Builds That Work Get the Daily Build
Make Support triage Professional services

Support AI tuned in Make: resolution rate from 34% to 92%

Source-checked Case study

Partner #makeitfuture used Make to test ranking and filtering strategies for a client's OpenAI support assistant without hard-coding each change.

Source: Make, Make case study, 2025

1What it does

Partner #makeitfuture used Make to test ranking and filtering strategies for a client's OpenAI support assistant without hard-coding each change.

2Who it is for

Industry
Professional services
Platform
Make
Category
Support triage
Tools used
MakeOpenAI

3How it works

Every step, in order, is in Workflow at a glance, pinned on the right as you scroll.

Workflow at a glanceMake scenario
  1. 01 Trigger

    Customer support question.

  2. 02 Action

    Filter the data the assistant sees.

  3. 03 Action

    Test ranking strategies for which data to use.

  4. 04 Action

    Ship the best one.

Trigger
Customer support question.
AI step
Unknown.

4How it went

Per Make: 55% better response accuracy, 11 ranking strategies tested in 3 weeks instead of 6 months, resolution rate from 34% to 92%.

Only what the source reports. Numbers above come from the source.

5Worth knowing

  • The win came from what data the AI sees, not the prompt.

6Recipe

Adapted from source Reworded or reconstructed from what the source shows.

Pattern: question -> Make filter/rank step -> top N docs -> OpenAI answer -> log result for comparison.

7Source

Open the original Make case study by Make →

Source opened and read on 2026-10-11.

Related

More Support triage builds

All builds →
Make Support triage

Support emails summarized by ChatGPT and threaded into Slack

Gmail forwards support emails to a Make mail hook. ChatGPT writes a short summary, and Slack gets a threaded message so nothing gets missed.

Agency Written by Builds That Work
Make Support triage

Dealer group SMS triage: AI intent and sentiment auto-closed 18,000 messages

Stellantis &You UK built two-way SMS with ClickSend and Knack in Make, then added OpenAI to classify intent and sentiment. Messages that need no reply get auto-closed. Subtle unhappiness gets flagged for a person.

Automotive dealership Written by Builds That Work
Make Support triage

Make AI Agent that answers FAQ emails and escalates the rest to Slack

A Make AI Agent with an FAQ as context reads new customer emails, replies to routine questions and posts complex ones to Slack for the team.

Ecommerce Written by Builds That Work

Support AI tuned in Make: resolution rate from 34% to 92%: workflow

  1. 01 Trigger

    Customer support question.

  2. 02 Action

    Filter the data the assistant sees.

  3. 03 Action

    Test ranking strategies for which data to use.

  4. 04 Action

    Ship the best one.