Skip to content
AGNT
All playbooks
Onboardingintermediate

Migration Playbook

Migrate a venue from OpenTable, Resy, or manual booking to AGNT — extract, restructure, optimize.

AGNT Onboarding Desk3 steps60-90 minutesClaude Opus 4.6, Claude Sonnet 4.6

Why this playbook

Venue migration is the most sensitive part of onboarding — you're replacing a working system. Downtime means lost bookings, and data loss means angry customers asking 'where's my reservation?'

This playbook extracts existing data from the current platform, restructures it into AGNT's format (reusing the same intake prompt from standard onboarding), and then configures the booking optimizer for the transition period — including dual-running with the old system during cutover.

Designed for venues switching from OpenTable, Resy, or manual booking systems. The extraction step handles each platform's export format.

Prerequisites

  • Claude Code with filesystem access
  • Admin access to the venue's current booking platform
  • Data export from the current platform (reservation history, menu, contact info)
  • AGNT venue admin panel access for the target venue

Input requirements

InputTypeRequiredDescription
Platform exportCSV, JSON, or directoryYesExported data from the current booking platform: reservation history, venue details, menu, customer contacts. Format depends on the source platform.
Source platformstringYes'opentable', 'resy', 'sevenrooms', 'manual', or 'custom'. Determines which extraction rules apply.
Cutover strategystringNo'immediate' (switch in one step) or 'dual-run' (run both systems for 1-2 weeks). Default: dual-run, which is safer but requires more configuration.

Step-by-step workflow

1

Extract and normalize data from current platform

Open prompt

This prompt reads the platform export and normalizes the data into a common format: reservation history, venue metadata, customer contacts, and booking policies. It handles platform-specific quirks (OpenTable's CSV encoding, Resy's nested JSON, SevenRooms' XML exports).

The output is a clean intermediate dataset that the next step can process. The prompt flags data quality issues: missing fields, encoding problems, duplicate reservations, and orphaned customer records.

2

Restructure into AGNT's venue knowledge format

Open prompt

This is the same intake prompt used in standard venue onboarding, but fed with normalized data from step 1 instead of a raw website scrape. It produces the pack.json, validates all fields, and flags anything that needs human review.

The advantage of going through the standard intake process: the venue's data structure is identical to a fresh onboarding, which means all downstream tools (agent config, booking optimizer) work without modification.

Normalized venue data from step 1
3

Configure booking for migration transition

Open prompt

The final step configures the booking optimizer with migration-specific settings: dual-run mode (if chosen), reservation history import for the agent's context window, and customer notification templates for the platform switch.

If using dual-run mode, the optimizer configures automatic sync between both systems during the transition period and sets up alerts for booking conflicts (same slot booked on both platforms).

pack.json from step 2Reservation history from step 1Cutover strategy

Expected outputs

JSON

Normalized platform export

Clean, standardized dataset from the source platform: reservations, venue metadata, contacts, policies.

Produced by step 1
JSON (pack.json)

AGNT venue knowledge pack

Standard AGNT venue format ready for agent configuration — identical to a fresh onboarding.

Produced by step 2
YAML

Migration booking config

Booking optimizer config with dual-run settings, history import, conflict alerts, and customer notification templates.

Produced by step 3

Tool requirements

  • Claude Code with filesystem access
  • Export from current booking platform (CSV, JSON, or API access)
  • AGNT venue admin access

Troubleshooting

OpenTable CSV has garbled characters in venue names
OpenTable exports use Windows-1252 encoding, not UTF-8. The extraction prompt handles this automatically, but if you pre-processed the CSV, re-export from OpenTable with UTF-8 selected (Settings > Export > Character encoding).
Reservation history is too large (50k+ records)
Filter to the last 12 months before importing. The booking optimizer only uses recent history for context — older reservations add noise without improving recommendation quality. The extraction prompt accepts a date-range filter.
Dual-run mode shows booking conflicts
This means the same timeslot was booked on both platforms. The migration config includes a conflict resolution rule: AGNT booking takes priority, old-platform booking gets a courtesy notification. For the first week, set the rule to 'flag-for-review' instead of auto-resolve.
Customers receive duplicate confirmation messages
Disable booking confirmations on the old platform during dual-run. Only AGNT should send confirmations. The migration config generates a platform-specific disable checklist for each source system.

Share as social post

Migration Playbook 3 steps, 60-90 minutes. Migrate a venue from OpenTable, Resy, or manual booking to AGNT — extract, restructure, optimize. https://agntdot.com/playbooks/migration-playbook

191 / 280 chars

Related playbooks

Run the playbook.

Open each prompt in order, feed the outputs forward, and ship the workflow end-to-end.