Speak retweeted
Rolling out our AI Tutor Call feature today for Spanish learners. Check it out!
In our initial market Korea learning languages over the phone is very common and this feature is designed to bring that same form factor to everyone at a fraction of the cost or inconvenience.
On the surface, this might feel like just another AI chatbot. The more interesting story is the harness under the hood.
To get this to feel better than just talking to ChatGPT, we needed to solve bilingual speech, ability to drive real teaching moments in a structured manner, a specially trained turn taking model, and a supervisor agent to ensure we're giving the right level of corrections.
This is part of a much wider rearchitecture currently under way at @speak and I'm excited for what we'll be launching over next few months!
Introducing AI Tutor Call. Personal language tutoring, one tap away.
Talk about any topic. Get help in your native language when you’re stuck, with live feedback as you go.
Live today for English and Spanish learners, with more languages coming soon. Your tutor is in your practice tab. Give them a call. 📞
Speak retweeted
Speak is getting an interactive new way to practice your language skills androidauthority.com/speak-a…
Speak retweeted
It's clear to anyone in harness land that there's going to be 6-12 domain specific harnesses that become dominant at the application layer.
One for each major domain and category of human behavior.
If models are intelligence, harnesses are the digital bodies that perform the actually useful work. But in the real world, that work can be extremely different across domains and require fundamentally different "bodies" to be successful.
At @speak, for example, we're building what we think of as the world's best teaching harness: specialized skills for curriculum planning and feedback, a huge swath of tools for interactive lessons, longitudinal learner memory, and validated proficiency assessment backed by 3rd party studies.
These tools mean that we build trust that we'll actually help you achieve your goal in the long run and our system is accountable for it.
Vertical personal agents and account ownership
I don't buy that there will be one general-purpose personal agent. I think there will be specific agents for:
•productivity / work
•food and grocery
•money
•health
Each of these will be owned by whoever controls the system of record and the right to act. Take money. The money agent should be your bank account and execute transactions on your behalf. It has the ledger, the license, the fraud models, and the liability when something goes wrong. A general agent that "logs into" your bank is a screen scraper with ambition. The bank that becomes an agent is the agent.
The same pattern repeats:
•Food and grocery: Whoever owns the cart, the inventory, and the last mile. The agent knows what's in your fridge because it put it there.
•Health: Whoever holds your records, your labs, and a relationship with a licensed provider. Advice without the chart is a guess. Action without a license is a lawsuit.
•Productivity / work: Whoever owns your calendar, inbox, and docs. Context is the moat, and it already lives somewhere.
Agents win on three things: data, permission, and accountability. A horizontal agent can borrow the first, beg for the second, and has none of the third.
Trust isn't general either. I'll let an agent book my dinner without asking. I won't let the same agent wire $50K or change my medication. Each domain has its own risk tolerance, its own regulation, and its own definition of "done." One agent can't hold all of those thresholds at once.
The general-purpose agent doesn't disappear. It becomes the router. It's the front desk that knows which specialist to call. That's a real product, but it's thin. Routers commoditize, and the specialists capture the value.
The question isn't "who builds the best assistant?" It's "who owns the account the assistant acts on?" The incumbents with licenses and ledgers have a bigger head start than they realize. The startups that win will either become the system of record in a vertical or make themselves impossible to route around.
The future isn't one agent that does everything. It's a handful of agents that each own something.
