This started as an analysis of Andrej Karpathy's excellent overview of AI's capabilities:\nhttps://x.com/karpathy/status/1979644538185752935
"Ghosts" is a brilliant metaphor for what we've created. They're not animals, and they're definitely not human. They are imperfect replicas of us. Karpathy describes them as a "statistical distillation of humanity's documents".
AI channels that distribution more effectively than any human and can beat us at Go, schoolwork, analyzing medical images, and many well defined tasks. At the same time, it lacks the reward systems that humans use to improve including curiosity, empowerment, play, intrinsic motivation, and culture.
As a result, AI's capabilities are limited. If you watch how the best work gets done with AI, it happens in chunks where a human supervises the output and gives iterative feedback to the AI. Large scale autonomous agents are brittle and fail quickly. Watch anyone vibe code an application with any level of novel complexity.
AI also faces integration barriers into existing organizations. AI is not capable of pulling a lot of the levers you need to be effective like coordinating with multiple stakeholders, building trust, authenticity, and interacting with different modalities across time and space. You could argue that the average human doesn't either, but people know when they're interacting with an AI and don't allow it the same agency as they do to people. The most successful AI B2B companies actually need more humans to integrate what they've built (forward deployed engineers) than traditional B2B SaaS.
Now, that all said, I think AI will completely reshape SaaS. Incumbents will be killed by those who know how to leverage AI. I've seen countless homepages talk about being the "AI platform for AI agents" but can't even string a demo together. Meanwhile they're trying to pitch a future where fully autonomous entities collaborate in parallel to write all the code and humans are useless. They will be the first to be replaced when they get surpassed by AI native companies.
The investors blindly throwing money into companies at 100x multiples are going to lose their money. I spoke with one of the most disciplined investors I know last week. They said they felt they had to play the game on the field even though they knew it didn't make sense. And this was from someone who is closer to the technology than 95% of investors.
On the other hand, companies who deeply understand AI will win everything. Cursor, Glean, Decagon, Sierra, Linear, Lovable, Replit, Bolt, Granola, and many AI natives are off to a great start.
While 90% of incumbents haven't adapted, it is possible. Figma, Notion, Vercel, Box, and Intercom have done a great job of tearing down what they have and rebuilding AI native products. They have teams who are close to the current capabilities of the models. They also understand their problem domain and as new capabilities come out know what capabilities will map well to what problems in what way. They are able to deliver on AI's promise to their customers. Whereas the majority of existing companies will die.
Within Amplitude's space (analytics), the door is wide open. In spite of many of our competitors filling up their homepages with the text "AI", I haven't seen a single compelling demo. We're still working like it's 2015.
We have spent the last year at Amplitude rebuilding our team to be AI native. We've learned about what models are capable of, how to write prompts, and how to leverage evals for building great products. We've worked with our customers to see what gets used in practice and what doesn't.
I know HN isn't one for hype, but we're going to be coming out with a lot over the next 6 months and I genuinely think it will change analytics."},"title":{"matchLevel":"none","matchedWords":[],"value":"The State of AI in SaaS"}},"_tags":["story","author_sskates","story_45634211","ask_hn"],"author":"sskates","created_at":"2025-10-19T14:03:46Z","created_at_i":1760882626,"num_comments":0,"objectID":"45634211","points":2,"story_id":45634211,"story_text":"SaaS is Dead, Long Live SaaS
This started as an analysis of Andrej Karpathy's excellent overview of AI's capabilities:\nhttps://x.com/karpathy/status/1979644538185752935
"Ghosts" is a brilliant metaphor for what we've created. They're not animals, and they're definitely not human. They are imperfect replicas of us. Karpathy describes them as a "statistical distillation of humanity's documents".
AI channels that distribution more effectively than any human and can beat us at Go, schoolwork, analyzing medical images, and many well defined tasks. At the same time, it lacks the reward systems that humans use to improve including curiosity, empowerment, play, intrinsic motivation, and culture.
