Anapoly AI Labs

Levi Swims Lab Brief

Transparency label: human-led (human-authored, with AI input limited to suggestions, edits, or fact-checking)

The Scenario

Levi Swims is a micro-enterprise; it provides swimming lessons to children and young people, with a particular emphasis on those who have special educational needs and disabilities (SEND). Lessons are delivered one-to-one or in small groups and are tailored to individual needs.

The enterprise has begun a progamme to develop a system better able to support its operation. Phase 1 of this system is described in a Concept of Operation. It is being kept deliberately simple. Parents and carers will be able to learn about Levi Swim services via a WordPress.com website, request lessons through a booking form, and receive personal confirmation directly from Levi. Bookings will continue to be recorded in a shared bookings calendar, implemented, as now, in an Excel spreadsheet stored on Google Drive. Invoicing and payments will continue to be handled as a third-party service through Community Support Team (CST), using their own accounting tools.

Phase 1 does not include automated booking, live availability display, integrated payment processing, nor automated reminder functions. These capabilities are recognised as desirable but are planned for later phases.

Purpose of the Lab

To explore the potential usefulness of AI to the enterprise, judging this against what a micro-business can sustain without unreasonable risk.

Opportunities for use of AI

Levi Swims is a classic micro-enterprise built around a single skilled practitioner. The work involves a few high-level operational processes: lesson delivery, booking and calendar management, parent communication, coordination with third-parties, and the handling of SEND/safety information. Phase 1 will rely on manual routines and tools: a static website, standard email capabilities, and a spreadsheet-based bookings calendar.

Strengths of AI in this context

Creating simple procedural workflows that reduce human load.

Capturing structured information from unstructured messages.

Drafting clear, consistent customer communications.

Triage: spotting missing information, contradictions, or conflicts.

Scheduling optimisation and constraint-checking (within limits).

Automating recurring admin: reminders, confirmations, follow-ups.

Summarising or condensing overlong information into actionable points.

Template-driven document or message generation.

Basic data hygiene (spotting duplicates, inconsistencies).

Mapping of AI strengths across high-level processes

A. Booking and Calendar Management

Tasks Levi performs:

  • Receiving booking requests.
  • Extracting key details.
  • Spotting missing information.
  • Checking availability in the spreadsheet.
  • Negotiating alternatives.
  • Confirming bookings.
  • Maintaining the calendar manually.

AI match: very high. It can parse requests, extract details, flag missing fields, draft responses, and detect clashes. It cannot replace the spreadsheet calendar or make safety decisions.

B. Parent Communication

Tasks Levi performs: initial responses and clarifications, explaining SEND considerations, confirmations, reminders, handling cancellations.

AI match: high. It can draft replies, generate explanations, and maintain consistent tone. Levi must always send messages himself.

C. Third-Party Coordination

Tasks Levi performs: communicating invoice details, responding to payment flags, ensuring calendar consistency.

AI match: moderate. It can draft summaries for CST and detect mismatches. It cannot integrate with Xero or make credit decisions.

D. Lesson Planning and Delivery

Tasks Levi performs: planning and delivering lessons, tailoring approach, ensuring safety.

AI match: very low. AI can suggest ideas or summarise notes but cannot replace judgment.

E. SEND and Safety Information Handling

Tasks Levi performs: sifting information, extracting essentials, writing concise notes, deleting unnecessary material.

AI match: high but delicate. AI can summarise long messages and highlight relevant points while Levi retains all safety decisions.

Analysis:

Where could AI Provide Benefit?

  • The biggest operational pain point is booking administration. Many use cases and scenarios revolve around manual admin, potentially creating delays and errors. This problem is well suited to AI.
  • Parent communications are frequent, repetitive, and not safety-critical; AI could support drafting and clarification.
  • SEND/safety note extraction could benefit from AI’s ability to condense overshared information.
  • Third-party coordination could benefit to a small extent.
  • Lesson delivery gains nothing.

AI could assist in booking admin, parent communication, and SEND/safety information triage.

Constrain the AI’s role to that of administrative augmentation, not automation, the goal being a safe, valued, low-risk application.

Approach to Lab Work

Model Levi Swim’s business operation, defining processes and creating a minimal synthetic data environment.

Apply the principles of the three-layer context model to produce the context window components needed to configure a Booking Admin Copilot.

Deploy the Copilot and then run a progression of experiments to:

  • extract structured data,
  • identify missing information,
  • draft clear responses,
  • summarise SEND notes,
  • flag potential scheduling conflicts, and
  • prepare invoice requirements.

Print 🖨 PDF 📄