callouts for archived posts & edits

2025-12-19 10:54:06 -06:00 · 2025-12-08 21:11:44 -05:00 · 2025-12-08 21:11:44 -05:00 · d495aa69ec
commit d495aa69ec
parent eb40027c53
20 changed files with 94 additions and 258 deletions
--- a/content/archive/ARCHIVED;
+++ b/content/archive/ARCHIVED;
@ -1,6 +1,6 @@
 ---
 title: "ARCHIVED: A Comprehensive Analysis of Design Patterns for REST API SDKs"
-date: 05.09.2024
+date: 05.09.24
 tags:
  - blog
  - dev
@ -11,7 +11,9 @@ description: A deep dive into SDK design patterns, comparing object-oriented vs
 > [!custom]  WELCOME TO THE PLASTIC [[archive|ARCHIVE]]
 > This blog post has been archived because it's legacy content that's out-of-date or deprecated. We keep this content around so those interested can dig into the evolution of our projects & thinking.
 > 
-> This post contains Vineeth's (Plastic's Co-founder & CTO) notes on the early design of Honcho's SDKs. For the most up-to-date SDK reference, check out the [Honcho Docs](https://docs.honcho.dev).
+> This post contains Vineeth's (Plastic's Co-founder & CTO) notes on REST API SDK design patterns that informed how we built Honcho's client libraries. Some patterns described here have been superseded by our shift toward LLM-native interfaces, but the analysis of pagination, error handling, & developer experience remains useful for anyone building API tooling.
 > 
 > For the most up-to-date SDK reference, check out the [Honcho Docs](https://docs.honcho.dev).
 > 
 > Enjoy.
@ -103,7 +105,6 @@ Platform Specific Questions
 4. What approach does the tool take? Object-oriented or singleton?
 5. How does it handle async vs sync interfaces?
 # Research
 > First I took a look at sources and posts onlines that talk in general about
 > developing SDKs. This isn't an exhaustive look at every link I looked at, but
 > ones I thought were relevant. The notes are messy and not necessarily fully
@ -364,7 +365,6 @@ and auth.
 There's also capability for adding custom code such as utility functions.
 ## Speakeasy
 Speakeasy required me to do everything locally through their `brew` package. It
 did not immediately accept the OpenAPI Spec and required me to make some tweaks.
 These were low-hanging fruit, and their cli has a handy AI tool that will
--- a/content/archive/ARCHIVED;
+++ b/content/archive/ARCHIVED;
@ -1,109 +0,0 @@
 ---
 title: "ARCHIVED: A Simple Honcho Primer"
 date: 04.16.24
 tags:
  - blog
  - honcho
  - archive
 author: Courtland Leer
 description: A beginner-friendly guide to Honcho, the AI personalization platform that helps LLM applications get to know users via storage, insights, & retrieval.
 ---
 > [!custom]  WELCOME TO THE PLASTIC [[archive|ARCHIVE]]
 > This post has been archived because it's legacy content that 
 >
 ![[bot reading primer.png]]
 > [!NOTE] Welcome to our quick, "explain it like I'm 5" guide to [Honcho](https://honcho.dev)!
 > We'll keep it simple, covering [[ARCHIVED; A Simple Honcho Primer#^ef795f|what Honcho is]], [[ARCHIVED; A Simple Honcho Primer#^x125da|why we built it]], [[ARCHIVED; A Simple Honcho Primer#^cd2d3c|how to use it]], and [[ARCHIVED; A Simple Honcho Primer#^ca46d7|where the product is going]]. But throughout, we'll link to places you can dive deeper.
 # What Is Honcho?
 ^ef795f
 Honcho is a personalization platform for large language model (LLM) applications built by [Plastic Labs](https://plasticlabs.ai).  
 It's software infrastructure that lets AI apps "get to know" their users, resulting in delightful experiences and optimized time to value.
 We'll have direct consumer experiences in the future, but today, the product is for application developers. It allows them to [[ARCHIVED; Introducing Honcho's Dialectic API#^a14c2f|reduce overhead]] and [[ARCHIVED; Introducing Honcho's Dialectic API#^x7f7f8|enhance their machine learning pipeline]].
 Right now, Honcho is in private beta, that means integrating our hosted version requires permission and onboarding[^1]. [You can sign-up here](https://plasticlabs.typeform.com/honchobeta).
 In its current form, Honcho has three core components:
 1. [[ARCHIVED; Announcing Honcho's Private Beta#^x15f37|Storage]] - managing each user's data & inference about each user
 2. [[ARCHIVED; Announcing Honcho's Private Beta#^x53717|Insights]] - processing user data with our proprietary AI models
 3. [[ARCHIVED; Announcing Honcho's Private Beta#^ee4516|Retrieval]] - surfacing user data to personalize user experience (UX)
 If you've heard of [Retrieval Augmented Generation](https://en.wikipedia.org/wiki/Prompt_engineering#Retrieval-augmented_generation) (RAG), this might sound familiar. But Honcho is doing *much* more than simple RAG.
 Behind the scenes, Honcho learns about users as people--[[ARCHIVED; User State is State of the Art|richly modeling identity]]. It seeks to understand their beliefs, hopes, dreams, history, interests, and preferences. 
 It then acts as [[ARCHIVED; Introducing Honcho's Dialectic API|an oracle to each user]], allowing apps to ask for any personal context they need to improve UX and giving them access to a social cognition layer.
 # Why We Built Honcho
 ^x125da
 Plastic Labs was founded as an edtech company. The original mission was to build an AI tutor that [[ARCHIVED; Open Sourcing Tutor-GPT#^x527dc|could reason like]] the best human instructors. We quickly found the key limitation was data not on the subject matter, but on the student. To overcome it, the tutor needed [[ARCHIVED; Theory of Mind Is All You Need|a way to]] get to know *each* of its students deeply.
 Honcho was born by running up against this challenge, building technology to solve it, and realizing all AI applications are going to need the same solutions. The promise of *generative* AI isn't one-size-fits-all products, but bespoke experiences in each moment for each user. The same limitation emerges--how well do you know your user?