Speak retweeted
If I had to bet on the shape of AI at the application layer in a few years
I'd guess there will be ~5 distinct harness categories that reshape the world
And a tutor harness will be one of them
If you want to work on those class of problems check out @speak
45+ startups hiring in this thread:
1. @turbopuffer - remote, NYC, Sf
2. @convex - SF
3. @fal - SF / remote
4. @firecrawl - SF / remote
5. @attio - SF, NYC, London + remote
6. @Linear - North America & EU remote
7. @warpdotco - NYC
8. @speak - SF / Seoul / Tokyo
9. @BrightHarborCo - Austin
10. @Vizcom - SF & US remote
11. @PostHog - NA + EMEA
12. @sievedata - SF
13. @artie_labs - SF & remote
14. @concurrencehq - NYC / SF
15. @spacexai - global
16. @EnclaveAI - NYC, SF, TLV
17. @tweetsbyport - US, TLV
18. @RevelHQ - LA / SF / NYC
19. @SnorkelAI - SF & NYC
20. @neatlogs - remote
21. @BabaHQ - NYC
22. Tremendous - remote (Americas)
23. @nucleussec - Florida / fully remote
24. Searchable - London / Salt Lake City
25. @flytbase - Pune, India
26. Indices - London / SF
27. Duna - across Europe
28. @aiunderwriting - San Francisco
29. @zenml_io - SF
30. @nexcade_ai - London
31. Social Fetch - remote
32. @gumloop - SF + Vancouver
33. @LulaConvenience - US remote
34. CLAR AI - Munich
35. @valkai - NYC / SF
36. @KeycardAI - SF
37. Latent Defense - NY
38. @DataUAcademy - Southeast Asia + global
39. @Nooqbook - remote
40. @saasflashstudio - remote
41. @SkydioHQ - US
42. @NotionHQ - global
43. @Replit - global
44. @forus - NYC
45. @harmonic_ai - NYC
Speak retweeted

Andrew Hsu@adhsuSep 29
United States
United StatesConnected via United States App StoreAccount-level information, not a live location or per-post device.
Come build a superhuman tutoring agent with me @speak
45+ startups hiring in this thread:
1. @turbopuffer - remote, NYC, Sf
2. @convex - SF
3. @fal - SF / remote
4. @firecrawl - SF / remote
5. @attio - SF, NYC, London + remote
6. @Linear - North America & EU remote
7. @warpdotco - NYC
8. @speak - SF / Seoul / Tokyo
9. @BrightHarborCo - Austin
10. @Vizcom - SF & US remote
11. @PostHog - NA + EMEA
12. @sievedata - SF
13. @artie_labs - SF & remote
14. @concurrencehq - NYC / SF
15. @spacexai - global
16. @EnclaveAI - NYC, SF, TLV
17. @tweetsbyport - US, TLV
18. @RevelHQ - LA / SF / NYC
19. @SnorkelAI - SF & NYC
20. @neatlogs - remote
21. @BabaHQ - NYC
22. Tremendous - remote (Americas)
23. @nucleussec - Florida / fully remote
24. Searchable - London / Salt Lake City
25. @flytbase - Pune, India
26. Indices - London / SF
27. Duna - across Europe
28. @aiunderwriting - San Francisco
29. @zenml_io - SF
30. @nexcade_ai - London
31. Social Fetch - remote
32. @gumloop - SF + Vancouver
33. @LulaConvenience - US remote
34. CLAR AI - Munich
35. @valkai - NYC / SF
36. @KeycardAI - SF
37. Latent Defense - NY
38. @DataUAcademy - Southeast Asia + global
39. @Nooqbook - remote
40. @saasflashstudio - remote
41. @SkydioHQ - US
42. @NotionHQ - global
43. @Replit - global
44. @forus - NYC
45. @harmonic_ai - NYC
45+ startups hiring in this thread:
1. @turbopuffer - remote, NYC, Sf
2. @convex - SF
3. @fal - SF / remote
4. @firecrawl - SF / remote
5. @attio - SF, NYC, London + remote
6. @Linear - North America & EU remote
7. @warpdotco - NYC
8. @speak - SF / Seoul / Tokyo
9. @BrightHarborCo - Austin
10. @Vizcom - SF & US remote
11. @PostHog - NA + EMEA
12. @sievedata - SF
13. @artie_labs - SF & remote
14. @concurrencehq - NYC / SF
15. @spacexai - global
16. @EnclaveAI - NYC, SF, TLV
17. @tweetsbyport - US, TLV
18. @RevelHQ - LA / SF / NYC
19. @SnorkelAI - SF & NYC
20. @neatlogs - remote
21. @BabaHQ - NYC
22. Tremendous - remote (Americas)
23. @nucleussec - Florida / fully remote
24. Searchable - London / Salt Lake City
25. @flytbase - Pune, India
26. Indices - London / SF
27. Duna - across Europe
28. @aiunderwriting - San Francisco
