Prison Phones and the Problem with Profits: Can an upstart tech nonprofit take on the $1.4 billion prison communications industry?

A Dispatch from Maine Correctional Center

The trees thin out on my drive from downtown Portland to Windham, Maine, a small town home to Maine Correctional Center, a 650-person medium and minimum security facility. I’m here as “the press” to write and report about Ameelio, a rapidly growing tech nonprofit hoping to reduce barriers families face trying to stay in touch with their incarcerated loved ones. It’s a few minutes before noon, and I sit in a rented Nissan Sentra in the small parking lot before others arrive.

I text a bit with Andrew Lama, one of their program managers, as the team drives in — Uzoma Orchingwa (who goes by Zo) and Gabe Saruhashi, the nonprofit’s two founders, are flying straight from a fundraising trip and are tied up at the airport. They’ve all come to demo Ameelio Connect, a video call, voice, and messaging platform they’re aiming to pilot here in Maine.

It’s a compelling proposition. Under current pricing schemes imposed by large prison contractors, video visits and phone calls can cost more than $1 per minute, keeping them well out of reach for families that are already disproportionately likely to live below the poverty line. It’s also a devastating missed opportunity; a 2020 study by the Minnesota Department of Corrections found that “the hazard of recidivism [for an incarcerated person] decreased by 3.1 percent for general reconviction” for every additional video visit they received.

Ameelio hopes to eliminate this barrier — by leveraging a wealth of open source projects and software’s ability to scale rapidly at low cost, the nonprofit is able to offer their platform for free for the families and incarcerated people who use it. They’re off to a promising start: While we tour Maine’s facilities, which they hope to add with this trip, they’re already piloting in Colorado and live in all nine of Iowa’s facilities.

Zo and Gabe arrive a few minutes later and we walk with corrections staffers to a cinder block-walled classroom in MCC’s Dorm 6, where the team will test the platform on demo calls Andrew has prearranged with family members. Gabe tells me he’s been living out of a suitcase, but it doesn’t show; I watch as he and Andrew deftly arrange two WiFi hotspots and laptops, a lightweight trial before setting up on the facility’s larger network. It seems like a well-oiled machine — Jon*, a soft-spoken “resident” (Maine’s official term for incarcerated people), is too absorbed talking to his older brother, who’s calling from his lunch break, to notice the team carefully monitoring the demo, which runs without incident.

Are there other advocacy efforts that might be well served by familiarity with and clever application of the MERN stack?

When the calls end, we follow Jon and two other residents into another room to ask about their experiences in online college courses, feedback which will inform work on Learn, Ameelio’s effort to integrate educational and vocational training into its platform. Learn is aimed at further reducing recidivism — a 2013 RAND study found that incarcerated people who participate in correctional education programs have 43 percent lower odds of returning to prison than those who do not, and that every $1 investment in prison education programs corresponds to a $4 to 5 reduction in incarceration costs during the first three years post-release of a resident. Maine’s facilities seem to place particular emphasis on such programs. Jon, for example, is finishing up his Masters of Public Administration online and took the LSAT in the facility; he plans to get involved in criminal justice reform and restorative justice work when he’s released.

The rest of the day is packed with tours and meetings. We demo the platform with two more residents in a small conference room — again, it’s flawless.The team asks the residents more about their pain points staying connected with their loved ones. Both tell horror stories about recently installed Global Tel Link (GTL) video visitation kiosks that work so unreliably that many others in the facility have given up on using them. When the system fails, they aren’t refunded for the time they’d paid for.

We briefly linger with some corrections officers before touring the next wing, and one of the officers talks about their cumbersome current visitation system. Visitors have to manually re-enter scheduling info for each individual and time and day through the Maine.gov website, a convoluted process that takes numerous clicks and confuses some potential visitors. Even this, however, is an upgrade — until 2015, the department coordinated visits by hand on paper. The process remains incredibly time intensive for corrections staff, and these organizational challenges can mean turning families away last minute from visits. Ameelio’s app attempts to simplify the process, handling in-person visits in the same place as video calls and messaging. (The streamlined interface came from early feedback in Iowa.)

The team’s attention to detail is clear — they walk with a sort of bookish obsessiveness, each employee stopping occasionally to jot something down in a small notebook. I regularly see one or two steal off to ask a resident about a tablet they’re using, to ask an official about administrative protocol, or to look more closely in a rec room or library.