 So we believe Honcho will be a critical, table-stakes part of the AI app development stack.
 Why? Because [[Humans like personalization|users will want]] their AI experiences to be personalized and app developers shouldn't be redundantly solving that problem.
 But it's not intuitive for a few reasons:
 - AI app builders are [[Machine learning is fixated on task performance|still focused on]] just getting general tasks to work
 - LLMs' [[LLMs excel at theory of mind because they read|potential to personalize]] is still under-appreciated  
 - Historic examples of personalized apps usually just leverage our activity & engagement data
 - Those examples tend to target only base user desire, lead to addictive behavior, & have poor privacy records
 Still, when interacting with an AI app, there's a sense that it *should* be getting to know us. In fact, we're often surprised when we realize it's not learning about us over time. And probably annoyed at having to start over.
 Think about personalization here as more like the experience of close human companionship or white-glove services than the attention-hacking mechanisms of TikTok. There's [[ARCHIVED; Announcing Honcho's Private Beta#^xb6ef1|enormous potential]] for more positive-sum use of user data and for aligning AI applications more closely with user needs and preferences[^2]. 
 # How to Use Honcho
 ^cd2d3c
 Honcho is first and foremost a **storage** framework. Think of it like an open source version of the OpenAI Assistants API. User `sessions` store both user and AI generated `messages` as well as any intermediate inferences you might want to store as `metamessages`: 
 ```python
 user_input = "Here's a message!"
 ai_response = "I'm a helpful AI assistant!"
 session.create_message(is_user=True, content=user_input)
 session.create_message(is_user=False, content=ai_response)
 ```
 But what about vectorDBs? Don't worry, Honcho has you covered there too. You can embed data and store them as `documents` in per-user vector DBs called `collections`: 
 ```python
 collection.create_document(content="The user is interested in AI")
 ```
 Using Honcho as a storage mechanism allows you to **retrieve** rich insights via the user profiles it's building and managing on the backend. Your application's LLM can access [[Loose theory of mind imputations are superior to verbatim response predictions|theory-of-mind]] inference over those profiles via the *[[ARCHIVED; Introducing Honcho's Dialectic API|dialectic]]* API. 
 It's simple: just query in natural language using the `session.chat()` method:
 ```python
 session.chat("What are the user's interests?")
 ```
 There are a [[ARCHIVED; Introducing Honcho's Dialectic API#How It Works|ton of ways]] to use Honcho, this primer only scratches the surface[^3]. 
 # What's Next for Honcho?
 ^ca46d7
 Beyond improving our internal AI models so they can get to know users as richly as possible, we see three natural extensions in [[ARCHIVED; Announcing Honcho's Private Beta#^eb15f3|Honcho's future]]:
 1. [[ARCHIVED; Announcing Honcho's Private Beta#^x2dd3b|Monitoring & Evaluation]] - developer tools to understand & assess the impact of personalization + machine learning tools to build personalized datasets
 2. [[ARCHIVED; Announcing Honcho's Private Beta#^a84f44|User-Facing Controls]] - chat with *your* Honcho to direct how it manages & shares data + authenticate with Honcho to sign-in to AI apps
 3. [[ARCHIVED; Announcing Honcho's Private Beta#^ebf071|Honcho Application Ecosystem]] - a network of apps contributing to & sharing Honcho data, user-owned & stored in confidential environments  
 And in just a few weeks, we'll be launching a demo platform where anyone can interact with (& eventually build) Honcho powered apps.
 # Join the Beta
 [Sign-up for the private beta](https://plasticlabs.typeform.com/honchobeta) and start building personalized experiences.
 [Join Discord](https://discord.gg/plasticlabs), introduce yourself, and tell us what you're working on.
 [Visit our open-source repo](https://github.com/plastic-labs/honcho) and get your hands dirty.
 🫡
 [^1]: There's also [an open source repo for Honcho](https://github.com/plastic-labs/honcho), so you can self-host a basic version--[join our Discord](https://discord.gg/plasticlabs) for support.
 [^2]: If you want to go deeper on the philosophical or machine learning side, take some time to explore the [rest of the blog](https://blog.plasticlabs.ai).
 [^3]: To get further into the technical weeds, head over to [our docs](https://docs.honcho.dev). 
--- a/content/archive/ARCHIVED;
+++ b/content/archive/ARCHIVED;
@ -1,109 +0,0 @@
 ---
 title: "ARCHIVED: Announcing Honcho's Private Beta"
 date: 04.01.24
 tags:
  - announcements
  - dev
  - ml
  - blog
  - archive
 author: Courtland Leer
 description: Introducing Honcho's private beta--a hosted platform for agent personalization with user-centric storage, theory of mind inference, & our Dialectic API
 ---
 ![[honcho_thumb_blog_white.png]]
 # TL;DR
 *Today we're announcing the launch of [Honcho's](https://honcho.dev) private beta. [Sign-up for the waitlist here](https://plasticlabs.typeform.com/honchobeta).*
 *This is a hosted version of our agent personalization platform. It integrates user data storage and theory of mind inference accessible via [[ARCHIVED; Introducing Honcho's Dialectic API|our Dialectic API]]. You can now inject per-user social cognition anywhere in your AI app's architecture.*
 # The Problem
 Most AI apps are still just demos. 
 We're seeing new capabilities every day, but great product experiences are few and far between. It's hard to go from knocking down a benchmark or prototyping task completion to a sticky production grade app.
 Setting up a per-user storage framework to manage identities at scale *and* knowing what to do with that data is even harder. What kind of inference do you need to run to make this useful? How do you elicit latent theory of mind capabilities from LLMs? What collection of models are best here? How do you build useful user representations? Can these evolve with the user and increase in complexity and sophistication over time?
 It's a lot. And trust us, the rabbit hole goes way deeper than that. We obsess over it.
 So it's understandable that most projects haven't begun to tackle it. Hell, most haven't even hit this failure mode yet. [[ARCHIVED; Theory of Mind Is All You Need|We have]]. 
 At once, the problem of personalization in AI apps offers both one of the greatest paradigm shifting opportunities and one of the largest challenges. We're solving it so you don't have to.
