What this tool is
This calculator estimates what a jail or prison would pay each year to run video, audio, and phone services for the people held there and the families staying in touch with them.
It compares five options on a level playing field:
- Keeping Families Connected (KFC) — the platform being built by Washtenaw County Sheriff's Office. It's assembled primarily from Cloudflare's communication, storage, and AI building blocks, with Plivo handling calls to outside landlines.
- Four commercial vendors that sell the building blocks for the same kind of platform — Amazon Chime SDK, Vonage, Twilio, and Zoom — costed out as if a facility built their own platform on top of each.
Below the vendor comparison, a "Status quo" section shows what families currently pay under a typical incumbent contract. Those dollars come out of household budgets; under a facility-paid model like KFC, they go to zero.
Before you start
A few caveats
- These are list prices. Vendors often offer non-profit, government, civic-tech, or volume discounts that aren't shown.
- This is a cost model, not a full budget. It doesn't include the engineering work to build the platform, tablet hardware, ongoing maintenance, or staff time.
- "Steady state." Storage costs are shown after the retention window has filled up. Year 1 is cheaper because nothing has accumulated yet.
- Numbers update live. Every input on the left recomputes the whole screen as you change it.
Quick start — how to read the screen
- Set your facility size on the left — average daily population for main housing and for ITR.
- Read the vendor comparison table in the middle. The "Per year" column is what the facility pays. The "vs KFC / yr" column shows how much more or less each vendor costs compared to the WCSO build.
- Read the "Status quo" section below it. That's what your community currently pays under a per-minute family-pay contract for the same call volume.
- Toggle services off on the left to see what each vendor charges when you strip out things like transcription or long-term storage.
Glossary — terms that appear in the calculator
Technical acronyms
These show up in vendor names, product lines, and footnotes throughout the calculator.
- API (Application Programming Interface) The way one piece of software talks to another. When a vendor "exposes an API" for video, it means developers can plug their app into that vendor's video service. Twilio Video API and Vonage Video API are examples.
- SDK (Software Development Kit) A packaged set of code that developers use to build apps on top of a vendor's platform. Chime SDK is Amazon's set of code for building video apps using Amazon's infrastructure. Zoom Video SDK is the same idea from Zoom. The practical difference: an API is the door, an SDK is the door plus a furnished room.
- PSTN (Public Switched Telephone Network) The traditional phone network that connects regular landlines and cellphones. When the platform "calls a PSTN number," it's dialing a normal phone number.
- VoIP (Voice over IP) Voice calls that travel over the internet instead of the regular phone network. Modern audio calls inside an app are usually VoIP.
- WebRTC (Web Real-Time Communications) The underlying technology that lets live video and audio work directly inside a web browser, without installing a separate app.
- HD (High Definition) Higher-resolution video. Vendors typically charge more for HD than for standard definition.
- kbps (kilobits per second) A measure of audio quality. Higher kbps means better-sounding audio and slightly larger recording files. The default for in-app audio (32 kbps) is "voice-quality" Opus.
- GB / MB (gigabyte / megabyte) File sizes. 1 GB = 1,024 MB. Recording storage gets billed by how many GB are stored each month.
Kinds of calls
- Video call Face-to-face video between a person in the facility and a family member, through an app or web browser.
- Audio call (in-app) Voice-only call running over the internet inside the same app. No phone line involved, no per-minute charge to the family.
- Phone network call A call placed to a regular phone number (cell or landline) over the public phone network. The technical name is PSTN. KFC uses these for ITR-population calls to outside landlines.
- ITR (Intake, Transfer, and Release) A designated unit in the facility where people are processed in, moved between facilities, or released. This is the population that commonly makes calls to landline phone numbers to let loved ones know where they are or to arrange a ride home.
What the platform actually does
- Connectivity Placing and connecting the call. The core service.
- Call export (recording) Saving a copy of the call to a file after it ends.
- Storage Keeping that file around for some retention period (1, 2, or 3 years in this calculator).
- Transcription Automatic speech-to-text. Turns a recorded call into searchable written words.
- Translation Automatic conversion of a non-English transcript into English (or vice versa).
The pieces of the KFC build
- Cloudflare RealtimeKit (RTK) The live video and audio engine.
- Cloudflare R2 Long-term storage for recordings.
- Cloudflare Workers + D1 The back-end web service and database that run the platform.
- Whisper Cloudflare's speech-to-text model. Does the transcription.