As a result, AI's capabilities are limited. If you watch how the best work gets done with AI, it happens in chunks where a human supervises the output and gives iterative feedback to the AI. Large scale autonomous agents are brittle and fail quickly. Watch anyone vibe code an application with any level of novel complexity.
AI also faces integration barriers into existing organizations. AI is not capable of pulling a lot of the levers you need to be effective like coordinating with multiple stakeholders, building trust, authenticity, and interacting with different modalities across time and space. You could argue that the average human doesn't either, but people know when they're interacting with an AI and don't allow it the same agency as they do to people. The most successful AI B2B companies actually need more humans to integrate what they've built (forward deployed engineers) than traditional B2B SaaS.
Now, that all said, I think AI will completely reshape SaaS. Incumbents will be killed by those who know how to leverage AI. I've seen countless homepages talk about being the "AI platform for AI agents" but can't even string a demo together. Meanwhile they're trying to pitch a future where fully autonomous entities collaborate in parallel to write all the code and humans are useless. They will be the first to be replaced when they get surpassed by AI native companies.
The investors blindly throwing money into companies at 100x multiples are going to lose their money. I spoke with one of the most disciplined investors I know last week. They said they felt they had to play the game on the field even though they knew it didn't make sense. And this was from someone who is closer to the technology than 95% of investors.
On the other hand, companies who deeply understand AI will win everything. Cursor, Glean, Decagon, Sierra, Linear, Lovable, Replit, Bolt, Granola, and many AI natives are off to a great start.
While 90% of incumbents haven't adapted, it is possible. Figma, Notion, Vercel, Box, and Intercom have done a great job of tearing down what they have and rebuilding AI native products. They have teams who are close to the current capabilities of the models. They also understand their problem domain and as new capabilities come out know what capabilities will map well to what problems in what way. They are able to deliver on AI's promise to their customers. Whereas the majority of existing companies will die.
Within Amplitude's space (analytics), the door is wide open. In spite of many of our competitors filling up their homepages with the text "AI", I haven't seen a single compelling demo. We're still working like it's 2015.
We have spent the last year at Amplitude rebuilding our team to be AI native. We've learned about what models are capable of, how to write prompts, and how to leverage evals for building great products. We've worked with our customers to see what gets used in practice and what doesn't.
I know HN isn't one for hype, but we're going to be coming out with a lot over the next 6 months and I genuinely think it will change analytics.","title":"The State of AI in SaaS","updated_at":"2026-03-05T22:50:27Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"zebirdman"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["decagon"],"value":"Say a company (eg Delta Air Lines) wants to use genAI for their customer service. OpenAI/Anthropic (and maybe Gemini too) can help with this and partner directly with Delta and call them a customer. But there are also specific companies (i.e., application layer companies) that use OpenAI's and/or Anthropic's LLM to build a product that uses genAI for customer service. Examples are Decagon, Sierra, Cresta, MavenAGI, etc. \nThe same applies for other use cases beyond customer service too: legal tech, enterprise search, etc. \nWhy would Delta partner with an application layer company vs. the LLM themselves?"