Walking through a hallway of the Women’s Center, Gabe spots a shiny metal console that sticks out from the lime green wall of a break room, and he asks if we can go in. It turns out to be one of GTL’s boxy communication kiosks, positioned confusingly about a foot below eye level. The team jumps on the opportunity to try out the machine, nimbly pressing buttons, but grimly amused by their shared inability to get the arcane system to work. “I think we’ve pressed every button,” Gabe quips. “It’s gonna take your five dollars, seven times.”

Incentives

Ameelio is up against behemoth competitors. A 2017 report from the Prison Policy Initiative (PPI) found that three companies — Securus, Global Tel Link, and CenturyLink — control between 70 and 85 percent of the $1.4 billion prison communications industry. In 39 states, it’s legal for these contracts to include “kickbacks,” deals where state facilities receive a percentage of revenue that contractors make charging families for their services. For example, as reported by the Verge, a 2013 contract between Alabama’s Baldwin County and CenturyLink would give an 84.1 percent commission on gross revenue back to the state alongside “a guaranteed minimum average Commission of $55 per inmate per month;” the state earned more than $3 million per year from inmate phone services in 2014.

Some contractors go so far as to offer “no-cost” bids — installing the equipment for free in exchange for a monopoly on a facility’s communications (and, consequently, a percentage of that revenue). As such, companies have no reason to lower prices or provide a high quality product; in fact, they’re often incentivized to do the opposite. For example, 74 percent of jails banned (free) in-person visits after implementing (paid) video visitation, according to a 2015 report from the Prison Policy Initiative: Securus explicitly requires the elimination of face-to-face visits in its contracts for video kiosks. The products are decisively worse; video products are often riddled by audio lags, pixelated screens, and full-minute freezes that make visits start late or end abruptly. Additionally, the lack of privacy was so egregious that criminal defense attorneys in Texas sued Securus for violating the constitutional rights of their defendants, alleging that the sheriff’s department recorded (what were supposed to be) confidential attorney-client calls.

When startups aim to disrupt stagnant, greedy incumbents, they often, a few years later, find themselves just as inefficient, mediocre, and complacent. Unfathomable, money-burning offers designed to juice adoption numbers (like $10 rideshares across town or $15/month unlimited movie plans) disappear as these companies are forcibly weaned from growth capital. Their ultimate ability to disrupt a social order is tamed by their incentives and status as corporations; the very investment that allows for such “revolutionary products” often demands returns that force companies to prioritize shareholder value over all else.

A for-profit startup would almost certainly have this fate; for the prison communication industry, it is the incentives themselves that cause the incumbents to build products so poorly. This is also why it’s so difficult to disrupt — per another PPI report, many of the kickbacks paid out by contractors often include campaign contributions or payments to organizations like the National Sheriff’s Association, which has consistently claimed the country’s lack of progress on prison phone regulation as a lobbying victory. This creates a particularly insidious collaboration between prison officials and industry — a clear manifestation of the prison industrial complex, where industry funds “tough on crime” politicians who pass laws to incarcerate more people and allow such predatory contracts, the profits from which businesses will reinvest in the same politicians’ campaigns.

It is this exact arrangement Zo wants to avoid: “We fear becoming a prison contractor, and the goal is to not exist in 15 years,” he tells me. Unlike for-profit companies which benefit from an increasingly large prison system, Ameelio can define its success in terms of providing service — right now, communication and education — that greatly decrease an incarcerated person’s chance of reoffending and, in turn, reduce prison populations. “[We want to] shrink the prison system over time enough to become irrelevant… Every nonprofit should ask ‘how can we not exist as quickly as possible?’”

Before the Startup

In Y Combinator founder Paul Graham’s essay “Organic Startup Ideas,” he writes:

So if you want to come up with organic startup ideas, I'd encourage you to focus more on the idea part and less on the startup part. Just fix things that seem broken, regardless of whether it seems like the problem is important enough to build a company on. If you keep pursuing such threads it would be hard not to end up making something of value to a lot of people, and when you do, surprise, you've got a company.

Graham fails to consider a complication — what if the problem exists because some other people tried to solve it by starting companies?

