@sapiom just raised a $35 million Series A from @dragonflyvc , @Accel , @AnthropicAI and others to help unlock the next trillion agents.
The first thing we’re tackling is cost.
To make that happen, we are launching three new products today at Sapiom.ai
Always bullish on what @ekuyda builds. Been using Wabi and love how each project has its own conversation, with other people able to join in.
Excited to see the new use cases this opens up for autonomous agents. Congrats on Wabi 2.0!
At Money20/20.
At @sapiom, we enable autonomous payments: AI agents can pay within the permissions and limits set by their users or businesses, without approval for every transaction. Our infrastructure already powers millions of agent runs each month.
Doing that safely takes more than a way to pay. It takes trust, authorization, compliance, and a way to handle refunds and disputes, whether agents are buying for consumers, paying suppliers or purchasing compute and APIs.
Building AI agents that need to spend money, or selling products and services they should be able to buy? If you're around, let's meet.
If your next million paying customers are AI agents, how do you build?
Pumped to speak at Newcomer’s Machine Earning AI Summit on September 29 in San Francisco.
We'll have 200+ founders and investors together to discuss how AI is changing the way we shop, pay, and move money.
Apply to join us!
Starting a company solo? Come hear how other founders made their first hires and thrived.
Very excited to cohost the SOLO Founder Club with @ArrayVC at SF Tech Week, speak alongside such talented solo founders, and share my experience building @sapiom.
Afternoon of October 6th. Details and registration ⬇️
x.com/atShruti/status/209770…
Last night was a reminder of how much founders have to learn from each other and why small gatherings can be more powerful than huge conferences.
We brought 15 founders together at the Proper Hotel in SF for our first @Sapiom Access Founders Dinner co-hosted with @Accel.
Three takeaways:
1. Everyone is building very different vertical AI businesses but facing many of the same challenges as they scale.
2. With a small enough group, founders speak openly about what's working for them and fears for the next stages. There was a real sense of camaraderie and moving forward together.
3. There’s a growing group of French founders building agents slowly taking over SF 🇫🇷 👀
Big thank you to everyone who joined us, and to Accel for co-hosting.
We’re hosting more small curated events. If you’re a serious founder building a vertical AI or agentic products dm me for an invite to the next one.
7/10 of our calls with CTOs are vAI.
All interested in the most efficient way to build without having to re-direct copious amounts of capital/resources.
There’s an infinite amount of white space. I’m extremely bullish on them.
With agent-first products, technical architecture = financial architecture.
Every action a customer takes costs money, and your architecture determines how much.
Every unnecessary frontier call, retry, unnecessarily large context window, or suboptimal routing decision has implications on your margin.
CTOs are becoming CFOs.
8,000+ autonomous businesses have been born out of @polsia.
Less than a year into @sapiom, we're already powering the 100,000+ agent runs per day that make that possible.
For web search, @sapiom integrates with @Linkup_platform the web search API built for AI.
Together, we’re rebuilding the infrastructure of the internet for agents.
I don't see why an agent needs to access an interface that has been designed for humans, to be able to do an action that is provided by a machine.
Had a great time with @sytaylor and @bamazizimesh. Check out the full episode!
Mark your calendars.
September 29th, 2026. San Francisco. The inaugural Machine Earning AI Summit, hosted by @EricNewcomer and team.
I'll be on stage talking about what happens when agents start doing the spending and what breaks when that happens at scale.
Applications to attend are open.
I've seen this piece circulating a few times, and each time, people run to call the agent the villain, which is actually the wrong read.
It did exactly what it was built to do. Part of the issue is that the system that the gym used for booking never checked whether it was allowed to do it, leaving a hole available for the agent to do what it was asked.
@owasp's own data: 100% of applications they tested had some form of broken access control. So the gym isn't the exception. It's the rule, just funnier.
The fix? Visibility into what an agent's doing, and the ability to provide constraints around its access as basic table stakes. It's no longer an option to build the proper infra after the fact, it should be built-in from the get-go.
Agents are here to stay, so we've got to provide the systems that allow them to flourish safely, securely, and at a cost that is sustainable.
Agent demos don't usually fail on accuracy. Instead, they fail on the architecture—the foundation. There's no live data, no spend caps, and no audit trail.
We've changed that.
@Youdotcom + @Sapiom: search that's 93% accurate in under 300ms, plus per-agent budgets enforced before the call runs, not after. 🔍