},"title":{"matchLevel":"none","matchedWords":[],"value":"Difference b/w partnering with OpenAI/Anthropic vs. an application layer company"}},"_tags":["story","author_zebirdman","story_41436688","ask_hn"],"author":"zebirdman","created_at":"2024-09-03T16:45:24Z","created_at_i":1725381924,"num_comments":0,"objectID":"41436688","points":1,"story_id":41436688,"story_text":"Say a company (eg Delta Air Lines) wants to use genAI for their customer service. OpenAI/Anthropic (and maybe Gemini too) can help with this and partner directly with Delta and call them a customer. But there are also specific companies (i.e., application layer companies) that use OpenAI's and/or Anthropic's LLM to build a product that uses genAI for customer service. Examples are Decagon, Sierra, Cresta, MavenAGI, etc. \nThe same applies for other use cases beyond customer service too: legal tech, enterprise search, etc. \nWhy would Delta partner with an application layer company vs. the LLM themselves?","title":"Difference b/w partnering with OpenAI/Anthropic vs. an application layer company","updated_at":"2024-09-20T17:44:19Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"polygot"},"title":{"matchLevel":"none","matchedWords":[],"value":"Show HN: Pedant: Linting for English texts"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["decagon"],"value":"https://github.com/decagon/pedant"}},"_tags":["story","author_polygot","story_12136165","show_hn"],"author":"polygot","children":[12137648,12152076,12152426,12161514],"created_at":"2016-07-21T11:41:14Z","created_at_i":1469101274,"num_comments":6,"objectID":"12136165","points":15,"story_id":12136165,"title":"Show HN: Pedant: Linting for English texts","updated_at":"2024-09-19T23:28:22Z","url":"https://github.com/decagon/pedant"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"polygot"},"title":{"matchLevel":"none","matchedWords":[],"value":"Hn-sonar: get notified for HN replies to your comments. No email address needed"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["decagon"],"value":"https://github.com/Decagon/hn-sonar/"}},"_tags":["story","author_polygot","story_15997605"],"author":"polygot","children":[15998969],"created_at":"2017-12-24T02:49:03Z","created_at_i":1514083743,"num_comments":1,"objectID":"15997605","points":3,"story_id":15997605,"title":"Hn-sonar: get notified for HN replies to your comments. No email address needed","updated_at":"2024-09-20T01:52:37Z","url":"https://github.com/Decagon/hn-sonar/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"polygot"},"title":{"matchLevel":"none","matchedWords":[],"value":"Show HN: An app to \u201cpredict\u201d hearing loss, records output volume everyday"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["decagon"],"value":"https://github.com/Decagon/can-you-hear-me"}},"_tags":["story","author_polygot","story_13286876","show_hn"],"author":"polygot","created_at":"2016-12-30T19:51:13Z","created_at_i":1483127473,"num_comments":0,"objectID":"13286876","points":3,"story_id":13286876,"title":"Show HN: An app to \u201cpredict\u201d hearing loss, records output volume everyday","updated_at":"2024-09-20T00:13:25Z","url":"https://github.com/Decagon/can-you-hear-me"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"polygot"},"title":{"matchLevel":"none","matchedWords":[],"value":"Bashreqs: like pipreqs but for bash scripts"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["decagon"],"value":"https://github.com/Decagon/bashreqs"}},"_tags":["story","author_polygot","story_18004917"],"author":"polygot","created_at":"2018-09-17T11:42:09Z","created_at_i":1537184529,"num_comments":0,"objectID":"18004917","points":2,"story_id":18004917,"title":"Bashreqs: like pipreqs but for bash scripts","updated_at":"2024-09-20T03:00:01Z","url":"https://github.com/Decagon/bashreqs"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"polygot"},"title":{"matchLevel":"none","matchedWords":[],"value":"Wick: browse and interact with the internet like your filesystem"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["decagon"],"value":"https://github.com/Decagon/wick"}},"_tags":["story","author_polygot","story_13361523"],"author":"polygot","created_at":"2017-01-10T00:47:39Z","created_at_i":1484009259,"num_comments":0,"objectID":"13361523","points":2,"story_id":13361523,"title":"Wick: browse and interact with the internet like your filesystem","updated_at":"2024-09-20T00:15:00Z","url":"https://github.com/Decagon/wick"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"polygot"},"title":{"matchLevel":"none","matchedWords":[],"value":"Show HN: Star.js: auto-correct typos in messages based on the next message*"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["decagon"],"value":"https://github.com/Decagon/star.js"}},"_tags":["story","author_polygot","story_12087337","show_hn"],"author":"polygot","created_at":"2016-07-13T16:12:34Z","created_at_i":1468426354,"num_comments":0,"objectID":"12087337","points":2,"story_id":12087337,"title":"Show