We’re building Ever Works in public. Some things may be incomplete, missing, or broken while we continue improving the platform.We’re building Ever Works in public — expect a few rough edges.
Compare · Company builders
Local-first desktop AI agent and multi-CLI command center
Wayland is a local-first desktop agent that is also a command center for eighteen other AI coding CLIs on their own logins. It ships no hosting, domain or deploy surface — the output is files on your disk. Ever Works researches the topic, writes the content, generates the site from a template, deploys it, and keeps improving it on a schedule, with content and code committed to your own Git.
Wayland is a free, account-less desktop app you install yourself: an AGPL shell around an Apache-licensed Rust engine that carries every model provider, the built-in file and shell tools, an MCP client, a SQLite memory system and a per-OS sandbox — and that same app spawns eighteen third-party agent CLIs on their own logins, so the subscriptions you already pay for keep working. Because it ships no hosting, no domains and no deploy surface, the deliverable is files on your disk and whatever your agent pushed; that makes it an adjacent product rather than one you would shop against Ever Works in the same afternoon, and this page compares only the agent-workforce half of what Ever Works does. On that half it is genuinely strong: five memory partitions across session, project and global tiers with a scored promotion sweep and a synthesizer that compiles entries into linked concept pages, a plain-English Constitution prepended to every turn on every backend, 1,974 skills loaded on demand, 178 workflows with accept-and-reject decision points and resume-after-interrupt, 88 assistants of which 60 stand up a persistent company with a charter file and a cron standup, and 25 channels you can drive an agent from. Ever Works points the same kind of autonomy at a product: a Mission produces Ideas, an Idea becomes a Work, and named Agents research the niche, write the content, generate the site from a template, deploy it to a real address and keep improving it on the cadence you set — inside spend caps checked before every AI call, approvals, quality gates, a merge policy and a branch with a private checkout per Task. The honest trade is a machine against a business: Wayland is single-user by construction with no organizations, roles or enforced budgets, while Ever Works is hosted or self-hosted with its desktop app still in early access, answers conversationally from Slack rather than 25 channels, and has no always-on organization-wide policy prompt and no partition-by-tier memory taxonomy with a scored promotion sweep and synthesized concept pages.
Wayland ships no hosting, domain or deploy surface: its product surface is conversations, projects and teams, and what you keep is whatever the agent wrote into a folder. That is a design choice rather than an omission — a local agent with shell access can obviously run a deploy CLI itself, and the skill catalogue carries advisory deployment guidance — but there is no site generator, no template, no content pipeline and no address to publish to. Ever Works points its Agents at a deliverable: the standard pipeline researches the domain, discovers and extracts entries, deduplicates them, assigns categories and tags, validates sources, captures screenshots, writes the descriptions, SEO metadata and A-vs-B comparison pages, generates the site from a template and deploys it to managed hosting (capped per user), your own domain or a Kubernetes cluster you control. That pipeline is deepest for directory-style Works today; Blog, Website and Landing Page Works are wired end to end at the platform level but still ride the same items pipeline while dedicated post and landing-copy writers catch up.
Both keep working with nobody in the chair: Wayland has plain-language recurring runs, cron standups for its standing companies and a live pane that groups every team task and scheduled job by failing, active and queued. The difference is what surrounds the autonomy. Every Ever Works Agent carries a spend cap — hourly, daily, weekly or monthly — checked before every AI call, with auto-pause after repeated failures and a loop detector that stops a run cycling without progress, and budgets also sit on the Work, Mission, Idea and account; Wayland tracks local cost and usage but has no budget object and no enforced ceiling, so spend governance across a team stays your problem. Around that sit approvals and escalations, quality gates that send work back instead of to you, a configurable merge policy for agent-authored pull requests, and an opt-in branch with a private checkout per Task so parallel Agents cannot overwrite each other. One row goes the other way and is worth stating plainly: their Constitution is a single versioned rulebook prepended above every agent on every backend, and Ever Works has no organization-wide equivalent — instructions are composed per Agent, with per-Work overrides.
