Description
Summary I run an Australian education product built on three Zapier Chatbots (all using Claude as the AI), each connected to its own Zapier Tables, with data flowing in from Google Calendar, Google Sheets, and Google Drive via Zaps. I need an experienced Zapier specialist for a short, clearly-defined job. There are two parts: fixing some date problems in my calendar sync Zaps (Issues 1 and 2), and investigating two platform questions and giving me clear written answers (Issues 3 and 4). Evidence requirement — applies to everything below: all fixes and investigation findings must come with evidence — screenshots from your testing showing the problem, what you changed, and the corrected result. I make decisions on verified evidence, not verbal confirmation. The fixes (a full technical brief is provided when hired): Fix one calendar sync Zap. When a calendar event is edited, the date saved in my table shifts by one day. The Zap also needs "update the existing row instead of creating a duplicate" logic rebuilt. Afterward, check the affected rows are correct (with screenshot evidence). Fix the calendar sync Zaps for the other two chatbots. These two Zaps are built identically to each other. On the first: repeating events (same name, different dates) are syncing with dates one day early, and all-day calendar events aren't syncing across at all — fix both problems. On the second: audit it for the same problems and fix if found. Evidence of corrected rows for both. I expect these fixes are achievable. If any turns out not to be, I need a written explanation of why — is it something about my setup, or a limit of Zapier itself? The investigations (the most important part — written reports required): AI settings, speed, and model options — applies to all three chatbots. Responses to bigger questions can be slow (some take well over 20 seconds). Zapier Chatbots doesn't show an AI "effort level" setting anywhere in its interface. I need to know: (a) Can the effort level be reached another way (behind the scenes / via the API)? Important: do NOT change the setting, even if you find a way — document exactly how I can change it myself, step by step. If it truly can't be reached, confirm that in writing. (b) Can the chatbots be upgraded to a newer or more capable Claude model than the one currently selectable? If so, what are the options and trade-offs (speed, quality, cost)? Again — recommend, don't change. (c) What can realistically be done within Zapier to improve response speed — especially on long answers such as listing many events? This is investigation only: a written list of options with the expected impact of each. Do not implement any speed changes without my approval. Missing data investigation — the problem exists on all three chatbots. When someone asks any of my chatbots a broad question like "list all upcoming events," they only return some of the rows from the table — e.g. 12 out of 18 on one bot, 11 out of 26 on another. Ask again and you get a different incomplete list. The missing rows are all healthy and correct — nothing is wrong with the data. Because it happens on all three bots (separate tables, separate calendars, separate prompts), it looks like platform behaviour, not a one-off fault. I need written answers to: (a) Does the chatbot actually see the whole table when answering, or only a portion it fetches per question? If only a portion, can that portion be made bigger? (b) The bottom-line question: is a fix or workaround possible, within Zapier, that would have the chatbot accurately read and list 50+ events — every single one, in correct chronological order, every time? Yes or no — and if yes, how, and does it hold up as the table keeps growing? (c) If the answer to (b) is no, at what data volume does reliable performance end? Demonstration requirement: any fix or workaround you propose for (b) must be demonstrated on a test table of 50+ event rows, with the complete correct listing shown on at least three separate repeated queries — screenshots of all three. Compatibility requirement: my tables also hold other row types (homework, assessments, curriculum), and the chatbots use other knowledge sources (FAQ documents, an email log). Any workaround must not break the chatbot's ability to answer questions about those. The written reports — the final milestone of this job. For Issues 3 and 4, the written reports ARE the deliverable, and the engagement closes with their delivery. Each report should be 1–2 pages, in plain English a non-developer can act on, covering: what is happening, why, what you tested (with evidence), what is and isn't possible within Zapier — and, if something can't be fixed within Zapier, what a proper fix outside Zapier would involve (in general terms: what kind of build, what it replaces). These reports decide the direction of my whole build, so they matter as much as the fixes. You'll get a detailed brief, account access, and a responsive client who has already tracked down and documented several related problems. Fast turnaround is valued — I'm working to a decision deadline about two weeks away.