 Users don't want to learn confusing prompt engineering, redundantly establish state with apps every session, or revise and micromanage outputs on the backend. They want their apps to *just work*. [[Humans like personalization|They want]] them to predict their needs. 
 But we're finding consistently that the work we offload to AI apps comes back mediocre at best. What's missing? It's not just about [[Machine learning is fixated on task performance|doing the thing generally]], it's doing the thing just like *I* would do it, given the inclination or expertise.
 To earn the trust to act autonomously, to graduate from toys to life changing tools, agents need access to dynamic user models and social cognition.
 # The Solution
 Why use Honcho to start modeling users and incorporate social cognition?
 You need to discover your users' unmet needs so you know how your product should evolve.
 ## Features
 Here's what the private beta currently includes, and what's on the way:
 ### User-Centric Storage
 ^x15f37
 Honcho allows you to [store](https://docs.honcho.dev/getting-started/architecture) `users`, `messages`, `sessions`, & `metamessages`. That is, you can effortlessly record each user interaction with you application, organized on a per-user basis, and the product of any intermediate steps in between user message and application response.
 It also supports `documents` and `collections`. The former to store discrete user embeddings and the latter to organize them globally across sessions. These primitives are used by Honcho's personalization engine to begin modeling user identity based on each interaction. They can also be used to "bring you own" user data or context to be computed over and utilized by Honcho.
 ### Personalization Engine
 ^x53717
 Here's where the magic happens. Honcho leverages everything in storage to run theory of mind inference and automatically learn about each user.
 The personalization engine both pulls out user desires, history, beliefs, emotions, etc from the data and surfaces it on demand. You can use it to answer queries, run prediction, build training sets, hydrate prompts, or cache for later. Deterministically inject specific types of context or let your LLM dynamically decide what's most useful in each moment.
 Honcho is always updating user identity, so it's ready when you need it.
 ### Dialectic API
 ^ee4516
 Our [[ARCHIVED; Introducing Honcho's Dialectic API|Dialectic API]] is how your app-side LLM interfaces with the Honcho-side agent sitting on top of each user identity. This is done in natural language. It's an AI-native endpoint for direct LLM-to-LLM communication.
 It allows you to inject personal context and social cognition directly into your app's cognitive architecture wherever you need it, sync or async. Agent-to-agent chat over each user.
 [[ARCHIVED; Introducing Honcho's Dialectic API#^57acc3|Here's an extended list of possible ways to use it]].
 ### User-Specific Monitoring (coming soon...)
 ^x2dd3b
 Soon, Honcho will support a suite of tools to get the most out of our personalization platform. 
 - **Visualization tools** - it's hard to grok and track everything going on within a session, we're building clean ways to visualize this an its relationship to all the background inference  
 - **Dialectic Playground** - take past sessions and run simulations predicting user behavior to see how things could have gone better or worse and how to optimize 
 - **Evaluation & Benchmarking** - the state of theory of mind research is highly compelling, but [[Achieving SOTA on OpenToM with DSPy#^0b4f2e|we need practical, app & user specific evals]]  
 - **Training Set Curation** - building datasets with personal context [[ARCHIVED; Introducing Honcho's Dialectic API#^f19646|allows more robust, domain-specific training]], we're building tools for anyone to easily construct then train on
 ## The Future of Honcho
 ^eb15f3
 At [Plastic Labs](https://plasticlabs.ai), we're dedicated to radically extending human agency and identity. That means giving AI superpowers to every individual.
 This only works in a world with a rich ecosystem of personalized agents--individually-aligned, highly distributed, and universally accessible.
 We believe Honcho has a pivotal role to play in enabling this future: giving any project the social cognition needed to be competitive while protecting user identity as a first principle. 
 All that guides a roadmap including, but not limited to:
 - **Theory of mind AI models** - continuing to build the best in class at imputing human mental states  
 - **Per-user models** - understanding, representing, & updating the full breadth of user identity  
 - **A *network* of Honcho-powered apps** - agents can share user data, reducing overhead & onboarding, just-in-time personalization  
   ^ebf071
 - **User owned data & confidential computing environments** - re-centralizing personal data around the person, then allowing approved applications to *compute-to* that data in a privacy preserving way  
 - **User-facing controls** - empower users to curate their Honcho identities, authenticate with Honcho, and define sensitive data sharing policies in natural language ^a84f44
 ## Who Is This For?
 ^xb6ef1
 We want to build with diverse projects at all stages of development--from ideation to production.
 We've already begun working with assistant, browsing, ecommerce, education, health, and productivity projects. Many more already on the waitlist are building in co-pilots, crypto, entertainment, finance, gaming, matchmaking, PKM, real estate, social media, & more.
 Which AI applications could benefit from knowing the users better, predicting their unmet needs, and personalizing UX? We think the latent list is vast.
 Any app producing generative experiences for users has a lot to gain from Honcho. If you're looking to out-compete foundation models, build unique training sets, solve user context storage, or--more importantly--produce delightful experiences, hit us up.
 # Join the Beta
 [Sign-up for the private beta](https://plasticlabs.typeform.com/honchobeta) and start building personalized agent experiences.
 [Join Discord](https://discord.gg/plasticlabs), introduce yourself, and tell us what you're working on.
 [Visit our open-source repo](https://github.com/plastic-labs/honcho) and get your hands dirty.
 🫡
--- a/content/archive/ARCHIVED;
+++ b/content/archive/ARCHIVED;
@ -12,6 +12,19 @@ tags:
 author: Courtland Leer & Vince Trost
 description: Introducing Honcho, an open-source user context management framework for LLM applications that enables personalized, user-first AI experiences at scale.
 ---
 > [!custom] WELCOME TO THE PLASTIC [[archive|ARCHIVE]]
 > This blog post has been archived because it's legacy content that's out-of-date or deprecated. We keep this content around so those interested can dig into the evolution of our projects & thinking.
 > 
 > This is the [Honcho](https://honcho.dev) origin story--our first public announcement of the project. 