While Zo had long been drawn to criminal justice work, he specifically learned about the brokenness of prison communications when he came across the Ella Baker Center’s landmark 2015 “Who Pays?” study shortly after finishing his Masters of Philosophy in criminology at Cambridge.

The report details myriad costs — financial, physical, emotional — imposed by mass incarceration, particularly on families. The “Costs of Maintaining Contact” section bore particular relevance. Zo’s criminal justice coursework had made clear to him how crucial social support could be — studies suggest that experiencing visitation resulted in a 26 percent decrease in recidivism (reoffending after being released) — yet calls remained financially out of reach. In fact, 69 percent of participants in the Ella Baker Center’s survey identified the financial cost of phone calls as a barrier, and 34 percent of families reported going into debt to pay for phone calls or visitation. “Families are often forced to choose between supporting incarcerated loved ones and meeting the basic needs of family members who are outside,” the study reads.

The problem, however, is notoriously difficult to dislodge via legislation. The most famous battle began in 2000, when Martha Wright filed a complaint against the Corrections Corporation of America (a large private prison company) for their phone arrangements on behalf of her incarcerated grandson. A federal court referred the case to the Federal Communications Commission (FCC) in 2003, which took no visible action for nearly 10 years. In 2013, the commission voted to approve rate caps for out-of-state-calls, and in 2015 voted for more caps, including in-state calls and “ancillary fees” often tacked on to call prices.

At the time, this legislation was a huge win. Before it, companies extracted even more money through ridiculous-sounding fees — often doubling the effective price of a call for incarcerated people and their families — on top of the commissions they passed on to facilities. For example, a 2013 PPI report found that the now-shuttered Legacy Inmate Communications could charge you a $2.50 per month “Network Infrastructure Fee,” $3.00 per call “Premise Impose Fee,” and $1.95 per month “Regulatory Compliance Fee” on top of the actual price of your call.

Many of these fees also took advantage of the precarity of family members living paycheck to paycheck, through strategies like maximizing the number of small payments people make, each carrying a fixed convenience fee. Per the same PPI report, in the case of TurnKey Corrections, this fee was $8 per transaction, nudged by a particularly suggestive and insidious “Please enter today’s amount,” in-app prompt.

The FCC’s actions were clearly long overdue. “In my 16 years as a regulator, this is the clearest, most egregious case of market failure I’ve ever seen,” FCC Commissioner Mignon Clyburn wrote in a statement in support of the reforms. By 2017, however, some of the rate caps had already been reversed.

Further legislative change felt distant and improbable when even its large victories could be quickly erased. Zo soon began to think about solutions that could deliver progress with more immediacy. He saw law as central to social change in the 20th century, with lawyers like Charles Hamilton Houston (the NAACP’s first general counsel) and Supreme Court Justice Thurgood Marshall as cornerstones of the civil rights movement. For Zo, technologists seemed to have the same sort of leverage in the 21st.

“I started trying to look around and learn about civic technologists that have built tech to solve social problems, not building technology just to create leisure for people that already have it,” he said.

Inspired by nonprofit internet platforms like Watsi and Kiva, he sought out a technical cofounder. At the time, Gabe was a Yale senior who’d spent two summers as an engineer at Facebook and had taken time off to be a product manager at a startup going through Y Combinator. Their complementary expertise lent itself to fruitful collaboration — Zo would handle programs and policy, while Gabe focused on technical development.

I first heard about Ameelio while watching the team present at the Mozilla 2020 Fix-The-Internet Incubator. Then, Zo and Gabe were just two Yale students who’d built an app that allowed family members to write messages the nonprofit would then print and mail to incarcerated loved ones. The scale was overwhelming enough that they considered throttling the service to keep up with postage or printing costs. While the service still exists (now called Letters) and allows loved ones to send two for free (according to Zo, they’ve now sent 1.8 million letters in total), the team’s core focus is now on Connect, their video calling product.

Their team, now around 15 people, is a cross between a startup-like product team of engineers and designers, and a program staff who work on policy initiatives and coordinate with departments of corrections. Their nonprofit structure necessitates a mission-oriented team; without profits, their employees can’t get the sort of equity or compensation similar roles at tech companies might fetch.

