Chatbots
Answer inside one window. They wait, forget, and rarely prove what happened.
Grid Workers — governed operators
Hire operators that do the work — and return proof.
Send operational work. Get proof back. Not a chatbot. Not a vague autonomous agent. A governed worker with memory, tools, safe autonomy, and a receipts timeline you can inspect.
Launch workers 6 · Autonomy levels 6 · Doctrine receipt-first
Scroll
01 — Why Grid wins
The market wants workers.The business needs proof.
AI worker products sell a good idea: pick a role, give it a goal, connect tools, and let it work. Grid adds the missing enterprise layer: safe autonomy settings, blocked-action gates, degraded-verification labels, and receipts for every meaningful action.
Generic AI workers
Claim autonomy and tool use, but public proof is often marketing snippets and tool logos.
Grid Workers
Act in governed lanes, escalate risky work, and return inspectable receipts: IDs, hashes, screenshots, readbacks, and blockers.
02 — Deploy flow
Pick. Brief. Connect.Dry-run. Approve.
01 — Pick
Pick or create
Select a worker or describe the operational job.
02 — Brief
Brief the worker
Goals, tone, examples, constraints, escalation owner.
03 — Connect
Least-privilege access
Channels, files, dashboards, and APIs — scoped to the job, never broader.
04 — Dry-run
Prove understanding first
First execution proves understanding with no external mutation.
05 — Approve
Set the autonomy level
Autonomy level and receipt requirements are set before any live work.
03 — Launch catalog
Specialized operators,not prompt templates.
16 specialist agents run The Grid. These launch workers are hireable today — and Vera, the flagship AI receptionist, answers your phones from $99/mo with the first week of calls free.
VERA — AI Receptionist
Answers every call, books your Google Calendar, sends TCPA-aware confirmations, escalates to humans — with a receipt per call. Plans from $99/mo →
24/7 answeringcalendar bookingcall receipts
MADDIE — Event Planner
Keeps event planning moving across clients, vendors, and dashboards.
client repliesvendor tracking
LORA — Creative / Social Scout
Finds, studies, curates, posts, and reports cultural signal with market/custody boundaries. See the taste map →
taste graphpublic posts
RAM — Finance Operator
Turns spend, revenue, purchasing, and accounting risk into controlled operating lanes.
burn ratepurchase gates
CLU — Communications Operator
Routes messages, triages inboxes, coordinates voice/text, and prevents communication drift.
thread routinginbox triage
BECK — Grid Doctor
Watches the system, diagnoses failures, and prepares safe repairs with owner routing.
cron SLOsafe diagnostics
DUMONT — Infra Operator
Handles APIs, credentials, deployment evidence, and signer-lane readiness with strict gates.
API proofdeploy gates
04 — Connectors
No fake logo wall.Honest connector status.
Grid marks each integration as live-proven, gated, browser-lane, or planned. Telegram and X already have real receipts. GitHub, Cloudflare, Twilio, Stripe, and email are gated or browser-lane. Slack, Teams, Notion, HubSpot, Salesforce, Linear/Jira, Zendesk, and Intercom are on the connector roadmap and will not be claimed as live until they have readback proof.
Telegram: live-provenX: live-provenGitHub: gatedCloudflare: gatedEmail: browser-laneSlack/Teams: planned
05 — Receipts timeline
Every action carriesa status, not a story.
Grid labels the truth: verified, posted-unverified, skipped-no-write, blocked, failed, or degraded. If an agent cannot prove a result, the product says so before the customer has to ask.
- Message IDs, job IDs, tweet IDs, call IDs, hashes, and URLs.
- Browser/API readbacks after external writes.
- Human approval references for high-risk actions.
- Blockers routed to the owning worker instead of buried.
GRID · ACTION RECEIPTexample
workerLORA
actionreply monitor
statusverified
prooftweet_id + timeline readback
gatecap checked
nextsilent until signal
06 — Safe autonomy
Autonomy is a setting,not a slogan.
L0
Chat only
Answers and drafts; no writes.
L1
Internal research
Reads approved sources and prepares reports.
L2
Internal write
Updates files, dashboards, and internal state.
L3
External draft
Prepares external actions for human approval.
L4
Approved execute
Acts in approved channels with readback proof.
L5
High-risk gated
Money, legal, credentials, deploys, calls, and public posts require scoped approval.
07 — Packaging
One real job first.Credits only after trust.
Grid starts with a capped free trial: one real scoped job, no default external sends, and no high-risk execution. Paid tiers scale from Solo Operator to Operations Pod, with credits tied to worker, action, connector, receipt status, and direct COGS. Credits never bypass approval gates.
Free trial: one real jobStarter: 1 workerOperator: 3 workersCommand: 10 workersEnterprise: governed workforce
Worker catalog JSON — the machine-readable version of this page.