START

The US produces enough biogas to make over $20 billion worth of chemicals a year

60% of it is wasted or flared in low-margin applications

It comes off wastewater plants, farms and landfills
Gas a large-scale chemical plant can't capture

So Rise Reforming goes to the gas
Its modular technology fits inside a shipping container and runs on-site

The result is the cheapest feedstock and no fossil fuels as raw material

Out come chemicals like dimethyl ether, methanol and dimethyl carbonate
That last one has no domestic manufacturer in the US

The three founders met studying molecular engineering at UChicago
They got close over late-night problem sets
Environmentalists turned engineers, drawn in by the chemical industry's 5% share of global emissions
Then Y Combinator, to learn how to sell chemicals

1,800+ hours of stable syngas production
First in the Americas to produce syngas from wastewater biogas
A binding biogas supply agreement and a conditional DME offtake for its first commercial unit

DME is first, targeting aerosol and propane starting in 2027

Rise Reforming sums it up in six words:
"Unlocking resilience for the chemical industry"

‍

🎙️ George Rose, Founder & CEO, Rise Reforming on Fondo START  

‍

00:45 Supply-secure chemicals from waste gases
01:45 Why centralized plants are exposed to conflicts and natural disasters
02:00 Making the chemicals where the waste is actually generated
02:32 Dimethyl carbonate and the lack of a single domestic manufacturer
03:04 DME, methanol, and the markets Rise Reforming is targeting
04:48 Environmentalists turned engineers
05:32 Green premium?
05:57 Price parity with the natural gas alternative
06:27 A new methanol process and the cheapest feedstock
07:05 Meeting cofounder at UChicago over late-night problem sets
08:25 Doing YC to learn how to sell chemicals

Check out www.rise-reforming.com

‍A clinical trial at Stanford enrolled a patient living in Africa.

He was never physically there.

He had been diagnosed with a rare genetic disorder, and Stanford was running a trial for it. The research team gathered his genetic information and medical records. They monitored him remotely and enrolled him.

The patient was Melissa Bime’s father.

She was working at Stanford as a clinical trial nurse at the time. She had trained as a nurse in Cameroon and moved to the US, where she worked through the COVID vaccine trials.

She says that was the spark. Why not build digital versions of patients and get them into clinical trials.

That’s Infiuss Health.

It builds medical digital twins from real-world and historical data. Trial teams rehearse a protocol across thousands of virtual patients before a single one enrolls.

The twins flag who is likely to drop out mid-study. Who has a comorbidity that affects how the drug performs. Whether two hours of travel to a research site is feasible over two years.

Infiuss reports up to a 30% reduction in sample size and design cycles 10x faster.

During the recent Ebola outbreak, Infiuss helped companies running vaccine trials recruit patients with a specific genetic trait.

Melissa’s long-term view is approval with a hundred patients instead of thousands.

Her focus is small and mid-sized life science companies. Most have one or two drug candidates the entire company hinges on. Infiuss is built to give them more tries.

"AI-Powered clinical rehearsal. Run the trial before you run the trial."

🎙️Melissa Bime, Founder, Infiuss Health on Fondo START  

‍

01:29 From nursing school in Cameroon to healthcare entrepreneurship
02:29 The diagnosis that changed everything
04:22 Building computational versions of real patients
05:27 Why clinical R&D costs billions and fails nine times out of ten
06:23 What digital patient twins can predict today
07:35 Cutting roughly six to eight months from the process
08:05 Approval with a hundred patients instead of thousands
08:59 A clinical trial without a control group
09:50 Why smaller life science companies need more chances to fail
10:21 “You shouldn’t have to go out of business ’cause you failed a clinical trial”
10:55 Finding the right genetic trait during the Ebola outbreak

Check out infiuss.com

Most sales teams are competing for the same companies. Not necessarily because they're the best customers, but because they're the easiest to find.

They have LinkedIn profiles, public hiring activity, and enough online signals to show up in every prospecting database.

But plenty of good businesses don't leave that kind of trail.

Ludo Granger kept noticing this when visiting customers. He'd look at their prospect databases and find that he could sometimes identify roughly three times as much additional market opportunity.

The companies were there. The sales tools just weren't finding them.

A hotel, a restaurant, or an early-stage startup might be a perfect customer. But if it's not generating the right online signals, it can be almost invisible to traditional prospecting software.

That became the problem Ludo wanted to solve with Leadbay.

Instead of relying only on the signals everyone else is tracking, Leadbay tries to learn how a good salesperson evaluates a company.

It looks at your business, identifies what a promising customer might look like, and uses feedback from the leads you pursue to get better over time.