- m2m100 Cloudflare's translation model.
- Plivo The phone-network provider that connects ITR calls to outside landlines.
How bills get calculated
- Per minute Most services charge by how many minutes were spent talking, recording, or transcribing.
- Per participant-minute Some services bill once for each person in the call, so a 10-minute call between two people is 20 participant-minutes.
- Per GB-month File storage is usually billed by how many gigabytes are stored, for each month they remain stored.
- Tiered pricing Some vendors charge a higher rate at low volume and a lower rate at high volume. The calculator does the math and shows the effective blended rate inline.
The vendor comparison table, column by column
- Vendor The platform we're costing out. Each name is a link that jumps to its source citations at the bottom of the page.
- Per video call The full cost of a single video call at your settings, including its share of any recording, storage, and transcription that's currently toggled on. The smaller number in parentheses is the per-minute equivalent.
- Per audio call Same idea, for in-app audio calls.
- Per phone network call Same idea, for PSTN calls.
- Per month / Per year Total facility bill at the volume you've set.
- % inv. & storage The share of this vendor's bill being spent on investigative and storage services — transcription, translation, recording export, and long-term storage. The flip side: 100 minus this number is the share being spent on actually connecting people.
- vs KFC / yr How much more (positive) or less (negative) per year that vendor would cost compared to the WCSO build.
Why the percentage matters. Investigative and storage services exist to monitor, review, or preserve calls — not to make them happen. A vendor that's 95% investigative is one where almost every dollar goes to surveillance and review infrastructure rather than communication. Use the toggles on the left to see how the percentage moves when you remove specific categories.
The "Include in costs" toggles
Four toggles on the left let you strip categories of service out of every vendor's price simultaneously. This is how you compare apples-to-apples when you're not planning to use everything a vendor offers.
- Transcription off — calls are still recorded, but no written transcripts.
- Translation off — transcripts exist, but non-English transcripts aren't translated.
- Call export off — calls happen, but no recordings are saved at all.
- Call storage off — recordings are produced, but not kept.
- All investigative & storage services (master) — toggles all four at once. Use this to see the pure cost of just connecting people.
The breakdown rows for each vendor and the percentage column both update live as you toggle.
The "Status quo" section — what families currently pay
Most jails contract with a single telecom vendor that charges families directly, per minute. Rates run roughly 10–25¢ per minute for video and 5–25¢ for audio. That money comes out of household budgets — disproportionately from low-income families.
This calculator uses Washtenaw County's actual rate schedule from a 2025-03-05 account-rep email: 12¢/min video, 7¢/min audio, first 30 video-minutes per person per month comped.
The "Annual cost shifted off our community" figure is the dollar amount the community stops paying when the facility takes the bill onto its own budget. Under KFC at the same call volume, the facility pays a small fraction of what the community used to pay — and the community pays nothing.
If you're using this calculator outside of Washtenaw County, the "Show Securus comparison" toggle in the Assumptions panel hides this section so the screen shows only the facility-side vendor comparison.
Why the prices differ so much
A few structural reasons account for most of the gap between KFC and the commercial vendors:
- Cloudflare R2 has no egress fees. Most cloud storage charges separately every time a file is downloaded for review. R2 doesn't. For a platform that re-plays recordings often, this can be a 5–10× difference compared to AWS.
- Cloudflare prices its AI models like raw compute, not like a managed product. Whisper at $0.00051/min vs Amazon Transcribe Standard Streaming at $0.075/min is about a 147× difference for the same speech-to-text job.
- RealtimeKit is priced to take share from incumbents. Cloudflare entered the live-video market recently and prices it aggressively against Twilio, Vonage, and Zoom.
- One platform vendor instead of stacked vendors. Twilio runs on AWS, so you pay Twilio's markup on top of AWS's markup. Cloudflare runs on its own network, so there's only one layer of margin.
- Plivo for ITR calls. Plivo is roughly 20% cheaper than Twilio for outbound calling to regular phones.
Methodology and sources
Every rate in the calculator is pulled from each vendor's public pricing page. The "Sources" card at the bottom of the calculator links to each one, organized by vendor, with the rate it backs.
The default audio-call usage (44.8 minutes per person per day) comes from Worth Rises' May 2026 report Critical Connections: The Power of Free Communication in Prisons and Jails, which surveyed facilities offering free communication and measured what people actually use when calls don't cost them anything.