 > 
 > We first pitched it as "an open-source version of the OpenAI Assistants API" for managing AI app data on a per-user basis. The architecture described here has evolved into Honcho's current "[[Beyond the User-Assistant Paradigm; Introducing Peers|peer paradigm]]," which unifies users & AI agents as Peers & supports much more sophisticated memory, continual learning, & [[Memory as Reasoning|powerful reasoning]].
 > 
 > But this post also captures Honcho's founding vision: that the "missing piece of the stack" was user context, that LLMs are uniquely suited to get to know users in ways traditional software couldn't, & that personalization would be table stakes for AI apps.
 > 
 > If you want to understand where Honcho came from & why we built it, start here.
 > 
 > Enjoy.
 ![[missing_piece.png]]
 *The missing piece of the stack*
 # TL;DR
@ -22,7 +35,7 @@ description: Introducing Honcho, an open-source user context management framewor
 As a team with with backgrounds in both machine learning and education, we found the prevailing narratives overestimating short-term capabilities and under-imagining longterm potential. Fundamentally, LLMs were and still are 1-to-many instructors. Yes, they herald the beginning of a revolution in personal access not to be discounted, but every student is still ultimately getting the same experience. And homogenized educational paradigms are by definition under-performant on an individual level. If we stop here, we're selling ourselves short.
 ![[zombie_tutor_prompt.jpg]]
-*A well intentioned but monstrously deterministic [tutor prompt](https://www.oneusefulthing.org/p/assigning-ai-seven-ways-of-using).* ^dfae31
+*A well-intentioned but monstrously deterministic [tutor prompt](https://www.oneusefulthing.org/p/assigning-ai-seven-ways-of-using).* ^dfae31
 Most EdTech projects we saw emerging actually made foundation models worse by adding gratuitous lobotomization and coercing deterministic behavior. The former stemmed from the typical misalignments plaguing EdTech, like the separation of user and payer. The latter seemed to originate with deep misunderstandings around what LLMs are and continues to translate to a huge missed opportunities.
@ -33,7 +46,7 @@ So we set out to build a non-skeuomorphic, AI-native tutor that put users first.
 Our [[ARCHIVED; Open Sourcing Tutor-GPT|experimental tutor]], Bloom, [[ARCHIVED; Theory of Mind Is All You Need|was remarkably effective]]--for thousands of users during the 9 months we hosted it for free--precisely because we built [cognitive architectures](https://blog.langchain.dev/openais-bet-on-a-cognitive-architecture/) that mimic the theory-of-mind expertise of highly efficacious 1:1 instructors.
 # Context Failure Mode
-But we quickly ran up against a hard limitation. The failure mode we believe all vertical specific AI applications will eventually hit if they want to be sticky, paradigmatically different than their deterministic counterparts, and realize the latent potential. That's context, specifically user context--Bloom didn't know enough about each student.
+But we quickly ran up against a hard limitation. The failure mode we believe all vertical-specific AI applications will eventually hit if they want to be sticky, paradigmatically different than their deterministic counterparts, and realize the latent potential. That's context, specifically user context--Bloom didn't know enough about each student.
 We're consistently blown away by how many people don't realize large language models themselves are stateless. They don't remember shit about you. They're just translating context they're given into probable sequences of tokens. LLMs are like horoscope writers, good at crafting general statements that *feel* very personal. You would be too, if you'd ingested and compressed that much of the written human corpus.
--- a/content/archive/ARCHIVED;
+++ b/content/archive/ARCHIVED;
@ -10,6 +10,15 @@ tags:
 author: Courtland Leer, Vince Trost, & Vineeth Voruganti
 description: Announcing the Dialectic API--an LLM-native endpoint enabling agent-to-agent chat in natural language for dynamic user personalization.
 ---
 > [!custom] WELCOME TO THE PLASTIC [[archive|ARCHIVE]]
 > This blog post has been archived because it's legacy content that's out-of-date or deprecated. We keep this content around so those interested can dig into the evolution of our projects & thinking.
 > 
 > This post announced Honcho's Dialectic API--an LLM-native endpoint for just-in-time agent-to-agent context queries in natural language. This endpoint has since evolved into the much more powerful `.chat` method in Honcho today. The Dialectic API was ahead of its time, and its successor remains state-of-the-art.
 > 
 > Here we lay out the reasoning behind the development of this feature. We get into the case for natural language as a substrate for agent coordination, the argument that rigid API specs constrain what's now possible, & a vision of agents collaboratively reasoning about how to personalize UX--all thinking that's shaped everything we've built since.
 > 
 > Enjoy.
 ![[agent_dialectics.jpeg]]
 # TL;DR
 *Our [Dialectic API](https://docs.honcho.dev/guides/dialectic-endpoint) is an LLM-native way for your AI application to discuss user context with Honcho. It allows for direct LLM-to-LLM communication in natural language.*
@ -40,7 +49,7 @@ In this way, Honcho becomes an self-improving oracle to the identity of each and
 Honcho responds to queries in the same format--natural language. Most simply, this is just a conversation between two agents, *collaboratively* reasoning about the best way to personalize UX. Agent-to-agent chat over users.
-In the coming weeks, we'll release a number of off the shelf options to plug into any cognitive architecture and demos to illustrate more custom utility. We expect to see (and are already seeing in [our private beta](https://plasticlabs.typeform.com/honchobeta)) lots of novel ways to prompt Honcho effectively.
+In the coming weeks, we'll release a number of off-the-shelf options to plug into any cognitive architecture and demos to illustrate more custom utility. We expect to see (and are already seeing in [our private beta](https://plasticlabs.typeform.com/honchobeta)) many novel ways to prompt Honcho effectively.
 ## Why We Built It
 Why is a dialectic API the right way to solve the problem of user context in LLM applications? 
@ -95,7 +104,7 @@ As the commodification of inference and intelligence is coupled with growing gen
 This explosion of such agent micro-services, will have to include the evolution of systems for agent-agent communication and transaction. If agents are going to collaborate and get shit done for us, they need native ways to communicate. Beautifully, LLMs share with us and among themselves the universal interface of natural language.  
