Latest posts
latest version: enough-016b-071626.zip (2.4 MB) it felt like time for a new post because, well, despite my comment in the philosophical questions category about not having a roadmap at launch … i have added a LOT of stuff, relatively speaking. wikisink, a mode to browse wikipedia offline, save/lock/comment on articles, and update the contents when new offline archives are published (or even more…
here are instructions for installing enough. enough-v015b-2.zip (1.1 MB) you’ll need Administrator privileges on MacOS 13+ and a machine with Apple Silicon, as well as ~50GB of disk space for all the model weights, though ~20GB should suffice for the two “smaller” weights options. you’ll also need to have homebrew and git installed; the homebrew setup will handle both if you don’t already have…
that is the question i asked myself in spring 2026. more precisely: “when will all of these large language model capabilities finally be ‘enough?’” and my answer was “maybe they already are.” i had worked with LLMs for 10 years in various creative capacities, but was feeling overwhelmed with how much capability i already have available to me and my many previously impossible dreams. the…
the name comes from the idea that we already have enough useful and usable text manipulation tools, which, due to their component parts’ licensing schemes, distribution mechanisms, and offline functionality, will still be accessible in 20 or 100 years. things like linux, HTML, python, offline weights, the vast expanse of Apache 2.0/MIT-licensed repositories, etc. ux-wise, i realized i wanted a…
i am definitely imposing one paradigm on users, namely: this is how i think user-configurable text planning and editing software might work better for writers. but that feels unavoidable. at its core, enough expects users to customize their experience by designing and using the workflows they prefer to use for planning or writing what they want to write the way(s) they want to do it, rather than…
it runs a server in your terminal, and its ui is your modern web browser (Safari, Chrome, Firefox, glad to look into fixing bugs on others if you find and report them). currently, it’s supported for MacOS on Apple Silicon only because that’s what i test it on. to get the most out of the offline models, you will need a relatively powerful computer for the time being. i develop and test enough on a…
first of all, don’t worry … openrouter seems safe, and it’s also probably “fine” to send your data to Anthropic or OpenAI or Google through their proprietary harnesses! but as soon as you start sending requests to a web server (or router), you are passing your data to openrouter, as well as any intermediaries openrouter might send your requests through, and of course any cloud model providers you…
an application wrapper – which would likely be a stripped-down custom browser anyway – is definitely a feasible alternative, and it’s something i’m still considering for the future. but the browser seems optimal with the existing feature set, for three primary reasons: a wrapper does not allow you to use enough side by side with other tabs, where you are likely to be doing research or using other…
in theory, you could likely set up paradigms and skills to do this pretty well. but there’s probably a much better option, because there are a LOT of coding-focused harnesses, and not only the corporate ones. enough is primarily oriented towards planning, writing and reviewing prose or poetry-type texts. also … well, probably no need to even open this can, but hey, it’s personal writing: i don’t…
it’s a great question; when i try to answer it inside my head, i get really nervous. the structure of enough suggests i am a particular “type” of person that i am not. i generally do not mind giving up my own personal privacy in exchange for access to the latest cool cloud computing tools. i do not write anything particularly sensitive, much less in need of obfuscation against the prying eyes of…
only you can answer that, but by way of explanation: for many years, i was terrified of the command line and unix/gnu/linux and all that, despite being fascinated by software development and publishing. as far as i can tell, nobody knew this about me, that i had this fear. in fact, many people apparently thought i was a “technical person.” and i mean, i was, but i “kept my hands clean” after ~2004…
even though i personally do NOT hate generative ai, as i feel i must disclaim, i do feel like i’ve already gotten tired of seeing it everywhere, generated prose in particular. i want more personal human writing. as a result, enough is designed to lean away from “generate me an entire {text type}” use cases. it gives you the ability to customize a planning, research, proofreading, and translation…
i have done everything i can to make using markdown more comfortable in enough. i too have never been a huge fan, though i’m coming around. i kind of prefer plaintext, tbh. which, it’s worth pointing out, works just fine in enough. but markdown is very useful for interacting with LLM agents, and hopefully with a little practice, the advantages to human readers (and writers) will become apparent. i…
the lowercase formatting helps to distinguish help content or documentation from system text. in the ui, the lowercase help text contrasts with the “proper” formatting in the markdown documents you edit and manage to guide enough’s agent behavior. it’s also just how i generally write, and i find it helps readers distinguish human writing from generated prose.
look, some stigmas you can’t just wave away. i personally love em dashes, or rather, i prefer to use what i call nn dashes – like that. even though google docs always converts them into good old — anyway, things are different now. by the way, it is totally possible that some lowercase ui guidance text may have been written by an LLM coding agent, reviewed, and deemed accurate by me, a human. i’m…
perhaps i will give it a shot. in that spirit, i think this quote from a prominent literary fiction author should not have caused them so much trouble. When writing my latest novel… I asked this advanced model what kind of songs my protagonists would be listening to at a dance, a few dozen years ago, and AI gave me a few titles … Often I just ask the machine, ‘darling, how could we develop this…
i try to suppress most certainties that arise within me, but yes, this feels comfortable, inoffensive, and consistent with enough’s approach. oh, and if you’re wondering, they were posted in reverse chronological order due to the way Discourse organizes lists of posts by default, and with replies disabled to, uh, maintain canonicality. all typos or bizarre turns of phrase are my own (i’m graham…
… and their answers, from Graham Smith, the writer behind enough. If I get asked any other questions (twice or more), I’ll add them.
A place to discuss enough and its various use cases.
This category is the repository of official documentation for enough, as well as targeted onboarding assistance, demonstration videos, FAQ type answers, and other helpful information. Note that new topics and replies are blocked, but the “enough discussion” category is a great place to bring questions needing answers.
This is a Civilized Place for Public Discussion Please treat this discussion forum with the same respect you would a public park. We, too, are a shared community resource — a place to share skills, knowledge and interests through ongoing conversation. These are not hard and fast rules. They are guidelines to aid the human judgment of our community and keep this a kind, friendly place for civilized…
Meta-discussion about the structure of this site, its categories and their organization, how narrow-domain conversations and skill-sharing work in real life vs. (the admin’s precious) theory, and how we all can improve and expand the forum’s utility for users.