HN: Star.js: auto-correct typos in messages based on the next message*","updated_at":"2024-09-19T23:29:53Z","url":"https://github.com/Decagon/star.js"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"Hexahgon"},"title":{"matchLevel":"none","matchedWords":[],"value":"RabbitIO: Unified authentication for PlayerIO"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["decagon"],"value":"https://github.com/decagon/rabbit"}},"_tags":["story","author_Hexahgon","story_10506947"],"author":"Hexahgon","created_at":"2015-11-04T14:54:38Z","created_at_i":1446648878,"num_comments":0,"objectID":"10506947","points":1,"story_id":10506947,"title":"RabbitIO: Unified authentication for PlayerIO","updated_at":"2024-09-19T22:26:51Z","url":"https://github.com/decagon/rabbit"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"bmaupin"},"title":{"matchLevel":"none","matchedWords":[],"value":"My Disastrous Experience with Ubuntu 14.04 LTS Trusty Thar (2015)"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["decagon"],"value":"http://www.deragon.info/ubuntu14.04.html"}},"_tags":["story","author_bmaupin","story_11488368"],"author":"bmaupin","children":[11488573,11488635,11488637,11488721,11488722,11488726,11488731,11488777,11488789,11488793,11488836,11488872,11488879,11488905,11488923,11488957,11488979,11488998,11489021,11489070,11489098,11489177,11489264,11489319,11489534,11490232,11491233,11494945],"created_at":"2016-04-13T14:20:39Z","created_at_i":1460557239,"num_comments":122,"objectID":"11488368","points":77,"story_id":11488368,"title":"My Disastrous Experience with Ubuntu 14.04 LTS Trusty Thar (2015)","updated_at":"2024-09-19T23:07:26Z","url":"http://www.deragon.info/ubuntu14.04.html"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"allleycat"},"title":{"matchLevel":"none","matchedWords":[],"value":"Calling All Outlier Founders"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["decagon"],"value":"https://medium.com/decasonic/calling-all-outlier-founders-e7542a717089"}},"_tags":["story","author_allleycat","story_28840777"],"author":"allleycat","children":[28841539],"created_at":"2021-10-12T15:17:54Z","created_at_i":1634051874,"num_comments":1,"objectID":"28840777","points":7,"story_id":28840777,"title":"Calling All Outlier Founders","updated_at":"2024-09-20T09:36:23Z","url":"https://medium.com/decasonic/calling-all-outlier-founders-e7542a717089"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"kawamuray"},"title":{"matchLevel":"none","matchedWords":[],"value":"Use and extend WebAssembly/WASI for language-agnostic Kafka consumer framework"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["decagon"],"value":"https://engineering.linecorp.com/en/blog/adding-experimental-webassembly-support-to-decaton---part-1/"}},"_tags":["story","author_kawamuray","story_24717868"],"author":"kawamuray","created_at":"2020-10-08T10:48:42Z","created_at_i":1602154122,"num_comments":0,"objectID":"24717868","points":2,"story_id":24717868,"title":"Use and extend WebAssembly/WASI for language-agnostic Kafka consumer framework","updated_at":"2024-09-20T07:05:00Z","url":"https://engineering.linecorp.com/en/blog/adding-experimental-webassembly-support-to-decaton---part-1/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"bbrookshier"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["decagon"],"value":"Dragonfly Energy files lawsuit against YouTube reviewer Will Prowse"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://www.sec.gov/Archives/edgar/data/1847986/000149315226026754/form8-k.htm"}},"_tags":["story","author_bbrookshier","story_48380423"],"author":"bbrookshier","children":[48380424],"created_at":"2026-06-03T05:53:53Z","created_at_i":1780466033,"num_comments":1,"objectID":"48380423","points":5,"story_id":48380423,"title":"Dragonfly Energy files lawsuit against YouTube reviewer Will Prowse","updated_at":"2026-08-15T04:18:53Z","url":"https://www.sec.gov/Archives/edgar/data/1847986/000149315226026754/form8-k.htm"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"misterballer"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["decagon"],"value":"I'm an engineering undergrad considering pursuing a PhD with the long-term goal of founding a hard-tech startup. I\u2019m undecided about the specific area and only have a vague sense of what interests me -- industrial decarbonization, mining / mineral processing, and advanced materials are a few areas that currently seem exciting.
