AI Customer Service RFP Template (Free Generator)
Generate a vendor-neutral RFP for an AI customer service agent: scope, must-haves, 28 questions and a weighted scorecard. Markdown, free.
- Free
- No signup
- Runs in your browser
# Request for proposal: AI customer service agent **Issued by:** [Company] **Responses due:** [date] ## 1. Scope [Company] is selecting an AI agent to answer customer conversations before they reach our team. - **Volume:** [conversations per month, by channel] - **Helpdesk:** [helpdesk] ## 2. Must-have requirements Answer each with Yes, Partial or No, and one sentence on how. 1. Answers only from content we approve, and says so when it doesn't know 2. Hands a conversation to a person with the full transcript 3. Opens tickets in our helpdesk 4. A hard monthly cap on spend 5. Role-based access and an audit log ## 3. Questions ### Answers and grounding 1. Where does the agent get its answers, and how do we control what it may say? 2. What does it do when the answer isn't in our content? Show a transcript. 3. How do we see which source an answer came from? 4. How quickly does a change to our content reach the agent? 5. How do we mark topics it must always hand to a person? ### Handover to people 1. What triggers a handover, and which of those triggers do we configure? 2. What reaches our team with it (summary, transcript, customer details), and how does it arrive in [helpdesk]? 3. What happens outside our business hours, on each channel? 4. How does a conversation return to the agent once our team is done? ### Channels 1. Which channels does one agent answer today: website chat, phone, email, others? Which are generally available, and which are beta or early access? 2. For phone: do we keep our number (forwarding, porting or SIP), can calls transfer live to a person, and is an AI disclosure spoken? 3. Is the knowledge and configuration shared across channels, or set up per channel? ### Pricing 1. Define your billable unit exactly. If it is a resolution or outcome, quote your contract's definition. 2. Are conversations the customer abandons, or that end in a ticket, billed? 3. Price one sample month for us: [conversations per month, by channel]. Show the arithmetic. 4. What are the seat fees, platform fees, minimums and overage rates? 5. Can we set a hard monthly spend cap, and what happens when it is reached? ### Security and data 1. Where is our data stored and processed, and which sub-processors see it? 2. Is our data used to train any model? Cite the clause. 3. How long are transcripts and recordings kept, and can we delete them? 4. Which certifications do you hold today (SOC 2, ISO 27001) and which are planned? Will you sign our DPA? 5. How do our staff sign in (SSO, MFA), and what roles and audit logging exist? ### Integrations 1. How does the agent connect to [helpdesk], and what does it create there? 2. Can the agent call our own APIs mid-conversation (order lookup, account status)? How are credentials stored? 3. Is there an API or webhooks for our own reporting? ### Rollout and support 1. What does a pilot look like: length, cost, success criteria, exit terms? 2. How long from contract to the first live channel, and what do you need from us? 3. Who supports us after go-live, and on what response times? ## 4. Evaluation | Criterion | Weight | Vendor A | Vendor B | Vendor C | |---|---:|---:|---:|---:| | Answer quality and grounding | 25 | | | | | Handover to people | 15 | | | | | Channels in scope | 15 | | | | | Pricing and predictability | 15 | | | | | Security and data handling | 15 | | | | | Integrations | 10 | | | | | Rollout and support | 5 | | | | | **Total** | **100** | | | | Score each criterion 1–5 and multiply by its weight. A pilot on our own content outranks any answer in writing. ## 5. How to respond Answer in this document's order and send it by [date], with a sample-month price and one transcript showing a handover.
This ran in your browser on what you typed. The product runs on your actual calls, chats and emails.
Run the pilot on your own content — start freeAbout this tool
Evaluating AI agents for customer service usually starts with a demo, and demos are where every vendor looks the same. The differences show up in writing: what the agent does when it doesn't know, what exactly is billed, where your data goes, how a conversation reaches your team. An RFP that asks those questions the same way of every vendor is the fastest way to see them side by side.
Fill in your company, your monthly volume by channel, your helpdesk and any must-haves. The template asks every vendor to price that exact month and show the arithmetic, quote their definition of a resolution, and show a handover transcript, alongside 28 questions on answers, handover, channels, pricing, security, integrations and rollout, and a weighted scorecard. It's vendor-neutral, WireDesk included. It's built in your browser and nothing is sent anywhere.
What the template asks about pricing
1. Define your billable unit exactly. If it is a resolution or outcome, quote your contract's definition. 2. Are conversations the customer abandons, or that end in a ticket, billed? 3. Price one sample month for us: 12,000 conversations a month: 60% chat, 25% email, 15% phone. Show the arithmetic. 4. What are the seat fees, platform fees, minimums and overage rates? 5. Can we set a hard monthly spend cap, and what happens when it is reached?
What each input means
- 01
Company and due date
Printed at the top and in the response instructions.
- 02
Monthly volume, by channel
The single most useful line in the RFP: every vendor prices this exact month, so their answers are comparable instead of each picking a flattering example.
- 03
Helpdesk
Named in the handover and integration questions, so vendors answer for your stack rather than in general.
- 04
Must-haves
One per line; each becomes a Yes, Partial or No question. Leave it blank for the five most teams add: grounded answers, handover with the transcript, helpdesk tickets, a spend cap, and roles with an audit log.
Questions about this tool
Your scope and volume, a short list of must-haves, and questions vendors have to answer with evidence: a transcript of the agent not knowing, a definition of the billable unit, a sample month priced, the sub-processors that see your data, and how a conversation reaches your team. Then a weighted scorecard, so the decision isn't made on the best demo.