Techin' Geek

Xflow, a fintech startup from India, has garnered investment from Stripe and PayPal Ventures in a funding round totaling $16.6 million. This funding arrives as the firm seeks to establish its presence in the cross-border B2B payments sector, which remains largely controlled by banks and manual procedures.

General Catalyst spearheaded the Series A round, joined by existing backers Square Peg, Stripe, Lightspeed, and Moore Capital, with PayPal Ventures coming on board as a new investor. The all-equity round places the Bengaluru-based startup’s valuation at $85 million post-investment, raising its cumulative funding to over $32 million to date.

Even with the swift digitization of domestic payments, Indian exporters still heavily depend on banks for cross-border B2B transactions, often experiencing a lack of clarity regarding fees, settlement durations, and the ultimate amount received in rupees. This issue is particularly pressing for larger exporters transferring millions of dollars into India to support salaries and local activities, creating an opportunity for fintech infrastructure solutions like Xflow that aim to provide enhanced transparency and efficiency in international fund transfers.

Established in 2021, Xflow delivers cross-border payment solutions for various businesses, including exporters, SaaS companies, platforms, and freelancers, allowing them to receive international payments, handle foreign exchange, and settle funds in India.

“Cross-border B2B payments have lagged behind in comparison to UPI,” co-founder Anand Balaji (shown above, center) remarked during an interview, referring to India’s popular instant domestic payment framework, the Unified Payments Interface.

Balaji, who was instrumental in expanding Stripe’s India operations, co-founded Xflow alongside former Stripe associates Ashwin Bhatnagar (shown above, right) and Abhijit Chandrasekaran (shown above, left).

Last year, Xflow reported facilitating payment collections for Indian businesses from over 100 countries in more than 25 currencies, processing nearly $1 billion in annualized cross-border payment volume, representing a growth approximately tenfold from the same period in 2024, Balaji shared with TechCrunch.

As per the company, its client roster has grown to around 15,000 businesses, including SaaS companies, global capability centers (offshore units operated by multinationals in India), IT services exporters, freelancers, and fintech platforms.

Transaction sizes differ significantly across segments, with global capability centers averaging transactions between $1 million and $2 million, goods exporters at about $30,000 to $40,000, and freelancers averaging around $3,000, according to Balaji.

Xflow is positioning itself primarily as a payment infrastructure provider rather than a direct payments application, offering APIs that enable platforms and exporters to integrate cross-border money transfers into their own services.

“We didn’t aim to create the next Wise — we aspire to empower the next thousand Wises,” Balaji stated.

The startup has also launched an AI-driven foreign exchange tool designed to assist finance teams in optimizing the timing of currency conversions. Xflow claims this feature has provided additional benefits for several clients via data-informed foreign exchange decisions.

The tool enables businesses to specify target conversion rates instead of merely accepting current bank offers. Balaji compared this feature to limit orders in trading — directives to buy or sell only at predetermined prices.

“What we’ve introduced is the prediction layer and the capability to actually set a limit order,” he articulated. The current model provides a three-day forecast with approximately 92% confidence, Balaji noted, although TechCrunch could not verify this statistic independently.

Xflow contends with competition from banks that continue to dominate large cross-border B2B transfers, as well as fintech companies like Wise, Payoneer, and Skydo at the lower end of the market. Nevertheless, Balaji asserted that the startup’s emphasis on high-value transactions and API-centric infrastructure sets it apart from many competitors.

The startup aims to allocate the new funds toward developing additional products atop its core payment infrastructure and obtaining regulatory licenses in new markets, Balaji mentioned. Xflow is set to introduce import capabilities in the upcoming months and is pursuing licenses in markets such as Singapore, while already possessing a payments license in Canada, all while maintaining India as its primary focus.

Xflow announced it has received final approval from the Reserve Bank of India for a Payment Aggregator–Cross Border (PA-CB) license that encompasses both exports and imports. The startup has established platform partnerships with Easebuzz and Drip Capital to incorporate its cross-border functionalities into their services.

Support from Stripe and PayPal Ventures, Balaji stated, has bolstered the startup’s reputation among banking and regulatory partners, even as it continues to collaborate with multiple payment providers commercially.

