Midflight.fm

Two Stanford robotics students build some of the cheapest serious robots in the field, then find a way to teach robots without using robots at all. In 2024 they put those ideas together, design a glove before they design a body, and send it into hundreds of homes.

This is the story of Sunday and its wheeled home robot, Memo, trained on recordings of paid “Memory Developers” doing real chores in gloves shaped like the robot’s hands. We follow it from Aloha, Diffusion Policy and the Universal Manipulation Interface, through a $1.15 billion valuation and a 99.1% laundry-folding claim, to a promise of robots in about 50 family homes by Thanksgiving. Then we ask the skeptic’s question: why would a robot on wheels, taught by people in gloves, beat the humanoids from 1X, Figure and Tesla?

Narrated by Maya, an AI voice. Researched and written by Charles-Olivier Amyot. Midflight is independent and is not affiliated with or endorsed by any company discussed.

In this episode

* Tony Zhao and Cheng Chi’s research: Aloha and ACT, Mobile Aloha, Diffusion Policy, and the Universal Manipulation Interface

* The Skill Capture Glove, the paid Memory Developers, and why Sunday built a hundred gloves before one robot

* Memo’s design: wheels, a telescoping spine, a silicone shell, and passive stability

* The $165 million Series B, the “dishwasher wars” with Figure, and the ACT-2 laundry result (778 of 785, company-reported and unaudited)

* The home robot graveyard (Willow Garage’s PR2, Jibo, Anki, Kuri, Pepper, and iRobot’s bankruptcy) and the napkin math behind a $1.15 billion valuation

The number to watchHow many Founding Family homes actually have a working Memo by the end of 2026, against a target of about 50, and whether Sunday publishes how often those robots need human help. As of October 9, 2026, no delivery has been publicly confirmed.

How we label factsCompany claims are called claims, single-outlet stories are called reports, and claims made in lawsuits are called allegations, alongside the company’s response. This episode is not investment advice.

Sources

* Sunday launch release, November 2025

* Humanoids Daily, Memo and the $200 gloves

* Humanoids Daily, 100 gloves before one robot

* Business Insider via AOL, gloves vs teleoperation

* GlobeNewswire, Series B release

* TechCrunch, $1.15B valuation

* Sunday, ACT-2 preview

* Physical AI Guide, what the ACT-2 numbers prove

* UMI paper (Chi et al.)

* ALOHA and ACT paper (Zhao et al.)

* MIT Technology Review, Mobile Aloha

* Electrek, Tesla engineers join Sunday

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit midflightfm.substack.com

Two founders who started out in finance and hiring sell their recruiting startup, then set out to build an aircraft. They go public through a blank-check merger, get sued by a Boeing-backed rival, lose a co-founder to robotics, and promise passenger flights by 2025.

This is the story of Archer Aviation and its Midnight air taxi. We follow it from a stealth hiring spree through slipped certification dates, Abu Dhabi, the Los Angeles Olympics, a defense deal with Anduril, and a twist in which Archer agrees to buy the company that once sued it. Then we ask the skeptic’s question: if helicopters never made flying over traffic ordinary, why would an electric air taxi, and why is a company that has never carried a passenger worth billions?

Narrated by Maya, an AI voice. Researched and written by Charles-Olivier Amyot. Midflight is independent and is not affiliated with or endorsed by any company discussed.

In this episode

* Adam Goldstein and Brett Adcock go from Vettery to Archer, and the hiring spree that shook up air taxi engineering salaries

* The Atlas Crest SPAC, United’s order, and Wisk’s trade-secret lawsuit (allegations, with Archer’s response) through to the 2023 settlement

* How Midnight is meant to work, the dates that slipped, and the short-seller allegations Archer disputes

* New York Airways, the 1977 Pan Am Building crash, and Archer’s four-part answer to why helicopters never became everyday transport

* Hawthorne Airport, the White House pilot program, Thunder and Halo with Anduril, the Boeing deal for Wisk, Insitu and SkyGrid, and the napkin math on cash and aircraft

The number to watchThe number of passengers, beyond Archer’s own pilots, who have flown in a Midnight. Today it is zero. Archer had aimed to carry its first passengers in 2026. Watch when, and where, that number moves.