“I suspect that Ameelio is an outlier in its ability to attract a large, smart, and motivated group of students and young people to be part of their team and to operate on a shoestring,” Mozilla Builders co-founder Bart Decrem told me. “Very few [similar efforts] are as well run or able to harvest as much energy as Ameelio. I think that’s one of their superpowers.”

A few months after finishing Mozilla’s incubator, Zo and their first full time hire, Andrew Lama, would move to Iowa to troubleshoot the first launch of their video calling product. Iowa was an unlikely choice — kickbacks are legal in the state. Iowa’s old provider, however, did not support video calling, so Ameelio could replace them based on service quality alone. The specifics of prison procurement are not readily available, so Ameelio figured out the details of how to structure a proposal by sending thousands of Freedom of Information Act (FOIA) requests to find the terms of past contracts before beginning their Iowa launch in June of 2021 — they’d spent months convincing the department and iterating on the product with them. They’ve also pursued more initiatives that build on their strengths, like mailing a voter registration guide and all of the necessary forms to all 1,709 of Maine’s incarcerated residents in the 2020 election.

Ameelio’s results have attracted the interest of huge funders, mostly technologists — Jack Dorsey, Reid Hoffman, Vinod Khosla, and Kevin P. Ryan contributed individually — alongside grants from the Shuttleworth Foundation, True Ventures, and Schmidt Futures. “Convincing tech founders and venture capitalists to invest in a nonprofit is challenging,” Zo said in a 2021 Freethink interview, in reference to Ameelio’s lack of financial returns. The framing of such donations as investments, however, gestures at an alternate version of philanthropy — as “risk capital for public goods,” as Nadia Asparouhova writes — that looks more like the high-upside bets on teams that angel investors make.

A common critique of techno-solutionism is that the (often elite, highly educated) builders of technical products have little in common with the real experiences of their users — in this case, that of having an incarcerated family member. But Zo’s approach is informed by his childhood and the friends he grew up around. When we talk by phone a few weeks after the Maine rollout, he tells me about a photo he keeps in his wallet of his seventh grade travel basketball team. He recounts that when he moved to Connecticut from Nigeria at age nine, he was nerdy and shy — basketball gave him confidence and a group of friends. In his senior year of high school, he got a call that one of those friends had been arrested on a gun charge.

“I knew about the high rates of incarceration in our community, but I’d never had a really close friend of mine be on the verge of incarceration,” Zo says. Of the eight other boys on that basketball team, four would go on to be incarcerated. “These are kids that I’ve known since I was 10, 11 years old...It shook me up. I understood that no one wants to be in such a terrible place that they’re committing crimes. Not to take away agency from anyone, but the conditions which we are born in contribute so much to the decisions we make later in life. That was one of the catalysts to recognize that I could’ve very easily been one of them.”

Building Product in a Software Non-Profit

Ameelio’s current focus is on achieving financial stability and scaling the organization’s reach. While their size means costs are far lower than those of their competitors — they previously considered removing the “Team” section of their website after a department of corrections feared they were too small to provide adequate service — they still remain reliant on donor funding. Zo spends much of his time fundraising and, as such, the organization is constrained in what it can do based on what it can successfully raise money for.

For example, while the first two letters per month sent through Ameelio’s Letters product remain free, users need to purchase additional mail pieces, although there’s also an option to request additional credits if “you truly can’t afford to send mail to your loved ones.” Because Ameelio bears printing and mailing costs, universally unlimited letters are not financially viable, but they still undercut the alternatives — a December 2020 survey of 300 Ameelio users found they reduced mail costs by 74 percent. They’re also exploring new potential revenue streams to be financially self-sustaining (instead of reliant on donor funding) — charging external providers to host training or educational programs on Ameelio Learn, selling the physical equipment needed to access their services (like custom Android tablets), or charging external professionals like lawyers to contact clients — that can still keep the digital platform completely free for users.

Lacking the signals most startups have to evaluate their success (like revenue), the Ameelio team established a multi-year partnership with the University of Chicago’s Crime Labs to study the platform’s impact. They track a variety of outcomes — one- to three-year recidivism rates, mental health conditions, and behavioral infractions, among others — across similar facilities, comparing those which have Ameelio’s technology and those which don’t. Ameelio is unusual among initiatives the researchers work with.

“It’s a young startup,” postdoctoral research fellow Nour Abdul-Razzak told me. “They take data very seriously.”