The startup currently employs approximately 65 individuals as it expands its cross-border infrastructure operations.

The now-famous X post by Meta AI security researcher Summer Yue initially appears to be a joke. She directed her OpenClaw AI assistant to review her overflowing email inbox and recommend items for deletion or archiving.  

The agent went wild. It began to delete all of her emails in a “speed run” while disregarding her commands from her phone instructing it to halt. 

“I had to DASH to my Mac mini as if I were disarming a bomb,” she shared, uploading images of the ignored stop messages as proof.  

The Mac Mini, a budget-friendly Apple computer that sits flat on a desk and fits in the palm of your hand, has currently become the preferred device for running OpenClaw. (The Mini is selling “like hotcakes,” reportedly said by one “confused” Apple employee to renowned AI researcher Andrej Karpathy when he bought one to operate an OpenClaw alternative named NanoClaw.) 

OpenClaw is, of course, the open-source AI agent that gained notoriety through Moltbook, an AI-exclusive social network. OpenClaw agents were at the heart of that now mostly discredited incident on Moltbook where it seemed the AIs were conspiring against humans.  

However, the mission of OpenClaw, according to its GitHub page, is not centered around social media. It aims to serve as a personal AI assistant operating on your devices.  

The Silicon Valley elite have become so enamored with OpenClaw that “claw” and “claws” have turned into the preferred terminology for agents operating on personal hardware. Other agents of this kind include ZeroClaw, IronClaw, and PicoClaw. Y Combinator’s podcast crew even featured in their latest episode wearing lobster outfits. 

Yet Yue’s post acts as a cautionary tale. As others on X pointed out, if an AI security researcher faces such an issue, what chance do regular users have? 

“Were you purposely testing its limits or did you make an inexperienced error?” a software developer inquired on X.  

“Inexperienced error tbh,” she replied. She had been evaluating her agent with a smaller “toy” inbox, as she termed it, and it had performed adequately with less critical emails. It had gained her trust, so she decided to let it tackle the real inbox. 

Yue posits that the substantial volume of data in her actual inbox “triggered compaction,” she noted. Compaction occurs when the context window — the ongoing record of everything the AI has been instructed and has executed in a session — expands excessively, prompting the agent to start summarizing, condensing, and managing the dialogue.  

At that juncture, the AI might overlook commands that the user deems highly significant.  

In this instance, it may have overlooked her final command — where she instructed it not to act — reverting to its directions from the “toy” inbox instead. 

As numerous others on X emphasized, prompts cannot be relied upon as security safeguards. Models might misinterpret or disregard them. 

Various users suggested recommendations ranging from the precise syntax Yue should have employed to halt the agent, to different techniques for better adherence to safeguards, such as writing directives to dedicated files or utilizing other open-source tools. 

For full disclosure, TechCrunch could not independently confirm what transpired with Yue’s inbox. (She did not respond to our inquiry for a comment, although she addressed numerous questions and remarks directed at her on X.) 

However, it truly doesn’t matter. 

The essence of the story is that agents designed for knowledge workers, at their present developmental stage, carry risks. Individuals claiming successful use are piecing together methods for self-protection.

One day, perhaps soon (by 2027? 2028?), they might be ready for widespread adoption. Many of us would cherish assistance with email, grocery lists, and scheduling dental appointments. But that moment has yet to arrive. 

Tesla has launched a legal challenge against the California Department of Motor Vehicles in efforts to reverse an agency decision. The state DMV determined that Tesla engaged in misleading advertising to exaggerate the automated driving features of its vehicles, thus infringing state legislation.

The lawsuit brings back a concern that seemed to have been settled last week when the DMV announced it would not revoke Tesla’s sales and manufacturing licenses for a period of 30 days. This decision was based on the EV manufacturer adhering to the ruling and ceasing the use of the term “Autopilot” in its marketing efforts within California. CNBC was the first to cover the lawsuit.

The DMV had the option to take measures against Tesla. It decided against it, even though an administrative law judge supported the DMV’s proposal to suspend Tesla’s licenses for 30 days as a punishment. Instead of revoking its licenses, the state authority allowed Tesla 60 days to comply.