How we label factsCompany claims are called claims, single-outlet stories are called reports, and claims made in lawsuits or by short sellers are called allegations, alongside the company’s response. This episode is not investment advice.

Sources

* Archer Q2 2026 results (SEC 8-K)

* TechCrunch, SPAC deal and United order

* BNN Bloomberg, Wisk v. Archer

* TechCrunch, Wisk settlement

* Archer, Midnight unveiled

* Gulf Times (Bloomberg), UAE timeline slips

* DroneXL, White House pilot program selection

* Aviation Week, Salinas to Monterey flight

* Inside Unmanned Systems, Boeing deal terms

* Gothamist, the 1977 Pan Am Building crash

* Airways, the state of eVTOL innovators

* Pickdrones, Joby v. Archer ruling

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit midflightfm.substack.com

Some companies have already won. Midflight is about the ones still in the air: companies building real machines in the physical world, where the outcome isn’t decided yet.

Each episode asks the hard question a skeptic would ask, gives the company its best answer, and follows the founder and the people who bet on them. Then it weighs the case for, the case against and the one number to watch. All in under thirty minutes.

Narrated by Maya, an AI voice. Midflight is independent and is not affiliated with or endorsed by any company discussed.

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit midflightfm.substack.com

In 2022, in Sunnyvale, California, Brett Adcock, a farm kid from central Illinois who grew up reading Asimov, put his own money into a company his own plan said had extremely low chances of success: a general-purpose humanoid robot.

He had already sold one company and taken another public. This time he hired a crew from Boston Dynamics, Tesla and Google, landed BMW as a first customer, signed with OpenAI, then walked away to build the robot’s brain himself. By 2025, investors valued Figure at $39 billion. Then came the hardest question in robotics. Why build a whole humanoid when a robot arm already does the work?

Today Figure says it builds a robot an hour and wants one in your home. How many are doing paid work for customers, it hasn’t said. The story isn’t finished. The robot is still learning.

Midflight tells the stories of founders whose bets are still in the air.

Written by Charles Amyot. Narrated by Maya, an AI voice. Nothing here is investment advice.

Thanks for reading Midflight.fm! This post is public so feel free to share it.

Chapters

* 0:00 Cold open: a humanoid on the factory floor

* 2:56 Title

* 3:23 Chapter one: The kid from the farm

* 6:07 Chapter two: The crew

* 8:55 Chapter three: Big partners

* 14:10 Chapter four: One body, every job

* 18:00 Chapter five: The hard part

* 24:24 Chapter six: The verdict

In this episode

* From Vettery and Archer Aviation to Figure

* Figure 01, 02 and 03, the Helix AI model and the BotQ factory

* The BMW Spartanburg pilot: what Figure reports and what BMW confirms

* A former safety engineer’s lawsuit (allegations, with Figure’s response) and strength versus safety

* Home ambitions, Chinese competitors, and the napkin math behind a $39 billion valuation

Videos

* Helix launch page with demo videos (Figure)

* Figure 03 at BMW Plant Spartanburg (BMW PressClub)

* Figure 03 cleans a living room with Helix 02 (Notebookcheck)

Sources

* UF Warrington profile of Brett Adcock

* TechCrunch, Adecco acquires Vettery

* AIN, Adcock leaves Archer co-CEO role

* Axios, Figure’s launch

* TechCrunch, $2.6B valuation and OpenAI

* Decrypt, the OpenAI split

* Fortune, BMW deployment details

* Yahoo Finance (Bloomberg), $39B Series C

* Figure, BotQ

* Figure, ramping Figure 03 production

* Notebookcheck, whistleblower lawsuit

* Figure, Catalyst Brands agreement

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit midflightfm.substack.com

In 2014, in California, a former pro climber named Keller Rinaudo Cliffton started building the half of a system nobody else would. In Tanzania, he had seen health workers text for urgent medicine that never came. He called it a database of death.

