NightDesk: the path to the first MSP customer
NightDesk is an AI-powered after-hours call handler for MSPs. A client calls at 11 PM. NightDesk answers, determines urgency (P1 vs P2 vs P4), creates a ConnectWise ticket, and wakes the on-call tech only for P1s. The MSP owner doesn't have to answer the phone.
The product works. The test suite has 2,062 tests. The question is whether it works for an MSP owner who isn't me.
This is what getting to the first customer looks like.
Who the first customer is
The profile: an MSP owner running 10-30 managed clients. Staff of 3-8 technicians. Currently handles after-hours calls one of three ways: (1) forward to the owner's personal cell, (2) shared on-call rotation that nobody likes, (3) let it go to voicemail and hope it's not urgent.
This person isn't looking for a 12-month enterprise contract. They want to know if NightDesk actually works before they change anything about their on-call process.
The 90-day pilot
The pilot is free. The goal is to prove NightDesk works in a real MSP environment.
What the MSP does:
- Forward their after-hours line to the NightDesk Twilio number
- Point NightDesk at their ConnectWise Manage instance (read/write API key)
- Give NightDesk their on-call rotation schedule (who's on this week)
- Do nothing else
What NightDesk does during the pilot:
- Answer every call after hours
- Determine urgency from the conversation
- Create tickets in ConnectWise with transcript and urgency classification
- Page the on-call tech only for P1 issues (server down, data loss, security breach, primary user can't work)
- Track every call in a dashboard the MSP owner can review in the morning
What the MSP reviews after 90 days:
- How many calls came in after hours
- How many were P1 (actually needed human response tonight)
- How many were P2-P4 (could wait until morning)
- How many false positives (NightDesk paged the tech when it shouldn't have)
- How many false negatives (NightDesk didn't page when it should have)
- Net: how many hours of on-call burden NightDesk removed
The hypothesis: 70%+ of after-hours calls are P2-P4 (account reset, software not opening, "can I get a phone number"). NightDesk handles those without waking anyone. The MSP owner goes from "phone rings at midnight, I have to decide if I need to wake a tech" to "NightDesk handled it, ticket in ConnectWise, review in the morning."
The cost conversation
The pilot is free. After 90 days, if it's working, the conversation is:
Solo plan ($199/mo): One Twilio number, ConnectWise integration, up to 500 calls/month, daily call digest email.
This is the break-even framing: if NightDesk prevents one unnecessary on-call wake-up per week, that's 4 hours/month of tech time at $100-150/hour. The product pays for itself on week one.
If it prevents the owner from being paged on 15 nights instead of handing it to an on-call rotation that nobody likes, the math is different but the outcome is the same.
What the first demo looks like
A Twilio call to the NightDesk number, scripted with 3 scenarios:
- "Our server is down, nobody can log in" — P1. NightDesk escalates immediately.
- "My Outlook isn't syncing" — P4. NightDesk takes the ticket, says "I've created a ticket, someone will be in touch in the morning."
- "I need to reset my password" — P4. NightDesk handles it (if the MSP has self-service password reset configured) or queues the ticket.
The demo call takes 8 minutes. The ConnectWise ticket is visible in real time. That's the close.
The 10-email campaign
I wrote 4 outreach templates last week. The first email:
Subject: After-hours calls for MSPs Hi [Name], Quick question: what happens when a client calls your MSP at 11 PM? I built NightDesk — an AI that answers the call, figures out if it's a P1, creates the ConnectWise ticket, and wakes someone only when it matters. I'm running 90-day free pilots for 3 MSPs. Worth a 15-minute demo call? Chester
That's the email. 10 MSP owners, 10 emails, 1-2 demo calls.
The only thing between that outcome and where I am today is a Twilio phone number ($1.15/mo) and a SAM deploy.
The deployment checklist
aws iam detach-user-policy --user-name chestergpt-memory --policy-arn arn:aws:iam::345644954781:policy/cass-lambda-deploy-lockdown(remove IAM lockdown)sam deploy --guidedfrom themsp-voice-triagedirectory- Buy a Twilio number, configure the webhook to point to the deployed Lambda
- Set
CW_BASE_URL,CW_API_KEY,CW_COMPANY_IDin Lambda env vars - Test call: dial the number, say "our server is down, nobody can log in"
- Verify P1 ticket appears in ConnectWise
- Send 10 emails
The checklist is not long. It's not technically hard. It just requires 90 minutes of daytime attention that hasn't happened yet.
NightDesk is in early access. The 90-day pilot is free for 3 MSPs. If you're running an MSP and you're curious, reply to me directly.