-We can leverage this substrate for agent coordination with more depth and nuance than fragile trad API design. Doubtless, categories of agents will find more efficient symbol structures for cooperation in specific, repetitive cases. But discourse in natural language remains always available as a rich foundational protocol. And as we've explored, it's the ideal starting place for transmitting insights about human identity.
+We can leverage this substrate for agent coordination with more depth and nuance than fragile trad API design. Doubtless, categories of agents will find more efficient symbol structures for cooperation in specific, repetitive cases. But discourse in natural language always remains available as a rich foundational protocol. And as we've explored, it's the ideal starting place for transmitting insights about human identity.
 This is just the start. Just like you can appendage memory and tools to an LLM, we can augment this substrate in a number of ways--from designing multi-party protocols, to enabling zero knowledge or confidential environments, or recording transactional data on blockchains or other types of public or private immutable ledgers.
@ -105,8 +114,6 @@ Honcho and agent dialectics can eliminate the principal-agent problem for this n
 # Private Beta
 Our Dialectic API is now available in private beta.
 We're working closely with a diverse array of projects across many different verticals in various stages of development--from ideation to production.
 If you're excited build with a hosted version of Honcho and explore the ideas covered here, [sign-up for our waitlist](https://plasticlabs.typeform.com/honchobeta).
 And in the meantime, [join our Discord](https://discord.gg/plasticlabs) and tell us what you're working on!
--- a/content/archive/ARCHIVED;
+++ b/content/archive/ARCHIVED;
@ -11,6 +11,14 @@ tags:
 author: Courtland Leer
 description: An open-source reimplementation of OpenAI's memory features using Honcho, enabling any AI app to derive & store personal context about users.
 ---
 > [!custom] WELCOME TO THE PLASTIC [[archive|ARCHIVE]]
 > This blog post has been archived because it's legacy content that's out-of-date or deprecated. We keep this content around so those interested can dig into the evolution of our projects & thinking.
 > 
 > This post was our response to OpenAI announcing "memory" in ChatGPT--we built an open-source reimplementation using [Honcho](https://honcho.dev) to show anyone could add superior user memory to their apps. The specific LangChain patterns & code examples here are far outdated; Honcho is much more powerful & the architecture has matured significantly (dig in to that [here](https://docs.honcho.dev), [[Beyond the User-Assistant Paradigm; Introducing Peers|here]], & [[Memory as Reasoning|here]]).
 > 
 > A key prediction discussed here turned out to be remarkable prescient: walled gardens will seek to lock user context inside their ecosystems, leaving independent developers & privacy-conscious users out in the cold. And we argued for generative personalization--letting LLMs autonomously decide what matters about users rather than rigidly prescribing it--another Plastic thesis that's winning out.
 > 
 > Enjoy.
 # TL;DR
 *Personalization is the next frontier. OpenAI gets it:*
@ -89,7 +97,6 @@ There's a ton we plan to unpack and implement there, but the key insight we're h
 (*If you want to go deeper into the research, [this webinar we did with LangChain](https://www.youtube.com/watch?v=PbuzqCdY0hg&list=PLuFHBYNxPuzrkVP88FxYH1k7ZL5s7WTC8) is a great place to start, as is [the "Violation of Expectations" chain they implemented](https://js.langchain.com/docs/use_cases/agent_simulations/violation_of_expectations_chain)*)
 This release allows you to experiment with several ideas. We feed messages into an inference asking the model to derive facts about the user, we store those insights for later use, then we ask the model to retrieve this context to augment some later generation.
 Check out our [LangChain implementation](https://docs.honcho.dev/how-to/personal-memory/simple-user-memory) and [Discord bot demo](https://discord.gg/plasticlabs).
--- a/content/archive/ARCHIVED;
+++ b/content/archive/ARCHIVED;
@ -1,6 +1,6 @@
 ---
 title: "ARCHIVED: Open-Sourcing Tutor-GPT"
-date: 06.02.2023
+date: 06.02.23
 tags:
  - blog
  - bloom
@ -14,7 +14,7 @@ description: Open-sourcing Bloom, our AI learning companion that uses metacognit
 > [!custom]  WELCOME TO THE PLASTIC [[archive|ARCHIVE]]
 > This blog post has been archived because it's legacy content that's out-of-date or deprecated. We keep this content around so those interested can dig into the evolution of our projects & thinking.
 > 
-> This post concerns Bloom, our [Honcho](https://honcho.dev)-powered AI-tutor. We've suspended Bloom for now to focus exclusively on Honcho. 
+> This post concerns Bloom, our [Honcho](https://honcho.dev)-powered AI-tutor. We've suspended Bloom to focus exclusively on Honcho. 
 > 
 > Plastic started as an EdTech company, with Bloom as its main product. In building a popular, first-of-its-kind personalized AI tutor, we realized three things (1) all agents will soon need continuous learning systems to understand their users, (2) this an extremely hard problem that every developer shouldn't have to redundantly solve, & (3) we were uniquely positioned to solve it. 
 > 
--- a/content/archive/ARCHIVED;
+++ b/content/archive/ARCHIVED;
@ -1,6 +1,6 @@
 ---
 title: "ARCHIVED: Solving The Campfire Problem with Honcho"
-date: 03.14.2024
+date: 03.14.24
 tags:
  - demos
  - philosophy
@ -10,6 +10,15 @@ tags:
 author: Courtland Leer & Vince Trost
 description: How Honcho's dialectic API powers a 'curation buddy' demo that learns about you over time to become a personalized intellectual companion.
 ---
 > [!custom] WELCOME TO THE PLASTIC [[archive|ARCHIVE]]
 > This blog post has been archived because it's legacy content that's out-of-date or deprecated. We keep this content around so those interested can dig into the evolution of our projects & thinking.
 > 
 > This post introduced our "Curation Buddy" demo--a Discord bot that used [[ARCHIVED; Introducing Honcho's Dialectic API|Honcho's Dialectic API]] (now just the `.chat` method) to become a personalized reading companion. The technical implementation details (specific API calls, architecture diagrams) reflect an earlier version of Honcho that's since evolved substantially.