29. @zenml_io - SF
30. @nexcade_ai - London
31. Social Fetch - remote
32. @gumloop - SF + Vancouver
33. @LulaConvenience - US remote
34. CLAR AI - Munich
35. @valkai - NYC / SF
36. @KeycardAI - SF
37. Latent Defense - NY
38. @DataUAcademy - Southeast Asia + global
39. @Nooqbook - remote
40. @saasflashstudio - remote
41. @SkydioHQ - US
42. @NotionHQ - global
43. @Replit - global
44. @forus - NYC
45. @harmonic_ai - NYC
Speak retweeted

Andrew Hsu@adhsuSep 10
United States
United StatesConnected via United States App StoreAccount-level information, not a live location or per-post device.
Replying to @speak
@speak, we've been working closely with OpenAI on GPT-Live-1 for the past few months, building with an entirely new class of voice model model that continuously listens and speaks at the same time. Here's some of what we learned.
Speak retweeted

Andrew Hsu@adhsuSep 8
United States
United StatesConnected via United States App StoreAccount-level information, not a live location or per-post device.
Congrats to the @indent team and @sthupukari on launching! We've been using Indent for well over a year now @speak and it's organically become one of our few internal agents adopted by almost the entire team, and it helps us make more data-driven decisions daily. I've been really impressed with the significant progress they've made on a lot of the nitty gritty hard parts of enterprise agent uptake including auth/IAM and team memory and context. Onwards!
Introducing @indent, one coworker for your whole company
Indent is a single agent that is built to work with every member of your team, simultaneously. Instead of every one setting up their own agents, use Indent to build one together.
Try it at indent.com
Speak retweeted

Andrew Hsu@adhsuAug 4
United States
United StatesConnected via United States App StoreAccount-level information, not a live location or per-post device.
Amazing technical writeup worth reading in full - kudos to the gpt-live team on this work! It's clear that full duplex/bidi is the future of voice and it's really cool to peek under the hood at the complex systems engineering that makes it all "just work". Some observations/notes for voice agent builders:
- The whole post is a huge flex, but especially turning the “improve startup latency” ticket into a new webrtc handshake protocol. Generational scope creep. @juberti what was the original estimate on that issue? :)
- The voice must flow. Nothing matters but continuous realtime audio flow and smoothness. Be extremely clear in your voice agent’s system design about division of responsibilities between the realtime band and async channels/sidecars, and treat the async RPC boundary as a first-class conversational & UX design principle.
- Many other details are direct consequences of optimizing for voice fluidity: prompt caching and avoiding audio-blocking prefills, live context compaction with blue-green-style cutovers, smoothly handling long-running voice sessions (that are much longer than gpt-live’s context window)
- Future models and versions of gpt-live will continue to become more intelligent and handle more tasks natively in the realtime band. As inference speed improves, the model will also be able to spend more tokens thinking and delegating within the realtime response latency budget. This means both the actual delegation and “perceived delegation” boundary will push out.
- Interesting details on the turn-transcript heuristics, e.g. “we prioritize coherence in the displayed assistant responses even when the user speaks in the middle.” The full bidirectional/duplex model arch breaks free of modeling the convo as a traditional ordered list of chat messages, but heuristic/projected semantic turn messages are still critical for most voice agent harness systems you’d want to build around the core live model.
