded deployments: Software vendors integrate Refold’s SDK, agents, and UI into their own products, delivering native connectivity to their users
2. Direct deployments: Enterprise IT and engineering teams use Refold internally to orchestrate workflows across complex internal systems

This dual model increases distribution and deepens usage. Each new integration, workflow, or module compounds the value. Some vendors are even launching new premium products powered by Refold’s platform.

Refold’s business model scales efficiently. It has low marginal costs and multiple monetisation levers, including usage-based pricing, hosting, services, and expansion tiers. It already delivers strong product currency through faster onboarding, cost savings, and increased revenue. In doing so, it is earning the right to become the integration partner of choice for some of the world’s most ambitious software platforms.

The Seed phase and beyond

This round will accelerate Refold’s growth. The company is scaling its engineering and GTM teams, expanding its library of connectors, and advancing its orchestration, memory, and reasoning capabilities. It is building the infrastructure needed to support and capture context for complex deployments across cloud, hybrid, and on-prem environments.

Over the next 18 months, the focus is on repeatability: turning strong early demand into scalable, defensible growth. Refold is not just streamlining enterprise integration. It is building the invisible logic layer that will power the next generation of software.

Global appeal

Integration is a universal problem. It slows innovation across every geography, industry, and infrastructure stack.

Refold’s platform is industry-agnostic, deployment-flexible, and designed for AI-first environments. It already has applications across analytics, fintech, business intelligence, and supply chain. Long-term, Refold’s ambition is to become the Model Context Protocol server that powers AI-native enterprise workflows.

This is not just a big market. It is foundational infrastructure for the future of software.

If you’re a visionary founder solving deep technical problems, we’d love to hear from you. Reach out via our website.

MediScan AI, a West Coast US startup is replacing the slow, manual admin behind personal injury and insurance claims with AI-first tools built for physicians: transforming a niche but essential industry that's long been underserved by modern software.

We’re thrilled to have led MediScan AI’s Seed round, partnering with the team to transform how independent medical evaluators do their work. Today, these experts rely heavily on outsourced admin services to collate records, manage paperwork, and structure reports. Many give up 25-30% of their revenue just to get the job done. These services are expensive, outdated, and inefficient. MediScan AI is building the software that replaces them. The product helps evaluators handle more cases, in less time, with no loss of quality.

Markets with tailwinds

The US personal injury market is under pressure. The number of insurance claims is rising, driven by increasing litigation and new sources of medical data. At the same time, the experts who assess these cases, known as independent medical evaluators or IMEs, are at capacity.

There are around 100,000 IMEs in the US, working within a USD 16 billion market. They are critical to the success of high-stakes legal and insurance claims, yet their workflow remains slow and manual. Documents arrive in disorganised batches, often handwritten, and it can take days to prepare a single report. With insurance losses reaching USD 143 billion last year, and evaluators struggling to keep up, the market is wide open for change.

Generative AI has reached a point where it can handle entire workflows, not just automate isolated tasks. MediScan AI is stepping in at exactly the right time.

Products that change the game

MediScan has built a platform that turns messy medical records into clear, structured reports. It handles everything from scanning and cleaning documents to pulling out key facts and creating medical timelines. Evaluators can search across an entire case using plain language. They might ask questions like “When did symptoms first appear?” or “What treatment was prescribed?” and get instant, accurate answers.

The most powerful feature is how the system learns. Every time a physician edits or refines the output, the software improves. This feedback loop is built directly into the product, meaning the quality increases with every case reviewed. Most competitors rely on manual checking and outsourced review. MediScan AI does not. Their model improves through direct use by the experts themselves.

The result is speed without trade-offs. Evaluators using MediScan AI report a dramatic reduction in admin time and a big lift in case volume. That translates to more income for them and better, faster outcomes for the lawyers and insurers who rely on their insights.

The most powerful feature is how the system learns. Every time a physician edits or refines the output, the software improves. This feedback loop is built directly into the product, meaning the quality increases with every case reviewed. Most competitors rely on manual checking and outsourced review. MediScan AI does not. Their model improves through direct use by the experts themselves.

The result is speed without trade-offs. Evaluators using MediScan AI report a dramatic reduction in admin time and a big lift in case volume. That translates to more income for them and better, faster outcomes for the lawyers and insurers who rely on their insights.

Founders that hustle

MediScan AI is led by co-founders Kavian Mojabe and Sean Podvent, based on the US West Coast. Kavian, CEO and CTO, is a software engineer with personal context. His father worked as a medical evaluator, giving him firsthand insight into the problems these professionals face. He has built systems at Amazon and in startups through to acquisition, bringing both technical skill and sharp product thinking.

Sean, COO, is a multi-time healthtech founder with a successful exit. He knows the sales cycle, understands healthcare customers, and has built and scaled SaaS businesses from scratch. His most recent company, Hygiene IQ, was acquired in 2023.

Together, they have kept the business lean and focused. They have won early customers through founder-led sales, used customer feedback to shape the product, and stayed disciplined on spend.

A compelling business model

MediScan’s first customers are individual evaluators and small groups. These professionals make independent buying decisions and are motivated to increase their efficiency. This makes them ideal early adopters.