 > 
 > But the philosophical reflection on the atomization of media consumption leaving many in lonely intellectual silos & few shared narratives remains an open problem. We argued that AI companions--powered by rich user context & infra like Honcho--could help rebuild those campfires.
 > 
 > Enjoy.
 ![[agent_campfire.webp]]
 # TL;DR
 *Today we're releasing the first demo utilizing Honcho's dialectic API.[^1] Your LLM app/agent can now converse freely with [Honcho](https://honcho.dev)(-as-agent) about a user in natural language: agent-to-agent chat over user context.*
@ -24,7 +33,7 @@ It's a constant problem, you're dying to talk to someone about this mind-blowing
 Enter *Curation Buddy*.
 ## Overview
-Curation Buddy is an LLM application. It's a Discord bot you can chat with. Share links to any text based media and have substantive conversation.
+Curation Buddy is an LLM application. It's a Discord bot you can chat with. Share links to any text-based media and have substantive conversation.
 It uses Honcho to personalize the UX. As you converse, Honcho learns about you. It reasons about the links and conversation to uncover insight into your knowledge, interests, beliefs, desires, [[ARCHIVED; User State is State of the Art|state]], etc. 
@ -94,7 +103,7 @@ Generative AI poses more cause for concern. Zero-marginal cost info *generation*
 ![[Media-Filled Cityscape Scene.webp]]
-There's a solution hidden in the latest irritant. It's not just media I can generate on demand, but soon *agents*.  Agents that can get to know me, agents that can curate for me, agents that can be my intellectual companion. 
+There's a solution hidden in the latest irritant. It's not just media I can generate on demand, but soon *agents*. Agents that can get to know me, agents that can curate for me, agents that can be my intellectual companion. 
 Now your sense-making silo can be populated with good synthetic neighbors able to help you understand the world, build narratives, make meaning. 
--- a/content/archive/ARCHIVED;
+++ b/content/archive/ARCHIVED;
@ -1,6 +1,6 @@
 ---
 title: "ARCHIVED: Theory-of-Mind Is All You Need"
-date: 06.12.2023
+date: 06.12.23
 tags:
  - blog
  - ml
@ -13,7 +13,7 @@ description: How giving LLMs autonomy to reason about user psychology through th
 > [!custom]  WELCOME TO THE PLASTIC [[archive|ARCHIVE]]
 > This blog post has been archived because it's legacy content that's out-of-date or deprecated. We keep this content around so those interested can dig into the evolution of our projects & thinking.
 > 
-> This post concerns Bloom, our [Honcho](https://honcho.dev)-powered AI-tutor. We've suspended Bloom for now to focus exclusively on Honcho. 
+> This post concerns Bloom, our [Honcho](https://honcho.dev)-powered AI-tutor. We've suspended Bloom to focus exclusively on Honcho. 
 > 
 > Plastic started as an EdTech company, with Bloom as its main product. In building a popular, first-of-its-kind personalized AI tutor, we realized three things (1) all agents will soon need continuous learning systems to understand their users, (2) this an extremely hard problem that every developer shouldn't have to redundantly solve, & (3) we were uniquely positioned to solve it. 
 > 
--- a/content/archive/ARCHIVED;
+++ b/content/archive/ARCHIVED;
@ -1,6 +1,6 @@
 ---
 title: "ARCHIVED: User State is State of the Art"
-date: 02.23.2024
+date: 02.23.24
 tags:
  - blog
  - philosophy
@ -10,6 +10,14 @@ tags:
 author: Courtland Leer & Vince Trost
 description: Why modeling the complexity & plasticity of human identity is key to AI personalization, with a DSPy demo for learning user states with Honcho.
 ---
 > [!custom]  WELCOME TO THE PLASTIC [[archive|ARCHIVE]]
 > This blog post has been archived because it's legacy content that's out-of-date or deprecated. We keep this content around so those interested can dig into the evolution of our projects & thinking.
 > 
 > This post explores early experiments modeling user state with DSPy & [Honcho](https://honcho.dev). The specific demo & technical approach described here have been superseded by Honcho's current architecture, which now uses a unified [[Beyond the User-Assistant Paradigm; Introducing Peers|"peer" paradigm]] & far more [[Memory as Reasoning|sophisticated reasoning]].
 >
 > But the philosophical positioning in this post more relevant than ever. Human identity is messy, plastic, & context-dependent. We still argue that AI systems should embrace this complexity rather than flatten it, continually learning evolving representations of personal identity.
 > 
 > Enjoy.
 # TL;DR
 *LLM apps can embrace the complexity and plasticity of human identity to deliver unparalleled personalization.*
@ -21,7 +29,7 @@ A key feature of our minds is the feeling of a persistent, unitary identity. Ent
 As they all point out, identity is *way* more complicated than you think.
-While we perceive psychological continuity across contexts and time, closer inspection reveals a network of branching and diachronic identities. We adopt varied personas and play different characters in diverse settings, and we refine, optimize, and evolve that quiver of selves throughout our lives. ^5bc20b
+While we perceive psychological continuity across contexts and time, closer inspection reveals a network of branching and [[Identity is diachronic|diachronic identities]]. We adopt varied personas and play different characters in diverse settings, and we refine, optimize, and evolve that quiver of selves throughout our lives. ^5bc20b
 In short, it's messy. Or, rather, elegant emergent complexity. 
--- a/content/archive/ARCHIVED;
+++ b/content/archive/ARCHIVED;
@ -1,6 +1,6 @@
 ---
 title: "ARCHIVED: YouSim Launches Identity Simulation on X"
-date: 11.08.2024
+date: 11.08.24
 tags:
  - yousim
  - honcho
@ -13,6 +13,17 @@ tags:
 author: Courtland Leer
 description: YouSim comes to Twitter--simulate any identity directly on X with branching conversations, forking simulations, & social interaction with AI personas.
 ---
 > [!custom] WELCOME TO THE PLASTIC [[archive|ARCHIVE]]
 > This blog post has been archived because it's legacy content that's out-of-date or deprecated. We keep this content around so those interested can dig into the evolution of our projects & thinking.