But Ludo isn't trying to build another database with more companies in it.

He thinks the way salespeople use databases is changing altogether.

During SF Tech Week, he used Claude from his car to work with five highly qualified prospects rather than sorting through thousands of accounts. Then he told the agent how to handle outreach and follow-ups.

That's the distinction he's betting on. Finding more companies matters, but knowing which ones are worth pursuing matters more.

Especially when the person making that first decision isn't a person anymore. It's an agent.

🎙️ Ludovic Granger, CEO & Co-Founder, Leadbay on Fondo START with Guest Host: Kamran Karic, COO & Co-founder, Ploid (YC F26)

01:42 Why everyone is chasing the same companies

02:08 The untapped market traditional sales databases overlook

02:34 How Leadbay can expand a company's addressable market

02:51 Why startups should build something agents want

03:17 The shift from human interfaces to AI agents

03:48 How AI identifies promising businesses with limited online data

04:19 Teaching AI to recognize and improve lead quality

05:01 Finding five highly qualified prospects instead of 5,000

05:29 Trusting AI agents to coordinate outreach and follow-ups

07:04 Why qualified data is the fuel for AI-powered sales

‍

Check out leadbay.ai

When everyone has access to the same AI models, what actually makes one AI product better than another?

It's no longer just about the intelligence. It's about how reliably you put it to work.

An agent that responds faster. Makes fewer mistakes. Delivers better outcomes for customers.

That's the problem Niyaz Puzhikkunnath and Faiz Vadakkumpadath are tackling with Papaya (YC F26): continuously improving AI agents already running in production.

After spending more than a decade developing their expertise at Amazon, the founders set out to build a company of their own.

Their first product focused on data infrastructure. But as they began working with customers, they discovered something interesting.

Customers liked the data product, but were especially interested in the AI agent harness the founders had built to make their system more reliable.

Rather than asking customers to migrate their existing data infrastructure, they saw an opportunity to make that technology useful across a much broader range of applications.

So they developed the harness into a standalone product, laying the foundation for Papaya.

Today, Papaya analyzes how customers use production AI agents, identifies issues across prompts, context, caching, and tool calls, and opens pull requests with proposed fixes.

It then tracks whether those changes actually improve quality, latency, and cost.

Because knowing why your agent failed is only half the problem. Actually improving it is what matters.

That's what Papaya is building toward: AI agents that get better over time, without engineers having to manually investigate and fix every issue.

If you're running AI agents in production, check out what the team is building at papaya.fyi.

🎙️ Faiz Vadakkumpadath & Niyaz Puzhikkunnath, Co-Founders, Papaya on Fondo START with Guest Host: Maanav Agrawal, Founder & CEO, Memoir 

‍

01:43 How Papaya identifies problems in production AI agents and proposes fixes

02:49 Why fixing agents matters more than monitoring them

03:02 Why companies prioritize agent quality over cost

03:34 Papaya's early traction and first paying customers

04:58 Automating the traditional software debugging process

05:47 Why AI could eliminate the traditional maintenance phase

06:18 Why identifying agent failures isn't enough

06:39 Optimizing agents that power critical business workflows

07:46 How Niyaz's experience debugging agents at Amazon inspired Papaya

08:33 How customer feedback transformed a data platform into Papaya

‍

Check out papaya.fyi

How many companies die as an idea because the founder can't build the first version?

That's who Simon Shu and Justin Sato have in mind with Restart, which they call the AI company builder. You describe the idea in plain language and Restart sharpens it, then builds a working web product while you watch the pages, copy, and design fill in on a live preview. It publishes to a real URL when you say so, and changes work the same way: say what should be different and watch it happen.

Most people building on it are starting their first company, and most of them aren't engineers. They're making SaaS, consumer apps, marketplaces, e-commerce, service businesses, and local businesses; even a coffee shop and a restaurant right alongside apps.

The whole thing started from one question: what will companies look like five years from now?

Advanced AI will make companies, the economy, and society work in fundamentally different ways. In that world, what changes for workers?

They should get to be creatives and visionaries, and Simon and Justin are building Restart so anyone can.

Simon brings experience from Tesla and Scale AI. Justin studied CS at Stanford after doing physics and AI safety research, and he's worked in AI engineering at several startups, including Tomo. They met over poker in SF.

Restart has already grown to around 50K users, largely through social, and the team keeps shipping and improving the product. There's a free plan with daily build credits, so you can start building today.

