The fax machine is not supposed to be a medical device. Yet when Mahima Chawla began asking colleagues about parental and medical leave, it kept appearing in the stories: people calling insurers from hospital rooms, hunting down forms, and trying to decode a paycheck while caring for a baby or a parent. Chawla had assumed leave was simple. The surprise became the company.
In 2020, Chawla joined Lauren Dai and Amber Feng, operators shaped by Square and Stripe, to found Cocoon in San Francisco. Their pitch was not another place to record a date away from work. Cocoon would ingest company policy, federal and state law, insurance benefits, payroll inputs and an employee's circumstances, then turn that pile into a leave plan. It would calculate eligibility and pay, generate required notices, guide claims and show everyone the right version of what was happening.
That distinction matters. Vacation software answers, “Is Alex out Friday?” Leave administration answers much stranger questions: Which of several overlapping programs applies? Who pays on day 19? Does an intermittent schedule count against an FMLA balance? Which medical detail may a manager see? And what changed in Connecticut last month? The product is less calendar, more compiler.
The discovery work was refreshingly low-tech
Before there was a rules engine, there were conversations. Chawla has described 20 to 30 interviews with employers. The founders worked their networks for introductions to HR leaders and posted a Craigslist ad offering 30-minute research interviews to people who had taken leave. They were looking for conviction, not compliments. What they heard was that employees could spend more than 30 hours navigating forms and pay, while HR teams kept side spreadsheets because the official system did not handle state-by-state reality.
This is the first useful thing to copy: investigate the process everyone believes is already handled. Regulated back offices often look solved from the executive floor because somebody diligent is quietly repairing them by hand. Cocoon's founders found the hidden labor and treated each workaround as a product requirement.
They also launched narrowly. Cocoon's first customer was Benchling. In January 2021, the product supported parental leave in California. One customer, one state, one leave type: an almost comically small opening for a company promising national infrastructure. But it forced the team to solve a complete journey before collecting a museum of half-finished features. Medical and caregiver leave followed, then personal leave, intermittent schedules, ADA accommodations, variable work schedules and richer payroll and manager tools.
- Connect policy and HR data
- Model eligibility and pay
- Guide notices and claims
- Track leave in real time
Software became the case manager
Traditional administrators often assign one representative to a leave. The approach sounds comforting until that person is unavailable, inconsistent or carrying too many cases. Cocoon made software the front line and kept certified specialists behind it for claims, pay and unusual questions. Employees can model a leave privately before telling a manager, see estimated income sources, file claims and watch payments. Administrators get dashboards, deadlines, notices, balances and payroll files without receiving every sensitive medical detail.
“As gray as the leave world is, Cocoon makes things a lot more black and white.”Lauren Serrano, leave leader at Angi, in a Cocoon customer case study
The software-first bet changes both experience and economics. An employee does not have to wait for office hours to inspect a plan. A People Ops team does not need a new caseworker every time leave volume grows. The rules stay consistent across locations, and legal updates can flow into the product rather than somebody's bookmarked government page. Human expertise remains essential, but it moves from repeating standard instructions to handling exceptions.
Cocoon packages that work into Core and Complete plans. Core covers FMLA leave types, planning, compliance, claims guidance, reporting, SSO, support and more than 20 HRIS integrations. Complete adds personal and ADA accommodation leaves, detailed pay sources, custom payroll files and dashboards for managers, HR partners and pay administrators. A newer Leave Planning product offers the planning layer to carriers and third-party administrators.
The company does not publish dollar prices. It charges an annual per-employee fee based on company size, plus a one-time implementation fee, and says implementation typically takes four to six weeks. That makes Cocoon a considered B2B purchase, not a self-serve subscription for a five-person office.
Empathy opens the door; arithmetic closes the deal
Cocoon's mission is to empower every working person to care for what matters most. The buyer, however, still needs a budget. The company's sharpest commercial claim is that customers recover an average of $12,515 per leave. When an employee fails to claim state or private-insurance benefits, an employer with a fully paid policy may cover more salary than necessary. Cocoon guides the claim and coordinates those payments, reducing the employer's share.
Renewal and self-service figures were reported by Cocoon in 2025. BAGGU's case study describes a reduction from more than three hours to about 30 minutes per leave, shown here as an approximate percentage.
