An AI Receptionist Only as Good as Its Last Manual Update

Kitchen Made New (KMN) is a kitchen fabrication and restoration company serving residential and commercial clients. As part of their customer experience strategy, KMN integrated an AI receptionist into GoHighLevel — giving customers real-time order tracking, service updates, and FAQ answers through both voice and web chat.

The AI was central to how KMN handled inbound calls — particularly from manufacturing clients whose end-customers called multiple times per day asking where their orders stood. The system worked. But only as long as someone kept manually feeding it accurate data.

Manual Updates Were Becoming a Bottleneck

Every time an order changed status, a staff member had to stop what they were doing, open GoHighLevel's knowledge base, find the right entry, type the updated information, and save it. With a growing volume of active orders across both residential and manufacturing clients, this manual step was becoming unsustainable — and the AI was starting to give customers outdated answers.

• ✕One manual update per order status change. Every time an order moved through production, a staff member had to interrupt their workflow to update the AI separately.
• ✕Outdated information reaching customers. Missed or delayed updates caused the AI to give callers incorrect order status — eroding trust on high-volume inbound days.
• ✕Unsustainable at scale. As order volume grew, the number of manual knowledge base updates per day grew with it — with no ceiling in sight.
• ✕Voice AI and web chat were not synced. Updating one channel did not update the other — customers could get different answers depending on how they contacted KMN.
• ✕No confirmation of accuracy. Staff had no way to verify what the AI was currently saying without logging into GHL and manually checking — adding even more overhead.

Google Sheet → One Button → Both AI Channels Updated in Seconds

We designed a centralized Google Sheet connected to a custom Google Apps Script. Staff input order changes in one structured table — the same place they were already working. With a single button click, the script reads the Sheet, formats the data, routes it through Make.com, and pushes it directly to GoHighLevel's AI knowledge base via the GHL API. Both the voice AI and the web chat widget update simultaneously from the same source.

• 1Structured Google Sheet as single source of truth. All order status and FAQ data lives in one table — the same one staff were already maintaining.
• 2'Update AI' button inside the Sheet. A custom Apps Script menu button triggers the full sync — no technical knowledge required, no GHL login needed.
• 3Google Apps Script reads and formats the data. The script parses the Sheet, structures it for GHL's API format, and fires a webhook to Make.com.
• 4Make.com routes to the correct knowledge base categories. Parallel PUT requests update both Order Status and FAQ sections simultaneously.
• 5Both AI channels sync from the same push. Voice AI and web chat widget both read from the same GHL custom values — one update, both current.
• 6Status confirmation after every update. Staff see a clear success or error confirmation without needing to verify inside GHL.

Three Working Sessions to Production

Before vs. After