Customer Support and Voice AI
AI receptionist for small business: calls, bookings and handoffs
What an AI receptionist does on a real call, the rules that apply, a sample greeting and a pass/fail checklist to launch one safely.
Support Operations
AI Support Knowledge Agent for Intercom
Closed Intercom conversations reveal the articles your help center lacks, and drafts arrive in Intercom ready for an editor.
Closed Intercom conversations show exactly what customers struggle with, but nobody has time to read them all. This agent groups closed conversations by topic using your tags, checks each topic against your Intercom help center, and writes a draft in your brand voice wherever no article exists. Drafts are saved unpublished in Intercom, ready for an editor to polish and publish.
- Closed conversations grouped into topics using your tags
- Missing help center topics ranked by how often they come up
- Drafts saved unpublished in Intercom for an editor
$4,400
20–30 working days
Optional maintenance $249/mo
What it is
AI Support Knowledge Agent for Intercom reviews closed Intercom conversations, groups them by topic using your tags, and checks each topic against your Intercom help center. Where no article answers it, a draft is written in your brand voice and saved in Intercom unpublished.
Why businesses buy it
In Intercom, a thin help center means more conversations for your team and slower answers for customers, especially out of hours. Writing articles is always the job that slips. After launch, you get a ranked view of the topics your help center lacks and drafts waiting in Intercom for review. The Scope Card sets the tags, destination collection and review step before you order, so you know exactly what gets drafted and that nothing publishes without approval.
Problems this solves
Closed Intercom conversations hold answers your help center never captured
The agent reviews closed Intercom conversations and groups them by topic using your tags. Each topic is checked against your Intercom help center to find what is missing.
Your Intercom bot keeps handing over on the same questions
Where no article answers a topic, a draft is written in your brand voice. Once approved, those articles give customers and your help center more answers to draw on.
Help content written by different people sounds inconsistent
Drafts follow your brand profile and up to 20 reference examples. They are saved in Intercom unpublished, so an editor checks tone and accuracy before release.
A good fit if
Your team answers customers in the Intercom inbox, and the same explanations keep getting typed out in conversation after conversation.
You tag Intercom conversations by topic or product area, even loosely, so there is a signal about what each one covered.
Your help center lives in Intercom and also feeds answers elsewhere, so a missing article costs you twice.
An Intercom admin can create an app access token and join a 30-minute call to agree the field mapping.
Not the right fit if
Zendesk is your help desk. See AI Support Knowledge Agent for Zendesk.
Freshdesk is your help desk. See AI Support Knowledge Agent for Freshdesk.
You want reply suggestions for agents inside live Intercom conversations. See AI Support Copilot for Intercom.
You want new customers guided through setup inside Intercom, not help center content. See AI Customer Onboarding Agent for Intercom.
How it works
- An Intercom admin creates an app access token, and on a 30-minute call we agree which conversation data, tags and help center collections are mapped.
- Closed conversations from the period you choose are read and clustered by topic, starting from your Intercom tags and refined by what customers actually wrote.
- Each topic is matched against your published Intercom articles; topics with no matching article, or one that customers still write in about, become candidates.
- Candidates are ranked by how many conversations they cover, so the first drafts address the questions your team answers most.
- Drafts are written in your brand voice from up to twenty reference examples and saved unpublished in the Intercom collection you picked.
- Your editor reviews, edits and publishes in Intercom; twenty-five sample drafts are agreed with you before the pipeline runs unattended.
What you provide within 72 hours
Your portal lists these the moment you order. Delivery starts the working day after everything is in and your access is checked.
- An Intercom app access token, shared securely in the way we agree at intake, never by email.
- Your Intercom admin for a 30-minute call on tags, collections and conversation data.
- Tone guidelines and ten to twenty Intercom articles that show how you like content written.
- The collection where drafts should land, and the editor who approves them.
- Tags or topics to exclude, such as billing disputes or account security.
- One decision-maker who signs off the sample drafts.
- An API key for your own AI model account (for example OpenAI or Google Gemini), created for this build with a spending limit.
The 72-hour rule. Send everything on your checklist, including access, within 72 hours of ordering so your build starts on time.
Timeline
Start
Intake verified, Intercom token tested, mapping call held.
Weeks 1–2
Conversation clustering and article matching built from your tags.
Week 3
Ranking, drafting and unpublished saving to Intercom connected.
Weeks 3–4
Twenty-five sample drafts reviewed together.
Week 4
Your feedback applied over two rounds of changes.