- Realtime full-duplex sessions load your voice agent application server’s CPU (stream handlers, queues, network paths) more heavily than almost any other ordinary application type. Think about the fact that each full-duplex session processes ~100 audio frames per second across both directions. This is a very nontrivial driver of higher infra/capacity costs when scaling realtime voice experiences.
- The bar has been set extremely high for long-running voice agent infra providers, with a lot of additional complexity to support very smooth realtime voice model inference with an explicit temporal dimension. It’ll be interesting to see the industry patterns that emerge here.
Just posted our technical deep dive on the GPT-Live system, by @zahanm and yours truly.
Speak retweeted

levithefirst@levithefirstJun 30
Africa!
AfricaConnected via Africa App StoreLocation may be affected by a proxy or VPN.Account-level information, not a live location or per-post device.
Tool of The Day - Day 19
name - Speak
→ what it does
an AI language learning app built around helping people actually speak.
instead of focusing mostly on vocabulary, streaks, or passive lessons, @speak pushes users into real conversations from day one.
it combines ┐
- structured lessons
- AI tutoring
- speech recognition
- pronunciation feedback
- roleplay scenarios
- personalized practice
into one speaking-first system.
the goal of that system is to learn a language by using it.
→ why people care
a lot of people spend months or years learning languages and still freeze when it's time to talk.
they understand it in videos.
they know the grammar.
they recognize the words.
but can't make a conversation with it.
Speak removes that gap by creating low-pressure practice where people can talk repeatedly, make mistakes privately, and get immediate feedback.
so instead of waiting for a tutor or language partner, users can practice whenever they want.
→ how people actually use it
people usually open Speak for short daily sessions.
common usage ┐
- practicing English before interviews
- preparing for travel conversations
- improving pronunciation
- simulating real-world situations
- replacing expensive tutors
- building confidence before speaking with native speakers
- maintaining language habits
many users treat it like a daily gym session for speaking.
→ extra proof
- founded by Connor Zwick (@connorzwick) and @adhsu
- backed by OpenAI Startup Fund, Accel, Khosla Ventures, and other major investors
- crossed $100M+ ARR publicly in 2025
- reached 10M+ learners across 40+ countries
- users reportedly spoke over 1B sentences in 2024
- scaled from YC W17 into a unicorn-valued company
levithefirst@levithefirstJun 25Africa!
AfricaConnected via Africa App StoreLocation may be affected by a proxy or VPN.Account-level information, not a live location or per-post device.
your browser makes money from your data, not you, here's a tool that changes that.
Tool of The Day - Day 18
name : 𝘔𝘺𝘯𝘦 𝘉𝘳𝘰𝘸𝘴𝘦𝘳
→ what it does
an AI-first browser and search project built around user ownership instead of advertiser incentives.
@mynebrowser combines ┐
- browsing
- search
- AI assistance
- local memory
- wallet functionality
- rewards
- agentic workflows
into one system.
its long-term vision is to let AI actively help with tasks like ┐
- booking travel
- shopping
- research
- managing calendars
- writing emails
- navigating the web
and it keeps your behavioral data local and under your control.
→ why people care
you pay for the internet with their data.
your searches, clicks, scrolls, and preferences generate enormous value for platforms and you rarely see any benefit.
AI increases that tradeoff because conversations reveal intent, context, and how people think.
Myne's idea is simple. if your data creates value, you should control it and participate in that value.
→ people that need it
- privacy-conscious users
- AI power users
- web3 users
- researchers
- founders
- students
- travelers
- indie hackers
- remote workers
- people tired of browser tracking
especially users who want AI assistance without handing over full behavioral profiles.
→ what problem it solves
- invisible profiling
- fragmented browser + AI workflows
- lack of user ownership
- centralized AI incentives
- hidden data monetization
- switching between assistants and browser tools
it aims to turn the browser into something that works for you instead of optimizing around advertisers.
