AI support agent proposal with a ticket mix chart, hand-over rules, pass marks before launch and three one-time options.
Start free. No card required.
A starting point for the scope, pricing and next steps you want to share
Keep useful pages and replace the content with your client details
Choose the colors, fonts and imagery that fit your business
An AI customer support agent proposal shows which tickets an agent will answer, which it hands to people and how answers are tested before launch. This template is written for a software company whose team answers the same how-to questions all day, and offers three one-time options by channel.
AI automation agencies and consultants building a first support agent for software or ecommerce companies on Zendesk or a similar help desk. It suits teams where a large share of tickets are repeat questions with documented answers.
Start with ticket volumes by type from the client’s help desk, then say plainly which types the agent answers, which it hands over and when. Show how answers are tested on past tickets with pass marks before customers see them.
Offer options by channel and depth, such as chat only, chat and email, and chat and email with account lookups. Split payment between acceptance and go-live, and offer tuning after launch and extra languages as add-ons.
Time to customise: about 20–30 minutes.
11 parts, in the order your client reads them. Every part is editable: keep what fits the job, rewrite the rest and add your own.
Your 14 support agents handle about 7,400 tickets a month in Zendesk. Two in five are how-to questions your help centre already answers, and chat is closed at weekends, so Monday starts with about 600 waiting emails.
From three months of Zendesk data and a day with your team.
By the end of month three we expect it to resolve about 30% of chats and 20% of emails, or roughly 1,900 tickets a month.
Everything in the recommended option, Chat and email.
Your support lead signs off each pass mark before launch.
We review your articles and macros with your support lead and agree the hand-over rules.
A payroll software company with a support team of 11.
Covers: The client, The result
Ran support teams for eight years before moving into AI agents, and has launched 11 of them for software and ecommerce companies since 2023.
Three options. We recommend Chat and email.
The customer can ask for a person at any time, and each answer links the article it used. Your team flags wrong answers in Zendesk, and we fix the article or the rule within two working days during launch.
Pick an option and any extras, then accept on this page.
Repeat questions with a documented answer, such as how-to and account questions. Complaints, refunds and anything needing judgement should go to a person.
Run it on a set of past tickets and score its answers against agreed pass marks, then launch on one channel and review real conversations before widening it.
No. It takes the repeat questions so people can handle the complex ones, and the hand-over to a person must be quick and keep the conversation history.
Access to past tickets for testing, the help articles it answers from and a way to hand conversations to the team. Account lookups need read access to the right system.
11 editable parts, in this order: Introduction, Your situation, Our recommendation, What is included, Testing before launch, Timeline, Results for a similar client, Your team, Options and fees, Questions and Next steps. Keep the parts that fit the job, rewrite the rest and add your own.
As three one-time options the client can choose between: Chat only ($14,000), Chat and email ($21,500, recommended) and Chat, email and account lookups ($32,000). Payment terms: 50% on acceptance, 50% when the agent goes live. The prices are examples; replace them with your own before you send it.
Plan on about 20–30 minutes to replace the sample client, scope, dates and prices with your own, then preview it as your client will see it.
Share it as one link your client can read on any device, comment on, accept and e-sign, or export a PDF for your records.
Add your scope, pricing and brand. Preview the client experience before sharing.