Go-live
Live on your schedule, handoff recorded, support starts.
| Option | Delivery time | Price |
|---|---|---|
| Standard | 20–30 working days | $4,400 |
| Express | 15–25 working days | $5,150 |
| Rush | 10–20 working days | $5,950 |
What you receive at handover
- The knowledge agent connected to your Intercom workspace, saving drafts unpublished
- Two-way Intercom connection with documented field mapping, plus up to one more system
- Topic grouping based on your existing tags, with up to three output formats
- Test report with 25 sample drafts reviewed and approved by you
- Admin guide, recorded 45-minute handoff session and 30 days of post-launch support
Expected benefits and how to measure them
- Your Intercom help center can grow from real customer questions.
- Articles can share one consistent voice.
- Your team can spend less time repeating answers in conversations.
- Articles published from drafts
- Count approved drafts published each month in your Intercom help center.
- Conversations on covered topics
- Use Intercom reports to compare conversation volume by tag for four weeks before and after each article goes live.
- Article views
- Track views of the new articles in Intercom's article reports each month.
Before you start: Record conversation volume by tag and your current article count in Intercom for the month before setup.
These are things to track, not promised results. Outcomes depend on your data, your team and your market.
Running costs
The build is paid once. Afterwards, AI model usage runs roughly $20 to $150 a month in your own accounts, rising with the number of conversations read and drafts produced. Your Intercom subscription is unchanged.
Usage during sample reviews is billed to you too, keeping the access token and every provider relationship with your team.
Ask about running costsCustom requirements
Drafting for more than one Intercom workspace, writing in a second language or mapping custom Intercom data beyond the agreement needs a written Change Quote. Add a checkout note and we send it to you.
Request a custom versionKeep it running
Monthly maintenance covers monitoring, fixes and change hours for work such as adding new tags to the clustering or a new target collection. It starts after the post-launch support period, bills monthly and can be cancelled any month.
$249/mo (the minimum fee, includes 2 change hours)
Optional. Starts after the post-launch support period. Cancel any month.
How maintenance worksHow this compares
| Do it yourself with a SaaS tool | Hire a custom agency | Upmark Labs | |
|---|---|---|---|
| Price | An AI content feature is usually priced per seat or per resolution | A technical writer bills by the article | Ours is one fixed price, shown on this page before you buy |
| Topic signal | Often keyword search only | A manual read of conversations | Ours clusters closed Intercom conversations starting from your tags |
| Priority | Rarely ranked | Decided by the writer | Ours ranks gaps by how many conversations they cover |
| Delivery | On today, tuned later | Per batch | Ours has a fixed delivery time shown before you buy, with faster options |
Questions about this product
How does the Intercom knowledge agent use our conversation tags?
As a starting point for grouping. Closed conversations are first sorted by your Intercom tags, then refined by what customers actually asked, so an untagged or mis-tagged conversation still lands in the right topic. You can also tell us which tags to ignore entirely during the 30-minute mapping call.
Are drafts saved directly into our Intercom help center?
Yes, unpublished. Each draft is saved into the Intercom collection you chose and stays invisible to customers until an editor publishes it. The agent does not edit or unpublish existing articles, so your live help center only changes when someone on your team decides it should.
Which gaps does the Intercom edition draft first?
The ones covering the most conversations. Topics are ranked by how many closed conversations asked about them in the period reviewed, so early drafts address the questions taking up the most inbox time. You can pin a topic higher if a launch or policy change makes it urgent.
What Intercom access does the knowledge agent require?
An Intercom app access token, created by an admin in your workspace and shared securely in the way agreed at intake, never by email. The connection reads conversations, contacts and tags and writes draft articles, only within the documented field mapping, and the token can be revoked whenever you choose.
What does the Intercom knowledge edition cost with express or rush delivery?
One fixed price covers the whole build and is shown on this page before you buy, with express and rush delivery listed beside it at a set extra cost. Every option includes the Intercom mapping, twenty-five reviewed samples, two rounds of changes and a recorded forty-five-minute handoff.
Do customer names from Intercom conversations end up in articles?
No. Drafts are written about the question and its answer, not about the people who asked. Contact data is read only to group conversations correctly, and the twenty-five sample drafts you review before launch are checked specifically for any personal detail carrying through.
Guides on this topic
All guides Write the Intercom articles your inbox keeps asking for.
Order the Intercom edition and closed conversations are clustered, ranked and turned into unpublished help center drafts in Intercom.
Works well with
Employee Lifecycle
AI Helpdesk Agent
Resolves employee IT and internal service requests and routes the rest to the right team.
Support Operations
AI Support Copilot
Drafts replies and surfaces relevant articles for human agents inside the helpdesk.
Support Operations
AI Support QA
Scores conversations against your quality rubric and highlights coaching points.