→ how people actually use it
current usage is still mostly early community participation and testing.
projected workflows include ┐
- browsing with persistent AI context
- asking AI to complete multi-step tasks
- managing internet activity from one interface
- participating in reward systems
- shaping browser features through community feedback
the product is still early, so many use cases remain in active development.
→ why this tool stands out
most browsers compete on speed or privacy.
Myne is trying to compete on incentives.
its central idea is that browsers should become aligned systems where:
- data stays local
- participation is visible
- rewards are explicit
- AI serves users directly
it's one of the few projects approaching browsers as an economic and philosophical redesign rather than another browser skin.
→ best features
- local-first user memory
- agentic AI assistants
- proactive task execution
- integrated wallet
- user-owned data model
- explicit value exchange
- community-led access
→ best use case
perfect for ┐
- privacy-first browsing
- AI-assisted internet workflows
- early adopters
- users interested in ownership-based internet products
- people exploring alternatives to ad-funded ecosystems
→ extra proof
- manifesto published May 2026
- 30,000+ people on the waitlist
- community roles and ambassador program ongoing
Speak retweeted
when it comes to learning with ai, it's still 1995 and amazon is an online book store
impossible to comprehend how different humans will approach learning when superhuman tutors are abundant
Over the last 200 years, we've automated away a lot of hard physical labour. But people still go to the gym.
Indeed, many people today are more physically capable than people in the past. We can train systematically for whatever physical goal we want, and it's more fun than hard labour on a pre-modern farm.
@karpathy's hope is that, in the future, the same will be true of learning.
AI tutoring that's tailored to each person will make learning easy, and more people will want to do it. We will be able to go much further than our ancestors.
Speak retweeted
Will be posting the best stuff live from the Cerebral Valley Voice Summit tomorrow at @NewcomerMedia
Speak retweeted

Andrew Hsu@adhsuMar 25
United States
United StatesConnected via United States App StoreAccount-level information, not a live location or per-post device.
Strongly agree with this K-shaped divergence happening on software teams. It's existentially urgent to move faster here than what many consider reasonable, because a decent-sized gap today will compound into an insurmountable chasm in output and execution speed and quality by the fall.
This means your team must simultaneously buy into the vision of what is possible and will be possible soon, while also being in the trenches on a daily basis wrestling with and steering agentic systems that still often make mistakes, overcomplicate things, and don't always do what the user is envisioning. Navigating this takes strong, continuous leadership that's also in the trenches.
We're doing a lot at Speak to stay on the cutting edge here as a team - a few examples:
- Continually thinking about what we're still doing manually and following that signal to agentify our repos and processes
- Retooling our engineering hiring process to explicitly test for agentic engineering skill and mindset
- Explicitly trying to build production features while keeping manual code as close to 0% as possible
If you're working at a company that isn't all-in, you should leave before it's too late. Accelerate!
Speak retweeted

Andrew Hsu@adhsuFeb 12
United States
United StatesConnected via United States App StoreAccount-level information, not a live location or per-post device.
As we all know on X, agentic AI coding crossed a capability threshold at the end of 2025 such that the era of software engineers writing code by hand was over. @speak, we're urgently embracing this fundamental change to the way we develop software. I wrote up a snapshot of what we’re observing and learning, and where we think things are going:
speak.com/blog/agentic-engin…
Speak retweeted
My podcast with @speak cofounder and CEO @connorzwick is live!
Highlights:
1. How Speak built an AI language learning app
2. Why users love speaking to AI language coaches
3. Growth tactics that drove Speak to a $1B valuation
Check it out on YouTube!👇
youtube.com/watch?v=WBhQ52z0…
Speak retweeted
This Duolingo alternative changed the way I learn languages androidauthority.com/duoling…
Speak retweeted
LIVE: How AI Tutor Works with Speak’s Connor Zwick xitter.cf/i/broadcasts/1MnGnPNYb…
