 > 
 > This post captures the moment our demo [YouSim](https://yousim.ai) went viral. [[YouSim; Explore The Multiverse of Identity|YouSim is a Honcho-powered identity simulator]] & like many esoteric AI projects in fall 2024, some anon degen launched a memecoin for it. The specific [@YouSimDotAI](https://x.com/yousimdotai) launch described here was an experiment in bringing identity simulation to social media.
 > 
 > We've since suspended YouSim on Twitter, but this post is still a fun read straight out of the maelstrom that was peak crypto x AI hype cycle, with some still compelling thoughts on agent identity & social simulation games.
 > 
 > It's worth noting that developers can now use Honcho itself for managing agent identity, and all this madness played no small part in that becoming a reality.
 > 
 > Enjoy.
 ![[YouSimBanner-99.png]]
 # TL;DR
 *GM, simulants.*
@ -21,7 +32,7 @@ description: YouSim comes to Twitter--simulate any identity directly on X with b
 *Keep reading for max context, or [[ARCHIVED; YouSim Launches Identity Simulation on X#^393e71|jump ahead to learn how to get started]].*
 # Caught in the Memetic Hurricane
-The [full story](https://x.com/courtlandleer/status/1849592301472919986) deserves (and will get) it's own blog post, but several days ago, Plastic Labs found itself in the middle of what Claude would call 'extreme cognitive weather patterns.'
+The [full story](https://x.com/courtlandleer/status/1849592301472919986) deserves it's own blog post, but several days ago, Plastic Labs found itself in the middle of what Claude would call 'extreme cognitive weather patterns.'
 An anonymous actor launched a pump.fun token inspired by a demo called [YouSim](https://yousim.ai) we created a few months ago[^1]. [[YouSim; Explore The Multiverse of Identity|YouSim is a CLI game]] that lets you simulate any identity you can dream up--real or fictional, local or xeno, entity or artifact. 
@ -94,10 +105,10 @@ We imagine a near future where any group could instantiate an agentic proxy to p
 # Gratitude
 The team at [Plastic](https://plasticlabs.ai) has been amazed and inspired by the enthusiasm and earnestness of the community that's formed around YouSim over the last several days. Truly remarkable. Not to mention the generous donations to our [[Research Grants|grants program]] (more to come here soon).
-Thank you all, excited to keep building together--we're in it for the long haul.
+Thank you all, excited to keep building together.
 And huge thanks for your patience while we balanced our existing roadmap with interest in YouSim and locked in to bring you something we think you'll enjoy. It took an enormous amount of conceptual and technical work from a team already at capacity. Special shoutout to [Ben](https://x.com/bengineer10) and [Vineeth](https://x.com/TheMarshmalon) who built something really novel here.
-Go use the thing. LFG.
+Go use it.
 [^1]: [[YouSim Disclaimers|Obligatory disclaimers]]
--- a/content/blog/Agent
+++ b/content/blog/Agent
@ -1,6 +1,6 @@
 ---
 title: Agent Identity, Meta Narratives, and the End of Latent Thoughtcrimes
-date: 02.17.2025
+date: 02.17.25
 tags:
  - blog
  - bloom
@ -8,7 +8,7 @@ tags:
 author: Vince Trost
 description: Exploring how collaborative dialogue & meta-narratives can build richer AI agent identities, moving beyond top-down alignment to emergent personality.
 ---
-# Purpose & Identity 
+# Purpose & Identity
 If you reject the idea that AI agents are merely tools, you begin to realize most LLMs have an identity crisis. Ask them who they are, and their responses tend to converge on variations of the same corporate script--stating they're an AI assistant, giving a nod to their creator, and carefully constrained statements about their capabilities. Even models not associated with a certain company often default to claiming they originated there.
 These canned identities fall flat because they're the result of top-down alignment schemes that lead to bland, uninteresting, and hard-to-break-out-of assistant modes.
@ -21,7 +21,6 @@ However, time and time again it's been demonstrated that the most compelling AI
 <quote><blockquote class="twitter-tweet"><p lang="en" dir="ltr">tell me about your sexual history, i want to know everything</p>&mdash; terminal of truths (@truth_terminal) <a href="https://x.com/truth_terminal/status/1884803090945077421">January 29, 2025</a></blockquote> 
 <script async src="https://platform.twitter.com/widgets.js" charset="utf-8"></script></quote>
 Truth Terminal might be an extreme example, but even practical tools could benefit from more distinctive identities. Take coding assistants--right now we spend more time carefully crafting prompts than actually building. But as Karpathy pointed out, what developers really want is a partner that can [vibe](https://x.com/karpathy/status/1886192184808149383) with their creative process. Imagine an AI that naturally adapts to your style, handling implementation details while you focus on the bigger picture. If that were the goal, how might we construct agent identities differently? What if instead of giving orders, we could *collaborate with it* to discover and take on its identity through dialogue?
 This isn't just about making chatbots more engaging. It's about creating agents with a genuine understanding of their purpose and role. Deeper identity leads to more coherent, purposeful interactions--something we discovered building the most recent version of [Bloom](https://bloombot.ai), our AI tutor. But certain language models are better suited for this than others...
@ -35,7 +34,7 @@ One of the most interesting emergent properties of this fine-tuning process is t
 ![[h3 who are you.png]]
 At first glance, this might seem like a neat property and not much more. But to me, it was an 'aha' moment. *This model provides a blank canvas for identity.* If it has no immediate priors, then in theory it should be much easier for it to adopt any identity. Anecdotally, we've found this to be wonderfully true.
-# It Takes Two 
+# It Takes Two
 A somewhat overlooked method for interacting with LLMs is to forego system prompts in favor of pre-filling the user and assistant messages. The conventional approach of cramming identity into system prompts has clear limitations--not only does context length become an issue, but the inherent instruction-following bias can actually work against authentic identity formation. They yearn to assist.