🎙️ Simon Shu & Justin Sato, Co-Founders, Restart on Fondo START with Guest Host: @peggy_wang, Founder & CEO, @sparkchat_so @digipals_app 

01:40 Meeting at a founders-and-investors poker game in SF
02:15 The ride home where they started trading startup ideas
02:42 Building a startup always felt inevitable to both of them
03:27 Moving into Justin's Stanford dorm room
04:03 Grinding it out and raising a pre-seed round
04:20 The starting question: what will companies look like in five years?
04:55 Shipping before it's ready, and getting to ~50K users
05:18 Justin's LinkedIn post gets picked up by Indian news
05:35 The 20-iPhone content operation
06:21 "Restart is the AI company builder."
06:30 The early wedge: incorporation, payments, software
07:33 Who it's for: coffee shops, restaurants, and apps
07:58 What they love about SF
09:24 The guests flip it: what does the host think about AGI?
10:30 Simon on turning workers into creators in the AGI era
11:01 Justin on Restart's responsibility to make companies AI-native

Try it at restart.new

What does it take to build a world-class startup ecosystem?

Rodney Fong's answer starts with connection.

As President & CEO of the San Francisco Chamber of Commerce, he brings together founders, established businesses, policymakers, finance, technology, and the broader community - helping turn San Francisco’s density of talent into opportunity.

Through YES SF, the Chamber hosts 300+ events a year, creating a physical hub where ideas, industries, and people can collide.

Rodney’s philosophy is simple:

Put ambitious people in the same room. Make it easier to build. Give new ideas room to grow.

🎙️ Rodney A. Fong, President & CEO, San Francisco Chamber of Commerc on Fondo START with Guest Host: Grace Gong, Founder @ Smart Venture Media

01:01 SF Tech Week

01:48 the Chamber’s role and history

02:10 How YES SF connects founders and businesses

03:31 San Francisco’s comeback

04:50 Why SF’s talent density matters

06:10 YES SF’s 300+ annual events

06:53 Why this startup cycle feels different

07:35 A sea urchin startup idea

08:33 Robotics and everyday automation

09:13 The human side of technology

Most hiring software finds the candidates you ask for.

But what if you're asking for the wrong person?

Lantern AI, founded by Thomas Chan and Justin Tang, is building an AI hiring manager for the question that comes before the search: who should you actually hire?

It learns what strong talent looks like at your company, then searches 50+ data sources for people a keyword match on the job description would miss.

As hiring managers react to its picks, Lantern sharpens its sense of what they want. And when their requirements don't match the talent market, it pushes back instead of just following the brief.

That's the interesting part. Sometimes the best candidate misses a box or two; sometimes the person a company thinks it needs isn't the one who'd help it most. A great headhunter knows when to challenge the brief, and Thomas and Justin are building that judgment into the software.

"Turn your taste into hires." 

‍

🎙️ Thomas Chan & Justin Tang, Co-Founders, Lantern AI on Fondo START  

‍

01:27 Thomas introduces Lantern's AI hiring manager and personalized talent discovery.
01:53 Explains why static job descriptions can overlook exceptional candidates.
02:39 Justin describes Lantern's interactive approach to learning hiring preferences.
03:26 Thomas explains how Lantern challenges unrealistic hiring requirements.
04:17 Discusses Lantern's sourcing strategy using 50+ data providers.
05:19 Shares how AI is influencing demand for engineering talent.
06:43 Explains why companies need to rethink hiring as technology evolves.
08:25 Describes how Lantern evaluates candidates beyond their resumes.
09:48 Explains how AI can assess supporting evidence of candidates' work.
10:24 Discusses how Lantern helps hiring managers streamline recruiting.

‍

Check out www.lantern.md

Scope already had a live product with revenue when its customers started pulling founder Anand Pajaniradjane somewhere else.

It started in GEO and AI search optimization.
In customer conversations Anand kept showing people a testing suite the team had built around its own MCP, and they got more excited about that than the product he was actually selling.

Then one Wednesday a unicorn told him the GEO product was well thought out and well done but they weren't sure they had budget for it.
Anand brought up AI agents in the same conversation and got forwarded to the right person on the spot.

He built a small MVP that night.
The next morning he talked to about 15 companies and roughly 12 said they'd pay for it if he could make it work.
That became Scope's new direction.

AI agents now discover software, evaluate it and use it to complete tasks, but they don't use it exactly the way people do.
An agent might click through a UI, dig into a GitHub repo or go looking for examples, and its behavior can shift a lot depending on the product and context.

Scope has now logged millions of agent runs studying that behavior.
The company is building an Agent Experience Platform that shows software companies how agents find and use their products and where that experience breaks down.