However, I'm unsure which type of PhD lab would provide better preparation: either a problem-first lab (e.g., like Yet-Ming Chiang at MIT) or a technology-first lab (e.g., your typical academic lab that focuses on novel science without a predetermined application, and where commercialization happens largely by chance \u2014 when a technology happens to have a valuable market application).
I've looked at advice online from successful hard-tech entrepreneurs, but the answers are conflicting. Some say to work backwards from a problem, but others argue that problem-first approaches often don't work since deep tech is inherently different: you often can't force a scientific breakthrough for a predetermined problem. Instead, they argue that most hard-tech companies are only founded because somebody made a scientific breakthrough and realized afterwards that there might be a commercial application. Indeed, the VC firm Pillar VC says that most deep-tech companies they know were "technology-first."
With that in mind, does anybody have any advice on which type of PhD lab to join?"},"title":{"matchLevel":"none","matchedWords":[],"value":"Ask HN: Should hard-tech founders join a problem or technology-first PhD lab?"}},"_tags":["story","author_misterballer","story_49299333","ask_hn"],"author":"misterballer","children":[49299385,49299389,49300534,49301299,49301600,49304426,49309130,49315066],"created_at":"2026-08-14T14:38:22Z","created_at_i":1786718302,"num_comments":14,"objectID":"49299333","points":13,"story_id":49299333,"story_text":"I'm an engineering undergrad considering pursuing a PhD with the long-term goal of founding a hard-tech startup. I\u2019m undecided about the specific area and only have a vague sense of what interests me -- industrial decarbonization, mining / mineral processing, and advanced materials are a few areas that currently seem exciting.
However, I'm unsure which type of PhD lab would provide better preparation: either a problem-first lab (e.g., like Yet-Ming Chiang at MIT) or a technology-first lab (e.g., your typical academic lab that focuses on novel science without a predetermined application, and where commercialization happens largely by chance \u2014 when a technology happens to have a valuable market application).
I've looked at advice online from successful hard-tech entrepreneurs, but the answers are conflicting. Some say to work backwards from a problem, but others argue that problem-first approaches often don't work since deep tech is inherently different: you often can't force a scientific breakthrough for a predetermined problem. Instead, they argue that most hard-tech companies are only founded because somebody made a scientific breakthrough and realized afterwards that there might be a commercial application. Indeed, the VC firm Pillar VC says that most deep-tech companies they know were "technology-first."
With that in mind, does anybody have any advice on which type of PhD lab to join?","title":"Ask HN: Should hard-tech founders join a problem or technology-first PhD lab?","updated_at":"2026-08-16T05:49:26Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"gukoff"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["decagon"],"value":"There Are Magic Hexagons of Every Order"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["decagon"],"value":"https://gukov.dev/math/2026/08/02/new-magic-hexagons.html"}},"_tags":["story","author_gukoff","story_49229174"],"author":"gukoff","children":[49229920,49230076,49230415,49230520,49230552,49230769,49231215,49231865,49233535,49233977,49236015,49237718,49256872],"created_at":"2026-08-09T07:19:58Z","created_at_i":1786259998,"num_comments":33,"objectID":"49229174","points":197,"story_id":49229174,"title":"There Are Magic Hexagons of Every Order","updated_at":"2026-08-14T23:26:53Z","url":"https://gukov.dev/math/2026/08/02/new-magic-hexagons.html"}],"hitsPerPage":50,"nbHits":20,"nbPages":1,"page":0,"params":"query=Decagon&tags=story&hitsPerPage=50&advancedSyntax=true&analyticsTags=backend","processingTimeMS":9,"processingTimingsMS":{"_request":{"queue":250,"roundTrip":20},"afterFetch":{"format":{"highlighting":1,"total":1},"merge":{"mergeLoop":{"prepareNextHit":2,"total":2},"total":3},"total":3},"fetch":{"query":4,"total":5},"total":9},"query":"Decagon","serverTimeMS":260}