How Hapex works.

Getting started

Hapex gives your business one always-on AI Operator that learns how everything works, remembers every detail, and runs the work for you. It is your new best employee. No code required. The flow is: tell it about your business, connect your tools, run a live test, and turn it on.

To get started:

1. Open your Operator at build.hapex.ai.
2. Tell it about your business in plain English. The more it knows about how you work, the sharper it gets. Be specific about where your data lives, what you want done, and how often it should run.
3. The Operator recommends a plan based on how much you run and routes each job to the right model automatically. It also shows you the projected monthly cost.
4. Connect your tools via OAuth. You will see exactly which permissions are requested and why.
5. Run a live test. Watch the work happen in real time. Refine what it does if needed.
6. Click Activate. Your Operator starts running on its schedule, around the clock.

Most setups take under an hour. From there it only gets sharper at your business every week. Start at build.hapex.ai.

Connecting your tools

Your Operator sees every account at once. It connects to your tools via standard OAuth flows, and you authorize only the minimum scopes it needs for the work it does. No blanket access.

Currently supported connectors:

• Gmail — Read and/or send access. Read-only when it only triages. Send access only when it produces outbound email. Managed via Dashboard → Connections.
• Google Calendar — Read-only. Used when scheduling context matters.
• Google Drive — Read-only. Used when it processes documents.
• Google Sheets — Read and write. Used for data input and result output.
• Slack — Send messages to channels or DMs. Used for notifications and briefings.
• SMS — Outbound only. Used for alerts and digests.

To manage or revoke connections, go to Dashboard → Connections. You can also revoke Google access at any time from myaccount.google.com/permissions.

Setting up your Operator

The more your Operator knows about your business, the better it runs it. The most important part of setup is telling it clearly what you want done: what to read, what to do with it, and where the result goes.

Vague: Summarize my emails.

Clear: Every morning at 7am, read my unread Gmail, skip newsletters and receipts, classify each remaining message as urgent or routine, draft replies for routine inquiries, and email me a summary listing urgent items first with one line per item.

The more specific you are about timing, source, filtering rules, and output, the sharper your Operator gets. It asks clarifying questions when key details are missing, and it remembers what it learns so you never have to repeat yourself.

Once it understands the job, your Operator will:

1. Recommend a plan based on how much you run, then route each job to the right model automatically (fast models for routine work, genius-tier for hard work).
2. Show the projected monthly cost.
3. Request the OAuth connections it needs.
4. Let you run a live test before turning it on.

Before anything risky, your Operator asks first. Examples of what it can take off your plate are at hapex.ai/use-cases.

Reading run history

Every run your Operator does is logged. You can see when it ran, what it processed, what actions it took, and whether it succeeded or failed. Nothing is forgotten. Logs are stored for 90 days.

To view run history:

1. Go to Dashboard.
2. Select the task you want to inspect.
3. Click the Run History tab.

Each run entry shows: start time, duration, number of items processed, actions taken (emails sent, rows written, messages posted), and the final status. Failed runs include an error message and the last successful step so you know where it stopped.

You can trigger a manual run from the Run History tab for testing without waiting for the next scheduled run.

Billing

Billing is monthly, charged to the card on file via Stripe. Your plan (Hapex Plus, Pro, or Max) sets the run quota for the month, or pay as you go with Hapex Flow credits. One Operator, one subscription. The billing cycle resets on the first of each calendar month.

Key billing facts:

• The projected monthly cost is shown before you turn your Operator on. You will not be surprised by the first invoice.
• If your Operator hits the monthly quota, it pauses automatically. You receive an email notification before the cap is reached.
• You can upgrade your plan at any time from the Dashboard. The cost difference is prorated for the remainder of the billing cycle.
• Cancellation takes effect at the end of the current billing period. You will not be charged for the following month.
• All charges are non-refundable. See the Refund Policy for the carve-out covering Hapex-side failures.
• Live spend visibility. Watch real-time cost per run on Dashboard → Spend: this month, last month, all-time, per-automation breakdown, per-capability breakdown.
• Budget caps. Each automation has a monthly run quota; you get a notification when usage crosses 80%, and a hard stop at 100% (auto-resumes when the period resets).

To view invoices or update your payment method, go to Dashboard → Billing.

FAQ

Can my Operator do more than one job?

Yes. Your Operator runs as many jobs as you want within your plan's monthly run quota, all under one subscription. Each job has its own run history, and every run is routed to the right model automatically.

What happens if a run fails?

The run is logged as failed with an error message and an error kind (auth, rate_limit, timeout, network, config, validation, upstream). Transient errors (rate-limit, timeout, network, upstream 5xx) auto-retry with 1s/5s/30s backoff. Auth and config errors skip retry because waiting won’t fix them, so you’ll get a notification with a one-click link to reconnect. Three consecutive failures auto-pauses the job so a crash-loop doesn’t keep hammering a broken provider. Your Operator never has an off day, and it flags trouble instead of hiding it.

Can I change what my Operator does after it is running?

Yes. Open the job in your Dashboard, edit what you want done, and save. The change takes effect on the next scheduled run. Nothing needs to be turned off to edit it, and your Operator keeps what it has already learned.

Is my data stored?

Run logs are stored for 90 days. OAuth credentials are stored encrypted (AES-256-GCM). The content your Operator processes (email bodies, sheet data) is used only within the run and is not retained beyond the log entry. See the Privacy Policy for full details.

Can I export a job definition?

Yes. From the job detail page, click Export. This downloads a JSON file with the configuration, instructions, connectors, and schedule. The export does not include run logs or OAuth credentials.

Something is broken. Who do I contact?

Email support@hapex.ai. Include your job ID (visible in the Dashboard URL) and a description of the issue. For urgent platform failures, check hapex.ai/status first.