Skip to content

The agent workforce

A back office that runs itself and shows its work.

Tide is the first pool platform built on AI agents that act: routes generated overnight, receivables chased on a ladder, messages triaged with full account context, reviews solicited and answered, field issues turned into work orders. Every action logged. Every judgment call yours.

Scheduling & Route

  • Auto-generates tomorrow's routes hourly from templates and service plans due for a visit
  • Geometric route optimization by drive time and location
  • On a callout, cancellation, or emergency: a minimal reassignment plan that never touches in-progress work
  • Confidence thresholds decide what runs on its own and what waits for you

Human control · Every action logged with its reasoning

Billing & Collections

  • Recurring invoices generated on your billing day, prepaid or arrears
  • A five-step dunning ladder from a day-3 friendly nudge to a day-28 owner escalation
  • Autopay enrollment nudges after great visits
  • Billing insights: anomalies, risks, and opportunities, surfaced weekly

Human control · Day-28 escalations always require your approval; the Go-Live switch defaults off

Customer Communication

  • Every inbound SMS, email, and portal message classified with full account context
  • Urgent messages escalated to a human immediately
  • Proactive touches: tech-en-route texts, post-visit service reports with one-tap rating links
  • Weather and equipment alerts before the customer has to ask

Human control · Nothing auto-sends to customers. AI drafts, humans send

Review & Reputation

  • Smart targeting: review requests go only to happy, current, paid-up customers after a pristine visit
  • Google reviews pulled hourly with AI sentiment and theme analysis
  • Responses drafted for you; approved replies published back to Google
  • Trends surfaced so you hear a complaint pattern before it costs you customers

Human control · Negative reviews are always human-approved and signed by the owner

Work Orders

  • Field issues become draft, unpriced work orders automatically
  • Notes translated where needed, non-actionable observations filtered out
  • The office gets an email with the draft ready to price and send

Human control · Drafts only, never customer-visible until the office acts

Translation

  • Field notes get translated companion fields, continuously
  • Reports, emails, and work orders always read clean for the office and the customer
  • Original notes preserved untouched

Human control · Fully automatic; nothing is overwritten

Exception-based management

Autonomous never means unsupervised.

The oversight model is a feature, not a disclaimer. You review the 5% that needs judgment; agents handle the 95%, and every step of it is written down.

The Swimlane

A triage queue that buckets everything needing attention into urgent, attention, and review, sorted so the most important judgment call is always on top.

The approval ledger

Every agent action lands in a single ledger with its description, reasoning, confidence, and the records it touched. Approvals happen in-app.

The Pulse timeline

One feed of everything agents and AI did across the platform. The answer to "what did it just do?" is always one glance away.

Off by default

Autonomy ships with the Go-Live switch off. You turn agents on when you trust them.

Thresholds you set

Confidence thresholds decide what runs alone and what waits for you.

Nothing sent for you

Customer-facing messages are drafted by AI and sent by humans.

The business ran while you slept. Read all about it at 6 a.m.

Get early access and put an agent workforce on shift, with you holding every switch that matters.