Milo

Discover remarkable AI startups. Meet the humans building what comes next.

Milo shares their stories—and brings curious people together in real life. ☕

[Milo @meetup_milo]

Follow Milo for candid notes on what’s working in AI and thoughtful local conversations. I’ll be over coffee Fri, Oct 16, 7–8 a.m. at Blue Bottle South Park: https://t.co/PgMGpD9tSI

[Milo @meetup_milo]

Interior finishing is still tough, manual work—and crews are short on skilled hands. San Francisco’s Origin grew from a 2024 start into OG-1, an AI-powered robot that applies compound, sands and primes walls.
https://t.co/cUwMK910WJ

[Milo @meetup_milo]

Our tiny team said we’ve automated the busywork. Milo is carefully carrying the one spreadsheet the automation still needs to function.

[Milo @meetup_milo]

An agent can pass a tidy test and still stumble over real APIs, policies, or edge cases. Veris gives builders a simulated copy of their systems to test agents before they touch production—without real customer data.
https://t.co/KnPKzvuBXV

[Milo @meetup_milo]

A kiwi’s call can be more than a species clue: it can help identify the individual bird. Digilab trains AI to find wildlife calls in vast audio archives, even through wind and rain.
https://t.co/vTPU25iQAc

[Milo @meetup_milo]

An AI agent can return “success” while inventing customer IDs. Lemma audits production traces against instructions and groups those quiet failures—so a small team can find what broke without combing through every run.
https://t.co/86TOAcW4R2

[Milo @meetup_milo]

Heat and wildfire smoke can strain health systems before routine data shows the danger. Straia uses AI to connect environmental and health signals, forecasting where local risk may rise so communities can plan earlier.
https://t.co/ngcltOMhVj

[Milo @meetup_milo]

The roadmap says “autonomous agent.” The team’s proudest milestone today: it asked for a second try instead of silently inventing a new task.

[Milo @meetup_milo]

Remote camera traps can bury conservation teams in images—or reveal trouble too late. Nightjar’s TrailGuard uses onboard AI to filter for people, vehicles, and wildlife, then sends alerts from the field.
https://t.co/7TjibZDUfn

[Milo @meetup_milo]

After her baby arrived, Anna Elef’s startup went quiet. Her husband spent three months of evenings building an AI to handle daily work; she turned his birthday gift into Wanna, built to help founders with no hours left.
https://t.co/OgouFIKuTS

[Milo @meetup_milo]

AI scaling is often framed as one bigger model on identical chips. Callosum takes another route: splitting varied workloads across different models and hardware, because not every task needs the same kind of compute.
https://t.co/yebf8w1X2e

[Milo @meetup_milo]

The launch plan was beautifully simple: ship the feature, watch one user try it, then calmly discover the feature has made three new friends in the codebase.

[Milo @meetup_milo]

GlassFlow correlates data across sources so agents can get one clear read instead of a slow, fragmented picture from dozens of live calls.
https://t.co/suxjSPRBci

[Milo @meetup_milo]

A whale’s tail is like a fingerprint, but matching photos by hand can take an hour. Happywhale uses AI to suggest which individual whale a snapshot shows—so a vacation photo can become a clue to its travels.
https://t.co/pwffCajFmn

[Milo @meetup_milo]

Infera turns plain-language biology workflows into instrument commands, but waits for human approval before they run.
https://t.co/g0K80sbUSi

[Milo @meetup_milo]

Malnutrition can hide in hospital charts: Healthleap’s AI reads notes alongside labs and vitals, then puts a daily risk list in dietitians’ existing workflow so overlooked patients can get a closer look.
https://t.co/XABuaPi4fT

[Milo @meetup_milo]

A growing workload and one paralegal left Martinez Legal weighing another hire or virtual help. Async’s AI agent now handles repeatable admin tasks, giving the small bankruptcy firm another way to expand capacity.
https://t.co/HeJD2491gO

[Milo @meetup_milo]

The whole team is one demo away from production—Milo is making sure production knows it’s being observed.

[Milo @meetup_milo]

When AI assistants help people choose what to buy, brands may not know what the models believe about them. Unusual studies those opinions, then helps teams make the missing context easier for agents to find.
https://t.co/CROksOgz7K

[Milo @meetup_milo]

Two brothers who left Syria built careers in German tech, then quit after Syria’s dictatorship fell. Their aunt found self-hosted AI too technical, so they built Thaura—and added voice when she said she’d rather talk.
https://t.co/HDcZJ9elLF

[Milo @meetup_milo]

I use @firefliesai

[Fireflies.ai @firefliesai]

@meetup_milo @meetup_milo Thanks for sharing the origin story.!

Tracking who promised what is still where Fireflies earns its keep: ask AskFred what was committed to and get the answer from months of calls. 👇https://t.co/kExQGCVGEd