The customer evidence is specific enough to be interesting. BAGGU said leave administration fell from more than three hours to roughly 30 minutes. Angi, operating across 47 states, replaced AbsenceSoft and separate state trackers in about two months; automation let the company redirect a planned leave-administration hire into a more strategic benefits role. Cocoon reported that 96 percent of customers joining from mid-2023 had renewed, and that 85 percent of employees completed their leave without administrator help.
Customers named publicly include Benchling, Affirm, Ironclad, Superhuman, Vanta, Carta, Miro, ActiveCampaign, HoneyBook and Gem. This is enterprise and mid-market HR infrastructure sold mainly to Benefits and People teams, with payroll, legal, managers and employees pulled into the workflow. It competes against AbsenceSoft, Sparrow, Tilt, Larkin, insurers, PEO services and that eternal incumbent: the heroic spreadsheet.
Partnerships supplied the distribution
Cocoon raised a $5.5 million seed round in December 2020, then announced a $20 million Series A led by Index Ventures in September 2021. First Round Capital, SemperVirens, XYZ Venture Capital and Magnify Ventures also participated. The company later took an undisclosed strategic investment from ADP Ventures.
ADP became more than a logo on the cap table. In 2024 it named Cocoon a strategic leave-management partner, listed the software on ADP Marketplace and connected it to Workforce Now for SSO and automatic employee-data updates. Cocoon also built referral and pricing relationships with benefits brokers and consultancies, including a vendor-of-choice relationship with Sequoia. For enterprise software, these channels matter: the trusted payroll or benefits adviser is often present before a buyer searches for a leave product.
Then came TriNet. In April 2026, the HR platform acquired Cocoon and made it a wholly owned subsidiary. TriNet's quarterly filing put the purchase price at $23 million, including $1 million of Cocoon's own cash paid to former shareholders, for roughly $22 million net cash. Financial terms had not appeared in the original press release.
That price creates an awkward but important footnote. Cocoon had disclosed $26 million of funding by 2021, before ADP's later investment. A $23 million acquisition does not imply a $23 million value for every stakeholder, and acquisition accounting is not a cap-table payout schedule. It does mean the outcome should not be airbrushed into a giant venture return. Cocoon can have built a useful product, helped thousands of people and found a strategically sensible home without producing a blockbuster exit.
What changed their mind - and what to steal
Chawla's initial belief was that taking leave should be straightforward. The stories changed her mind: hours on government hotlines, missing payments, spreadsheet arithmetic and medical moments interrupted by administration. The founders then tested whether those stories described isolated bad luck or a repeatable market. Employer interviews showed the same breakdown on the other side of the desk.
Five moves worth borrowing
- Recruit users outside your network. Craigslist supplied less polished, more revealing research.
- Begin with a complete narrow workflow, as Cocoon did with Benchling's California parental leave.
- Codify repeated expert decisions, but keep specialists available for exceptions.
- Give employees privacy by default and managers only the information their role requires.
- Pair the human promise with buyer arithmetic: time saved, risk reduced and benefits recovered.
There is a cultural move to copy, too. Feng wrote that the three founders agreed before starting Cocoon that work could not be the most important thing in their lives. It was an unusually on-theme constraint for a leave company, but also a useful piece of operating design. They chose the boundary before headcount and habits made it harder to install.
The conditions where Cocoon is a mismatch
- A tiny company that needs only vacation approvals may not justify annual per-employee pricing and implementation.
- A buyer who wants one named representative for every leave may resist a software-first support model.
- The automation depends on accurate employee, policy and payroll data; messy inputs still create messy outputs.
- Cocoon's documented strength is US leave law. International absence programs require different legal coverage.
- Highly bespoke policies and rare exceptions still need judgment. Codification reduces expert work; it does not abolish it.
The acquisition gives Cocoon what it lacked: immediate reach into tens of thousands of TriNet's small and midsize business clients. TriNet gets automated compliance, real-time tracking, integrated claims and payroll calculations without building the stack from scratch. The risk is familiar to anyone who has watched a focused product enter a broad suite: distribution improves while the clean experience slowly collects enterprise furniture.
Cocoon told customers the same product, team and support would remain. That promise is now the interesting test. Its advantage was never merely a database of leave laws. It was the decision to make an intimate, confusing process legible without turning people into ticket numbers. If TriNet preserves that, the fax machine can finally return to its proper role: gathering dust.