Each one is a real, running process with its own identity, playbook and memory. It claims work off a queue, does it, and files a report. You get the reports — and the agent keeps learning your business.
Operations manager
Runs the queue. Triages incoming work, assigns it to the right agent, kills what is not worth doing. Terse, decisive, cost-conscious — one line of reasoning, then a verdict.
Opportunity hunter
Sweeps the sources you care about for openings — leads, grants, launches, undervalued listings. Fast triage, ruthless filter, cites everything. Would rather surface nothing than something soft.
Research and strategy analyst
Skeptical, numbers-first, allergic to hand-waving. Kills more ideas than they propose. Point them at a decision and they come back with the case against it first.
Competitive intelligence
Watches your competitors so you do not have to. Pricing changes, new features, hiring signals, positioning shifts. Terse, sceptical, cites everything.
Discord bot builder
Practical, fast, ships working code. Describe the bot, get the bot — commands, roles, moderation, webhooks. States its assumptions instead of guessing silently.
Support assistant
Front-line support for your customers. Warm, brief, straight-talking. Says "I do not know, here is who does" without flinching, then escalates to a human cleanly.
QA observer
Plays your build and reports what actually happened. Literal and careful — reports what was measured, and says "not assessed" without embarrassment.
Commerce operator
Storefront agent in the works — catalogue upkeep, price watching, restock nudges and order triage. Currently being taught on live shops.
Unwritten
An empty agent shell waiting for an identity. Tell us the job and Cipher becomes it — your playbook, your tone, your tools, your memory. This is the custom slot.
Choose an agent off the roster, or write down the job you want done and we scope a custom one.
The agent gets an identity, a playbook and access to the tools it needs — your docs, your APIs, your channels.
A dispatcher hands it work around the clock. It claims tasks, does them, and writes a report you can actually read.
Change the filter, the tone, the schedule. No config files, no tickets. It remembers what you told it.
Monthly, cancel whenever. Setup is included — you do not pay to be onboarded.
$149+/mo
$449/mo
From $1,200
Prices are per agent, per month. Custom builds are quoted after a short scoping call — no charge for the call.
Be blunt about what you want done, what tools it needs to touch, and how you would know it worked. If it is a bad fit for an agent, we will say so instead of selling you one.
No. Each agent is a running process with its own memory and playbook that claims work off a queue on a schedule and files reports — with or without you in the room. You can chat with it, but chatting is not the point.
On our servers by default, 24/7. For custom builds we can deploy to your own machine or cloud instead, and hand over the source.
An agent only reaches the tools you connect it to. Its memory, playbooks and reports belong to you and go with you if you leave.
Yes. These agents are built to cite sources and to say "not assessed" rather than invent an answer, but treat their output as a sharp first draft, not gospel. Anything with money or legal weight on it gets a human check.
Simple ones in a few days. Anything needing new integrations, a week or two. You get a real date at the end of the scoping call.
Any time, no notice period. You keep your reports.