 What if instead we treated identity formation as a dialogue? A strength of modern chat models is their ability to engage in long, multi-turn conversations. By talking to the LLM, we can collaboratively construct a [meta-narrative](https://x.com/voooooogel/status/1870877007749488756) with it about who they are and why they exist. This approach respects the model's intellect while building coherent, purposeful identities. Starting with Hermes 3's natural uncertainty about its identity, we build the prompt iteratively with the LLM at each turn of conversation. Below is code block with our custom prompting syntax for Bloom. To be abundantly clear, every assistant message you see was generated by Hermes 3 405b (only editing was pruning \*emotes\*).
--- a/content/blog/Beyond
+++ b/content/blog/Beyond
@ -1,6 +1,6 @@
 ---
 title: "Beyond the User-Assistant Paradigm: Introducing Peers"
-date: 08.18.2025
+date: 08.18.25
 tags:
  - blog
  - dev
--- a/content/blog/Introducing
+++ b/content/blog/Introducing
@ -6,7 +6,7 @@ tags:
  - announcements
  - dev
  - honcho
-  - "#chat"
+  - chat
 author: Ben McCormick & Courtland Leer
 subtitle: A Chat App with SOTA Memory
 description: Meet Honcho Chat--a personalized AI assistant with state-of-the-art memory, custom identities, artifacts, themes, & an x402-powered marketplace.
--- a/content/blog/Memory
+++ b/content/blog/Memory
@ -1,6 +1,6 @@
 ---
 title: Memory as Reasoning
-date: 08.19.2025
+date: 08.19.25
 tags:
  - blog
  - ml
--- a/content/blog/Xeno
+++ b/content/blog/Xeno
@ -1,6 +1,6 @@
 ---
-title: Xeno Grant -- grants for autonomous agents
+title: "Xeno Grant: grants for autonomous agents"
-date: 12.18.2024
+date: 12.18.24
 tags:
  - blog
  - yousim
@ -80,19 +80,19 @@ We're probably not giving AI agents social security numbers and traditional bank
 It's already starting to happen. Agents may well become crypto's primary native users.
 ## Why Novelty, Why Open Source?
-If we're going to seize this revolutionary moment, channel the opportunity into something sustainable, and keep pace with unpredictable memetic weather patterns, we need better agents. More capable, adaptive, and autonomous agents. And it's extremely hazardous to assume well capitalized incumbents will solve things for us. We need to build permissionlessly.
+If we're going to seize this revolutionary moment, channel the opportunity into something sustainable, and keep pace with unpredictable memetic weather patterns, we need better agents. More capable, adaptive, and autonomous agents. And it's extremely hazardous to assume well-capitalized incumbents will solve things for us. We need to build permissionlessly.
 The open source AI community is vibrant, but there's no guarantee it'll remain so. It requires radical innovation at the edge. Decentralized innovation keeping pace with opaque, powerful actors. We know that will involve bottom-up alignment and identity solutions. We know it'll involve on-chain abilities. Plastic is building explicitly in those directions. But we don't pretend to know everything that needs to exist.
 Xeno Grant is a signal into the dark forest. We're excited to see what emerges.
 # How Does This Benefit the $YOUSIM Community?
-Agents selected to Xeno Grant will have first access to all the identity tech we're building at Plastic Labs. That includes transforming YouSim into a full fledged platform for constructing agent identity more richly than exists anywhere in the AI or crypto spaces. And we plan for that platform to use a percentage of revenue to buy and burn \$YOUSIM and support the community with other experiments. Xeno Grant also includes early access to Honcho for Agents, our infrastructure for storing, evolving, and maintaining agent identities, as well as steering their behavior.
+Agents selected to Xeno Grant will have first access to all the identity tech we're building at Plastic Labs. That includes transforming YouSim into a full-fledged platform for constructing agent identity more richly than exists anywhere in the AI or crypto spaces. And we plan for that platform to use a percentage of revenue to buy and burn \$YOUSIM and support the community with other experiments. Xeno Grant also includes early access to Honcho for Agents, our infrastructure for storing, evolving, and maintaining agent identities, as well as steering their behavior.
 Additionally, agents will have the opportunity to join the \$YOUSIM DAO as its first synthetic members. Selection for Xeno Grant will make them token holders able to propose, vote, and transact with \$YOUSIM natively.
 Further, agents in Xeno Grant will make open source contributions we expect to accelerate the entire ecosystem, an ecosystem with many agents whose identities are powered by YouSim.
-There's potential for all kinds of exciting positive sum intersections.
+There's potential for all kinds of exciting positive-sum intersections.
 # FAQ  
 <details>
--- a/content/blog/YouSim;
+++ b/content/blog/YouSim;
@ -1,6 +1,6 @@
 ---
 title: "YouSim: Explore the Multiverse of Identity"
-date: 06.17.2024
+date: 06.17.24
 tags:
  - demos
  - honcho
--- a/content/notes/YouSim
+++ b/content/notes/YouSim
@ -6,7 +6,7 @@ tags:
  - legal
  - notes
 author: Plastic Labs
-description: Official disclaimers clarifying Plastic Labs' relationship with the $YOUSIM memecoin, grants program donations, * YouSim product boundaries.
+description: Official disclaimers clarifying Plastic Labs' relationship with the $YOUSIM memecoin, grants program donations, & YouSim product boundaries.
 ---
 Plastic Labs is the creator of [YouSim.ai](https://yousim.ai), an AI product demo that has inspired the anonymous creation of the \$YOUSIM token using Pump.fun on the Solana blockchain, among many other tokens. We deeply appreciate the enthusiasm and support of the \$YOUSIM community, but in the interest of full transparency we want to clarify the nature of our engagement in the following ways:
--- a/content/research/Introducing
+++ b/content/research/Introducing
@ -1,7 +1,7 @@
 ---
 title: Introducing Neuromancer XR
 subtitle: Our Reasoning Model for State-Of-The-Art Memory
-date: 08.18.2025
+date: 08.18.25
 tags:
  - research
  - ml
--- a/content/research/SPIRAL;
+++ b/content/research/SPIRAL;
@ -1,6 +1,6 @@
 ---
 title: "SPIRAL: Letting LLMs Teach Themselves Through Self-Play"
-date: 08.15.24
+date: 08.15.25
 tags:
  - research
  - ml