Some of this interest in data reflects the fundamental challenge of researching the needs of incarcerated populations. Ameelio primarily does user research with family members on the outside, who often have packed schedules and limited data for video calls. Elizabeth Gray, Ameelio’s second hire who works as a product designer and researcher, has experimented with different compensation amounts for research calls, scheduling reminders for upcoming calls, and learning how families tend to fit these calls into their schedule in order to better learn their needs. While Elizabeth emphasizes Ameelio’s ability to function like a tech company, they’ve had to adapt many tech industry norms to new constraints. For example, the team stopped working on a sprint-like engineering schedule because many of their features are based more on the timing of contracts with corrections officials rather than clean two-week blocks.

Elizabeth also notes she’s had to work around huge constraints in reaching incarcerated individuals, in order to design the product with them. Up until her visit to Maine, most of her own knowledge about incarcerated people came from a combination of coworkers, loved ones she could talk to on the outside, and online discussions. “[That’s] better than nothing, but as someone who really cares about talking to the people that are impacted, it always felt not satisfactory to me and that’s something that I’m even still trying to think about how to do better.”

Ameelio’s current approaches include talking with formerly incarcerated people who were recently released. They’ve added a monthly check-in with a resident we met at Maine State Prison to show early iterations of new products, and to do usability testing in hopes of gaining more granular insight into how well the system works.

The Ameelio team attempts to use these check-ins to ensure the product they are building remains centered on the needs of the incarcerated people they serve, but they are still constrained by the prison’s restrictions. For example, a first iteration of the Ameelio video product included a half-hour delay feature on messages that was requested by the Iowa Department of Corrections, which “was something we at Ameelio obviously didn’t agree with, but was something we had to include in the first iteration to get our foot in the door,” Elizabeth said.

“I know that policy does a really good job of toeing the line between respecting a lot of DOC’s wishes while also slowly trying to push against extraneous surveillance tactics,” Elizabeth adds. In a recent usability test, two incarcerated people using the platform mentioned disliking the auto delay feature — they compared it to writing a letter versus sending a text. “I brought that back to the team and that, combined with some of the internal philosophy we have, we were able to push a facility to swap their default to have no delays.”

In writing this piece, I frequently wondered about the limits of such an approach, whether Ameelio might have to compromise on functionality or build features that harm incarcerated people at the behest of a facility. Elizabeth acknowledged that balancing these stakeholders, at times, presents tension, but that she gains more from viewing departments of corrections as partners in the platform’s roll-out. For example, she recalls building functionality to flag messages or prevent them from being sent, a feature requested by the Iowa DOC, typically for security reasons.

I’d find some sort of “gotcha” that proved it to be garden variety techno-solutionism, a prison contracting grift that had been elaborately social-justice washed.

“When I was deep into doing research and design for this feature, in my mind I was going back and forth between trying to empathize with DOC, and empathizing with family members and incarcerated people about how their experience would feel seeing that a message was rejected,” Elizabeth said. This is a necessary constraint of building technology in such facilities; because Ameelio needs the blessing of DOCs for their rollouts, they can’t include functionality that might be great for users if it runs too much up against the department’s wishes.

The bar was extremely low for building better functionality. Iowa’s previous message service would never notify incarcerated people whether a message was delivered or rejected, resulting in abrupt breaks in conversation that confused families or conveyed unintended terseness. In designing a functionality to indicate whether a message was delivered, Elizabeth said she initially assumed that people would rather “[be given] information” than “[left] them in the dark.” After testing with a few people, she confirmed the need for such a feature. “Even though it’s not great to have your messages surveilled [by the DOC], people definitely appreciated the transparency that came with knowing when their message was rejected.” In addition to the notification, the feature required the facility to give a reason for why messages were rejected. “We were excited about requiring that information [for why a message was rejected] to make sure that people on the outside were getting a substantial reason for why their messages weren’t going through.”

Towards Liberatory Technology

The evening after our visit to Maine Correctional, I met the Ameelio team at the only Korean barbecue restaurant in Portland, Maine. For a few team members, that morning had been their first ever site visit, and many had never met in person.