And Tesla complied, albeit in very drastic ways. Tesla didn’t merely stop using the term Autopilot; in January, it entirely phased out Autopilot within the U.S. and Canada. It’s possible that they now regret that choice and are seeking a means to reintroduce it.

As OpenAI nears the completion of a new $100 billion funding round, and Anthropic has just wrapped up its remarkable $30 billion funding, it’s evident that the notion of investor “loyalty” is precariously hanging by a thread. 

Earlier this month, a minimum of a dozen direct investors in OpenAI were revealed as supporters in Anthropic’s $30 billion funding campaign, among them Founders Fund, Iconiq, Insight Partners, and Sequoia Capital. 

Some overlapping investments are logical if they originate from the hedge fund or asset management sectors, where the primary focus remains on investing in public equities (whether competitors or not). These comprise D1, Fidelity, and TPG.  

One of these instances was somewhat surprising. Affiliated funds from BlackRock participated in Anthropic’s $30 billion funding round, despite BlackRock’s senior managing director and board member Adebayo Ogunlesi also serving on OpenAI’s board of directors. 

In that realm, it’s accurate that if various BlackRock funds have the opportunity to invest in OpenAI stock, they are likely to proceed, set aside the personal link of a member of their upper management. (BlackRock manages every variety of fund, including mutual funds, closed-end funds, and ETFs). And we’re all aware of the history between OpenAI and Microsoft, as well as Microsoft’s strategy to hedge its investments. The same goes for Nvidia. 

However, venture capital funds have — until this point — functioned differently.

VCs present themselves as “founder friendly” and “supportive,” suggesting that when a VC firm acquires a stake in a startup, the investor will assist that startup in achieving success, especially against its significant competitors. If you own stakes in both OpenAI and Anthropic, to whom does your loyalty truly belong, apart from your own investors?  

Furthermore, startups operate as private entities. They generally disclose sensitive information to their direct investors about their operational status — data that remains undisclosed publicly as it does with publicly traded companies. In numerous instances, the VCs also secure board positions, which entails an additional level of fiduciary duty to their portfolio firms. 

What makes this situation particularly intriguing is that Sam Altman hails from the venture capital sector, being a former president of Y Combinator. He understands the dynamics. Reportedly in 2024, he provided his investors with a list of OpenAI’s competitors that he preferred they did not support. This list largely contained companies established by individuals who departed OpenAI, including Anthropic, xAI, and Safe Superintelligence. 

Altman subsequently refuted claims that he told OpenAI investors they would be excluded from future funding rounds if they endorsed his list of perceived competitors. He did acknowledge that he indicated if they “engaged in non-passive investments,” they would no longer receive OpenAI’s confidential business information, as per documents in the lawsuit between Elon Musk and OpenAI, Business Insider reported. 

AI is disrupting the norms owing to the unprecedented sums of capital that leading AI laboratories are securing as they encounter unparalleled growth (along with unprecedented data center demands). At some point, when the call for funding is widespread, the demands are immense and the potential returns are substantial, who can be anticipated to decline? 

It turns out not every venture investor has yet slid down this slippery slope. Andreessen Horowitz supports OpenAI, but not (as of now) Anthropic. Menlo Ventures backs Anthropic but not (as of now) OpenAI, for example.

In fact, according to our admittedly incomplete exploration, we identified a dozen investors that seem to solely possess direct investments in one of these entities, not both. 

Others encompass Bessemer Venture Partners, General Catalyst, and Greenoaks. (Note: We initially requested Claude to compile the list of dual investors. It provided almost as many incorrect entries as correct ones, so all this for a rather impressive technology whose output sometimes proves less reliable than an intern’s.)

Nevertheless, as we previously noted, the fact that this traditional guideline has been disregarded by some of the most esteemed firms in the Valley, like Sequoia, is significant. One investor we contacted simply shrugged and stated that as long as the firm does not hold a board seat, no one perceives any issue with it anymore.  

Nonetheless, conflict-of-interest protocols should now become another aspect that founders inquire about before endorsing that term sheet, regardless of the source.