By his account, experts told him drone delivery was impossible, would never be allowed, and nobody would pay for it. Rwanda’s health minister told him to just do blood. In 2016, Zipline’s planes started dropping blood into Rwandan hospitals under paper parachutes. Then came the hardest question in delivery. When a van already comes down your street, why would anyone need a drone?

Today Zipline flies dinner into backyards in Texas, has signed with Uber, and is racing to make flying a five-pound order cheaper than driving it. The story isn’t finished. Neither is the climb.

Midflight tells the stories of founders whose bets are still in the air.

Written by Charles Amyot. Narrated by Maya, an AI voice. Nothing here is investment advice.

Thanks for reading Midflight.fm! This post is public so feel free to share it.

Chapters

* 0:00 Cold open: blood from the sky

* 2:22 Title

* 2:49 Chapter one: The climber

* 7:11 Chapter two: Just do blood

* 12:13 Chapter three: The American turn

* 16:39 Chapter four: Dinner, from the sky

* 21:10 Chapter five: Turbulence

* 27:10 Chapter six: The verdict

In this episode

* Keller Rinaudo Cliffton’s path from DNA computing and pro climbing to Romotive and a trip to Tanzania

* The Rwanda deal, and a peer-reviewed study linking drone blood delivery to 51% fewer in-hospital deaths from bleeding after childbirth

* Platform 2 and its tethered “droid” for backyard drop-offs

* Ghana’s unpaid bills, parachute landings in Houston, and the race with Wing and Amazon

* The Uber deal and the reported $20 billion fundraising talks

Videos (these links go to articles that embed the videos)

* Zipline Platform 2 end-to-end demo (via New Atlas)

* MKBHD visits Zipline’s test site (via DroneDJ)

Sources

* Zipline newsroom, 2M deliveries and $600M raise

* Gavi, Rwanda launch

* Wharton, Rwanda drone study

* Baillie Gifford podcast transcript

* Harvard International Review interview

* Sequoia podcast

* Contrary Research, Zipline

* TechCrunch, Uber adds Zipline

* STAT Times (Bloomberg report, $20B talks)

* GNA, Ghana Zipline debts

* DroneXL, Houston parachute landing

* TechCrunch, Amazon Prime Air expansion

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit midflightfm.substack.com

In 2014, in Moss, a small town south of Oslo, Bernt Børnich started building a robot that didn’t exist yet: a humanoid safe enough to share a home with a family.

Nobody made the parts he needed, so 1X built them itself. It took a decade, a detour selling security robots, and a bet from OpenAI before NEO went on sale in 2025: a $20,000 robot for your living room. Then came the hardest question in robotics. When NEO folds your laundry, who is really doing the work?

Today 1X Technologies has opened a factory in California and is racing to make NEO capable on its own. The story isn’t finished. Neither is the robot.

Midflight tells the stories of founders whose bets are still in the air.

Written by Charles Amyot. Narrated by Maya, an AI voice. Nothing here is investment advice.

Thanks for reading Midflight.fm! This post is public so feel free to share it.

Share

Chapters

0:00 Cold open: the robot in your kitchen

1:58 Title

2:38 Chapter one: The kid from Moss

5:14 Chapter two: The motor

8:47 Chapter three: The bet

13:22 Chapter four: NEO

17:30 Chapter five: The hard part

20:17 Chapter six: The verdict

The number to watch: confirmed paying customers with a NEO at home in the first half of 2027, and how many of those robots stay there.

Subscribe now

Videos

I Tried the First Humanoid Home Robot. It Got Weird. (WSJ)

Full WSJ interview with Bernt Børnich

NEO product page and launch video (1X)

NEO Beta unveiling, Aug 2024 (1X)

Sources

Contrary Research, 1X

Authority Magazine interview with Bernt Børnich

Robotics and Automation News, ADT contract

Eric Jang, leaving 1X

VentureBeat, $100M Series B

The Robot Report, NEO preorders

Engadget, NEO and teleoperation

GlobeNewswire, Hayward factory

Humanoids Daily, 50,000-robot target

Euronext (Reuters relay), SoftBank talks

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit midflightfm.substack.com