I was drawn to Ameelio for its application of tech toward more useful aims than those of the typical ad targeting or B2B SaaS tools. I was intrigued by the way that it defined metrics beyond financial returns — the sheer scale and quantity of calls, the potential reduction in recidivism. What I had ignored was the emotional, non-instrumental benefit of the tools, the essential service of offering social support to incarcerated people who need it just as much as anyone else does.

Maine’s department is unusually progressive, and the vast majority of corrections officers we met are, I think, earnestly dedicated to the support and rehabilitation of the facility’s residents; Maine’s Department of Corrections Commissioner Randall Liberty, for example, took the job because his father had been incarcerated. However, there are limits to how different Maine’s approach can be, simply given the nature of American mass incarceration. Up close, I was still put off by the core principles of prisons themselves, particularly around surveillance and limits on resident autonomy; these approaches seem to just re-inflict harm.

What makes Ameelio interesting, then, is not its technology, but rather its implicit view of incarceration as a product of circumstance that often results from material need or an absence of social support networks. The team’s educational and communications products earnestly seek to provide opportunities and support that could facilitate re-entry, rather than simply punishing people for their past offenses. The technology itself is fairly standard; one can easily find YouTube videos explaining how to build a video chat or messaging platform, or to create a dashboard capable of permissioning access. Gabe confirmed this in a 2021 interview with TechCrunch. “We leverage a lot of open source tech, which is why part of our costs are so low,” he said. “They use Twilio, we use mediasoup; the only thing we’re paying for is servers. And we use Kubernetes, so our total cost right now is like $100 a month.”

This, however, is far from a dig at Ameelio’s technology. Previous attempts by companies applying technology to prison communications have made things decisively worse — take, for example, Securus requiring the end of in-person visits when installing its video visit kiosks. Ameelio distinguishes itself in the way the team frames technology as a tool — one, in this case, that has been thoughtfully applied to advance crucial advocacy work, built on a nonprofit structure that doesn’t (by nature) pit Ameelio’s incentives against those of its users.

I came into this piece with some suspicion of Ameelio, too, a sense that, if I kept looking, I’d find some sort of “gotcha” that proved it to be garden variety techno-solutionism, a prison contracting grift that had been elaborately social-justice washed. Wasn’t there a necessary tension between “liberatory” and “funded by tech billionaires,” and wouldn’t I find a red herring (probably after locking myself in a room and reading a lot more theory) that proved this was too “reformist” of a reform and worked against “real” grassroots advocacy work? It’s certainly true that Ameelio has to make tradeoffs — in platform design, in funding — to exist, but it’s quite difficult for me to imagine that it isn’t ultimately an overwhelming net positive, both for incarcerated people and in reducing recidivism, a step toward broader decarceration (or, at least, shrinking prison populations). To even ask these questions is a paradigm shift. At the very least, I can’t help but think — wouldn’t we be way, way better off if more ambitious young computer science graduates looked to Ameelio as an example over, say, Palantir? Are there other advocacy efforts that might be well served by familiarity with and clever application of the MERN¹ stack?

---

Toward the end of our day at Maine Correctional Center, we run into Ana*, one of the women the team had demoed the product with a few hours before. She invites us to come see her room; that morning, she’d mentioned some familiarity with the app, noting that her husband often sent her photos through Letters.

Inside is a bulletin board covered in photos. She’d told us she was interested in becoming a lobsterman on release, and there are a few photos of lobsters, and of sunsets over the ocean. Many are of her kids — her daughter on an ice rink, her son sitting on a boat, a birthday party — and adorable spotted bulldogs, sitting on the couch or dog bed. The photos her husband got from a cumbersome print system at a local Walmart would always come out blurry, she said. Ameelio’s Letters had printed the photos in far better quality — she can see them more clearly.

Zo beams as she eagerly shares the stories behind each photo, like she’s introducing us to her family. “That one came out super nice,” Zo says, pointing at a photo of one dog with its tongue out. “Thank you so much for showing us this.”

*Names have been changed to protect the privacy of these individuals.

¹ MERN (MongoDB, Express, React, and Node.js) is a common software stack used to develop web applications.

Lucas Gelfond (he/him) is a writer and software engineer. He publishes an (irregular) newsletter about people and publications shaping discourse and culture called ZINE MUNCH (www.zinemun.ch). His writing has previously appeared in VICE, Logic, Dirt, Reboot, and the College Hill Independent.

---