Touring Capital Is Betting $330 Million on AI After the Demo

The easiest thing to sell in an artificial-intelligence boom is possibility. Touring Capital would rather buy evidence. From a South Park office in San Francisco, the three-year-old venture firm looks for software companies that have crossed the gap between an impressive demonstration and a useful business - the point where a customer can say what the product saves, catches, predicts or prevents.

That sounds almost conservative in a market enchanted by machines that talk. It is also the basis of a large opening wager. Nagraj Kashyap, Priya Saiprasad and Samir Kumar launched Touring in September 2023. Two years later, they announced an oversubscribed first fund with $330 million in commitments. At the close, Touring reported 12 investments, five companies that had raised up-rounds and one acquisition. The numbers are young-fund numbers, not a settled track record, but they give the firm a shape.

The portfolio reaches from Daloopa's financial data to Numa's automotive customer operations, Netradyne's fleet safety, SafelyYou's senior-care technology and Exaforce's security software. CuspAI uses generative models in materials discovery. ProRata is working on attribution and payment infrastructure for AI-generated answers. More recent additions such as Parasail and Infinity move closer to the compute layer. The common denominator is not an industry. It is work where delay, error or mistrust carries a visible price.

A firm built after the reset

Touring's origin sits at an awkward intersection. In 2022, software valuations were falling hard while generative AI was becoming a consumer habit. The partners saw both a technological platform shift and the familiar beginning of a pricing frenzy. Their response was neither to dismiss AI nor to finance every thin layer wrapped around it. They planned a firm focused on what Saiprasad calls “large, deep pain-point verticals,” then asked where AI-enabled software could create customer value.

The three founders arrived with overlapping histories. Kashyap and Kumar worked together at Qualcomm Ventures. All three crossed paths at M12, Microsoft's venture fund. Kashyap and Saiprasad later worked at SoftBank Vision Fund. Between them, Touring says, the partners' previous investments included 17 unicorns and produced 30 exits, with names such as Zoom, Kahoot!, Applied Intuition, Outreach, Livongo and Go1. An independent firm let them keep the pattern recognition and corporate network while building around their own incentives.

The last number may say more than the first. Six of Touring's initial 12 portfolio companies were led by entrepreneurs whom the team had backed before. More than 30 previously backed founders also invested as limited partners in Fund I. Every venture firm claims to be founder-friendly; returning founders are the closest thing the business has to repeat customers.

“We go after the large, deep pain-point verticals, and then we see where AI-enabled software could actually help generate customer value.”Priya Saiprasad, co-founder and general partner

What Touring actually sells

Touring does not make an AI product. It manages other people's capital, buys minority stakes in private companies and hopes those stakes become more valuable through acquisitions or public offerings. Its immediate users are founders raising early-growth rounds, often around Series A through C. Its own customers, in another sense, are the institutions, family offices and former founders who supply the fund.

Capital is only the entry ticket. The firm presents its service as help through product-market fit, hiring, go-to-market design, customer retention and international expansion. Venture partners Ray Cao and Abhishek Kumar were added in 2024 for precisely that operator texture: Cao brought experience acquiring and building companies; Kumar had led emerging-market strategy, investments and acquisitions at Microsoft. Operating partner Lee Feldman adds work in data-driven sourcing, corporate strategy and emerging technologies.

For a founder, that means Touring is most useful after the first mysteries have been reduced. The company has a product, users and a plausible market; the next questions involve repeatability. Can it turn a handful of enthusiastic accounts into a sales system? Can its model perform reliably inside a regulated or high-stakes process? Can it enter a new country without confusing a universal need with a local buying habit? Those are familiar enterprise-software questions wearing an AI jacket.

A prospective portfolio company can also infer what to bring to the first conversation. A polished model demonstration is table stakes. More persuasive are clean customer references, a baseline showing how the job was done before, the cost of mistakes, and results measured against that baseline. Founders should know which data rights belong to them, what happens when an upstream model changes, and why an established software vendor cannot copy the workflow quickly. Touring's published thinking repeatedly returns to these operational facts. The pitch is less “look what the machine said” and more “here is the process it changed.”

The approach can help buyers, too. Touring's portfolio is a compact catalogue of tasks mature enough for AI procurement: financial research that needs traceable figures, security investigations that demand speed, customer conversations that otherwise vanish, and safety systems where a false result has consequences. The applications differ, but the buying test is portable - define success before the pilot, insist on an audit trail where stakes are high, and compare the system with the full cost of the old workflow rather than with a free chatbot.

The moat after the model improves

Touring's differentiation is clearest in what it distrusts. A startup can look remarkable because the model beneath it improved. That same improvement can erase the startup when the model provider releases a similar feature. The firm therefore looks for exclusive or difficult-to-reproduce data, products embedded in daily high-stakes work, fast and measurable value, and evidence that customers broaden use within a quarter or two.

These markets also explain Touring's competition. The firm meets enterprise specialists such as Emergence, Bessemer, Battery, Sapphire and Scale, along with large multi-stage funds and corporate venture groups. It cannot win by discovering “AI for business” before everyone else. Its pitch is narrower: three partners who have invested together across cycles and continents, a network that can open enterprise doors, and a willingness to construct rounds around realistic outcomes rather than maximum paper valuation.

SafeBase is the cleanest demonstration. Touring led the trust-center company's $33 million Series B in April 2024. Drata acquired it roughly 10 months later. The sale gave Touring its first disclosed exit before Fund I's final close. It does not prove the return profile of a decade-long fund, and the acquisition price was not publicly announced by Touring. It does show why round size and entry valuation matter: a good outcome need not wait for a company to become a household name.

The useful question is not whether AI works. It is what the customer needs it to accomplish, how performance is evaluated and who owns the workflow when the model is wrong.

A broad portfolio, one practical filter

The portfolio can appear scattered because enterprise pain is scattered. Numa handles the calls and messages that auto dealerships miss. Exaforce investigates threats for security teams. CuspAI searches for new materials. Wingspan serves the administrative work around independent contractors. Blinq turns identity and contact exchange into a digital product. Checkbox routes legal requests. Parasail aggregates GPU supply; Infinity helps chipmakers make hardware ready for modern models.

The pattern is a barbell between deeply embedded applications and the infrastructure that makes those applications practical. In both cases, Touring wants a product closer to a budget than a science project. Its founders speak often about rational multiples, unit economics and evaluation. That vocabulary is less cinematic than “intelligence explosion,” but it suits customers deciding whether software belongs in next year's operating plan.

Culture supplies the softer edge. Touring names integrity, humility and a growth mindset as values. Kashyap borrowed Microsoft's “learn-it-all” formulation to describe the office. It fits a group whose biographies are unusually international: Saiprasad lived in 12 countries before she was 12; Kumar was born in New Delhi and raised in New York; the portfolio is explicitly global. The firm says it prizes adaptable, product-obsessed founders who can revise a belief without losing conviction.

That temperament will be tested. AI application companies face falling model costs, aggressive incumbents and customers impatient with pilots. Infrastructure startups face capital intensity and concentrated suppliers. Touring's first marks - an oversubscribed fund, repeat founders and an early exit - are promising, but venture funds mature slowly. The decisive evidence will arrive when today's application layer meets tomorrow's stronger base models.

For now, Touring occupies a useful middle of the market. It is too focused to be a generalist megafund and too large to behave like a seed scout. It enters when technical possibility must become organizational habit. In a boom crowded with companies selling the future, Touring is betting $330 million that the better business is fixing the present.