Keeping a copy in sync

Load the full dataset once. After that, each request returns only what was added, changed or removed since the last one, so you never need to delete and reload everything.

The three steps

1. Load everything once. Download GET /v1/exports/conferences (gzipped JSON Lines, one record per line) and insert every record into your table, keyed by id. Save the response's X-Sync-Token header.
2. Ask for what changed. As often as you like (daily, hourly), call GET /v1/conferences/changes?since=<saved token>.
3. Apply each entry, then save the new token. An entry with "deleted": true means remove that id; anything else is the full current record, so replace your row (insert it if it's new). Then save next_sync_token. If has_more is true, call again straight away with it.

The same works for journals. Keep one token per dataset.

Example (Python)

import gzip, json, os, requests

API = "https://api.orbytelgroup.com/v1"
HEADERS = {"Authorization": "Bearer " + os.environ["ORBYTEL_KEY"]}

def initial_load(db):
    r = requests.get(f"{API}/exports/conferences", headers=HEADERS, timeout=300)
    r.raise_for_status()
    for line in gzip.decompress(r.content).splitlines():
        db.upsert(json.loads(line))              # keyed by record["id"]
    db.save_token(r.headers["X-Sync-Token"])

def sync(db):
    token = db.load_token()
    while True:
        r = requests.get(f"{API}/conferences/changes",
                         params={"since": token}, headers=HEADERS, timeout=60)
        if r.status_code == 410:                 # license changed: start again
            return initial_load(db)
        r.raise_for_status()
        page = r.json()
        for record in page["data"]:
            if record.get("deleted"):
                db.delete(record["id"])          # fine if you never had it
            else:
                db.upsert(record)
        token = page["next_sync_token"]
        db.save_token(token)                     # after applying the page
        if not page["has_more"]:
            break

Good to know

• Changes appear about a minute after they're saved. The feed only hands out positions that are safely settled, so nothing is skipped.
• A record can arrive twice, for example in two consecutive syncs. Because every entry is "replace your copy with this", applying it again is harmless.
• A removal can be for a record you never had (for example one outside your license). Ignore it.
• Several changes to one record arrive as one entry, in its latest state.
• Tag and organization changes count. If a tag is added to a conference, or an organization is renamed, the affected records come back in the feed.
• 410 Gone means your license changed (for example a therapy area was added). Your copy may now hold records you're no longer licensed for, or lack new ones, so reload from /exports and continue from its token.
• No token yet? GET /v1/conferences/changes with no parameters returns a token for "now".
• Prefer dates? ?updated_since=2026-10-19 works instead of since, removals included. A token is still more precise, because it can't miss or repeat a boundary.

Sizes and limits

• The full conference export is about 9,000 records (a few MB compressed); journals about 3,000.
• /changes returns up to 500 records per call by default, 1,000 with limit=1000.
• A daily sync typically needs one or two requests.