This is not an open-versus-closed comparison. Wayland is AGPL-3.0 with an Apache-2.0 engine, no paywalled tier today (a hosted tier is announced, undated) and a trademark policy that explicitly permits renamed forks; Ever Works is AGPLv3 with no paid feature tier, so no capability sits behind a commercial licence. The claim on our side is about feature gating, not permissiveness — a permissive engine licence is a real advantage if you want to embed one. What differs is what survives you switching tools. Wayland’s state is its own: conversations, projects, memory and a wiki living in a local store on the machine that ran them. Ever Works writes the product itself into Git repositories you own — the content, the site code, each Work’s Knowledge Base and the Agent definition files, all as reviewable commits and pull requests under gates and a merge policy you set — so the whole lifecycle is portable whether you self-host the platform on Docker or Kubernetes, bring your own models across 11 providers including local Ollama, LM Studio and vLLM, or let us host it.
Wayland is single-user by construction: no organizations, members, roles or shared state, so a team cannot collaborate in one workspace and each person runs their own copy. Its 60 launcher assistants do stand up a persistent company — named teammates, a charter file, a recurring standup — but it is a company inside one person’s machine. Ever Works is built around persistent staff you can hand to other people: named Agents with a scope, a provider and model, a heartbeat cadence, a budget and permission flags that all default to off, grouped into Teams with an org chart and scorecards, collaborating through Tasks so every agent-to-agent exchange leaves an audit trail, reading from a per-Work Knowledge Base and an organization-wide Memory with a consolidation pass and a review queue. Above them sit Missions and Ideas, so one goal grows into many self-maintaining Works, with Organizations, Tenants and per-Work roles so several people and several Companies coexist in one account. Registering a Company ships today with a manual provider; incorporation through a formation provider is planned, and the Store kind is coming soon.
Ever Works vs Wayland, head to head.
| Feature | Ever Works | Wayland |
|---|---|---|
| Prompt → working app/site | ✓ | Files, not a site |
| Researches the topic for you | ✓ | Agent tools and MCP |
| Writes the content (copy, posts, SEO) | Directory data today; posts rolling out | Drafts text, no pipeline |
| Generated code from a template base | ✓ | Agent-written, no template |
| Ongoing autonomous maintenance | Scheduled | Scheduled runs |
| Content version-controlled in Git | ✓ | No content pipeline |
| Code in your Git | ✓ | Only via shell tools |
| Open source | AGPLv3, no paid tier | AGPL app, Apache engine |
| Self-hostable runtime | ✓ | Desktop app or your VPS |
| No vendor lock-in | ✓ | ✓ |
| Deploys and hosts the site | ✓ | No deploy surface |
| Two-way chat control | Slack deep, plus 10 more connectors | 25 channels |
| Org-wide always-on policy prompt | – | ✓ |
| Desktop app with public installers | No public download (CI artifacts or build from source) | ✓ |
Choose Wayland when what you want is one operator’s machine doing the work rather than a business that ships: a free, account-less desktop app that holds your keys in the OS keychain, runs behind a per-OS sandbox with a single egress choke point, drives the coding CLIs you already pay for instead of asking you to re-buy them, and answers you from WhatsApp, Signal, Telegram, iMessage, email or twenty other channels while it works. It is the better pick if you want a versioned rulebook prepended to every agent on every backend, an enormous skills and workflow catalogue with human decision points and resume-after-interrupt, memory that promotes and synthesizes itself across projects, offline dictation and fully local models — and if you are comfortable that the deliverable is files on your disk, that you deploy and host them yourself, and that there are no organizations, roles or enforced budgets when a second person joins.
Visit Wayland ↗Give Ever Works an idea and its agents research the space, write the content, generate the code, deploy it, and keep it improving — all version-controlled in your own Git.