What makes the model even stronger is how naturally it expands. MediScan AI is not just a tool for one step of the process. They are building an end-to-end system that can support the full lifecycle of medical evaluations.

• Act I: Give individual evaluators better tools to process records and write reports.
• Act II: Support physician management groups with scheduling, billing, and admin.
• Act III: Connect evaluators directly with law firms and insurers through a centralised platform.

By owning the core workflow, MediScan AI captures valuable medical insights that competitors cannot easily access. That data, structured and validated by physicians, is what makes the product stronger over time.

Global appeal

While MediScan AI is focused on the US today, the problem is global. Healthcare and insurance systems everywhere struggle with medical record reviews, especially in complex claims. The burden is high and the workflows are broken.

By proving their product in one of the most regulated and high-stakes markets in the world, MediScan AI is setting a strong foundation for international expansion. They are building with an AI-first approach and physician input at every stage, which positions them well to grow across jurisdictions.

The Seed phase and beyond

MediScan AI is early but executing with clarity. The funds from this round will go toward expanding the technical team, building out customer success, and adding features that improve collaboration and compliance.

This is a workflow that has long been overlooked. It has been buried in paperwork and powered by expensive overhead. MediScan AI is taking a software-first approach to reshaping it. They are not chasing trends. They are solving a deep, specific problem that sits at the heart of billion-dollar insurance and legal systems.

The opportunity in vertical AI is not simply about replacing manual work. It is about giving professionals more control, more leverage, and better tools to do what they do best. In MediScan AI’s case, that means helping physicians who move markets work faster, smarter, and on their own terms.

We are proud to partner with Kavian and Sean as they build the future of this category.

If you’re a visionary founder ready to make waves, please reach out via our website.

AIMon is tackling one of the biggest challenges in AI today: making Large Language Models (LLMs) safe, reliable, and enterprise-ready.

We’re thrilled to announce that we’ve co-led AIMonʼs Pre-Seed round alongside Bessemer Venture Partners. AIMon is addressing one of AI’s most pressing challenges: ensuring Large Language Models (LLMs) are safe, reliable, and enterprise-ready. Their platform empowers organisations to deploy AI confidently, mitigating risks like hallucinations, harmful content, and data leakage.

Markets with tailwinds

Generative AI is transforming industries at a breakneck pace, but deploying LLMs isn’t easy—one mistake in accuracy or safety can destroy trust and credibility. AIMon addresses this gap, providing the essential reliability layer enterprises need to adopt and build with AI confidently.

The generative AI market is expected to grow to $143B by 2027, with spending climbing at a staggering 73% annual growth rate. Regulatory frameworks like the EU AI Act are pushing organisations to prioritise transparency and safety, adding urgency to the need for reliable AI solutions. Yet, despite the massive opportunity, no major player owns the LLM reliability space—until now.

AIMon’s platform has the potential to be as indispensable to AI as Datadog has become to DevOps. It’s critical infrastructure for an AI-driven world.

Products that change the game

AIMon’s Hallucination Detection and Monitoring platform is the enterprise solution the AI world has been waiting for. It’s cutting-edge yet simple, designed to make AI deployment safe, scalable, and trustworthy. The platform delivers real-time insights, detecting hallucinations, toxicity, and sensitive data with low latency. It scales seamlessly on standard hardware, eliminating the need for costly infrastructure, and it’s customisable to meet the unique demands of enterprise applications.

By enabling organisations to deploy AI with confidence, AIMon transforms LLMs from experimental tools into dependable engines for innovation. This isn’t just a product; it’s the backbone of AI reliability.

Founders that hustle

AIMon’s co-founders, Puneet Anand and Preetam Joshi, are a powerhouse duo with a track record of delivering transformative technology. Puneet scaled multiple monitoring products at AppDynamics to $150M+ ARR, earning his reputation as a builder of must-have enterprise tools. Preetam, meanwhile, was instrumental at Netflix, where he co-created Metaflow, an open-source ML framework that’s still shaping the AI landscape for companies like Netflix, Intel, Porsche, and Goldman Sachs.

With their combined expertise in building scalable ML systems and solving hard technical problems, Puneet and Preetam are more than capable—they’re primed to lead AIMon to success. They don’t just build; they execute, iterate, and adapt with incredible speed.

A compelling business model

AIMon’s initial go-to-market strategy focuses on B2B sales, targeting enterprises that urgently need reliable AI tools. Over time, the team plans to target users bottoms-up, fostering a community-driven adoption model that scales.

Starting with monitoring, AIMon creates an indispensable wedge in the AI stack. As the platform evolves into a comprehensive evaluation and improvement layer, it becomes a key driver of enterprise AI performance. This dual approach positions AIMon to build a defensible, scalable business in a rapidly growing market.

The Seed phase and beyond

With its funding, AIMon is laser-focused on getting market traction for its product, converting warm leads into enterprise customers and growing its engineering team. Their goal is to prove product-market fit and achieve significant ARR within 18 months, paving the way for further growth.

AIMon isn’t just building a product—they’re creating the foundation for trustworthy AI at scale. As generative AI adoption accelerates, AIMon is perfectly positioned to lead the charge, making safe and reliable AI a reality for enterprises worldwide.

If you’re a visionary founder ready to make waves, please reach out via our website.