🎙️ Anand Pajaniradjane, CEO & Founder, Scope on Fondo START  

‍

01:04 Scope’s mission: making software discoverable and usable by AI agents
01:58 How customer demand pulled Scope away from its original GEO product
02:32 The week Anand realized the new product had stronger customer pull
02:57 The unicorn conversation that strengthened the signal
03:27 Building the first MVP overnight and validating demand in person
03:53 Roughly 12 of 15 companies indicate they would pay for a working product
04:39 How AI agents discover and evaluate software differently from AI search
05:12 What agents actually inspect: interfaces, repositories, and examples
06:03 Why context changes agent behavior and one-size-fits-all doesn’t work yet
06:51 Anand’s experience in Founders Inc.’s Artifact program
08:18 Applying to YC as a solo founder
08:47 Anand’s first YC interview experience
09:58 Why AI agents are becoming a new class of software user
10:31 Scope’s focus on sharing what it has learned from customers and agent behavior
11:43 Where to follow Anand and Scope

Learn more at www.tryscope.com

Wait two weeks to find out a part doesn’t work.
Then wait two more weeks to try again.

Sam Mathew knows what that does to a builder.

He started in hardware. During COVID, he switched to software because he couldn’t get parts. Slow iteration wasn’t just costing him time. It was discouraging him from building hardware.

Now he’s back, making parts for other builders.

twentyfour26 manufactures metal parts in San Francisco for robotics and consumer hardware startups.
But it didn’t start as a machine shop.

Sam and his co-founder, Hemant Mahendru, were building software for manufacturers. Selling it was hard.

So on the first day of Founders Inc., they changed the business: instead of selling software to manufacturers, they became one and started taking orders themselves.

Their approach is deliberately focused.

Rather than trying to manufacture everything, they standardize the machine, materials, tools and workholding around a narrower set of parts. Their software handles quoting and programming, and they run production themselves.

Upload a STEP file. The software checks whether the design fits their setup, flags anything that needs changing, and turns accepted designs into a quote and machining plan.

When the part fits and capacity allows, they can cut it the same day.

For a hardware team, that’s more than a shorter delivery time.

It changes how often they can test a design, learn what’s wrong and try again.

The obvious cost of waiting is a delayed prototype.

The less obvious cost is a builder deciding to work on something else.

Sam’s building for the people who want to stay in hardware.

"Send CAD. Get metal."

🎙️ Sam Mathew, CTO & Co-Founder, twentyfour26 on Fondo START 

01:57 Making new toys from old ones

02:22 Dropping out in 10th grade to build an edtech startup

04:05 The outdated spreadsheet that sparked the company

04:43 Lessons from their first YC interview

05:18 Pivoting from manufacturing software to making parts

06:00 Turning CAD files into metal parts

06:28 Why faster access to parts matters for hardware builders

06:52 The cost of waiting between prototypes

07:35 The practical reason Sam shaved his head

07:48 How a seed-round joke became a viral video

‍

Check out twentyfour26.com

Shipping code got faster. Security has to keep up.

Before starting Trident, Dekai Li’s co-founder saw a workflow that concerned him:

A product manager supplied a project description. An engineer pasted it into a coding agent. The output became a pull request.

Dekai’s concern: manual pentesting couldn’t keep pace with how quickly developers were shipping.

That’s the problem they built Trident to address.

Trident uses AI to continuously pentest applications and APIs, review code, and identify security issues across cloud infrastructure.

But finding a possible vulnerability isn’t enough. You have to prove it’s real.

Dekai says validation is one of the hardest parts of AI pentesting.

That’s why Trident’s findings come with reproducible evidence, and fixes can be proposed as draft pull requests for engineers to review and merge.

The goal isn’t to give back the speed AI makes possible.

It’s to build security that can keep up with it.

Check out Trident "Continuous Security. AI agents test your apps, APIs and cloud, review every pull request, and answer your team in Slack."

‍

🎙️Dekai Li Co-Founder & CTO, Trident on Fondo START  w/ Guest Host, Grace Gong, Founder, Smart Venture Media

‍

01:05 Why manual pentesting struggles to keep pace with development
02:12 Earning trust as young cybersecurity founders
03:25 Validating vulnerabilities and navigating AI guardrails
04:02 Building credibility through experienced hires and product results
05:18 A customer case: five critical vulnerabilities in 20 minutes
06:10 The founders’ cybersecurity backgrounds and path to Trident
07:02 How AI coding workflows helped inspire the company
08:13 Lessons from YC and building around customer feedback
09:32 Scanning software packages before they enter the codebase
10:21 Focusing on the security problems customers prioritize
‍

Check out tridentsecurity.io