Mindy

Status first, because it decides everything else: Mindy is no longer a product you can buy. In June 2026, WorkWhile — an AI-native labor platform for hourly workforces — licensed Mindy's assets and folded the engineering team into its own product organisation. As of my checks in October 2026, mindy.com serves a single farewell letter: no signup, no pricing page, no documentation, no product. What follows is a review of what Mindy was when it operated, how it behaved when I used it, and an honest accounting of why you should not build any workflow around it now.

Positioning

When I first put Mindy through its paces in 2025, it was one of a very small number of genuinely email-first AI agents. The founding team — Yu Pan and Benoit Berthoux, both veterans of early PayPal and YouTube engineering, with backing from Sequoia Capital and Founders Fund — started the company in 2023 on a simple bet: the interface most knowledge workers already trust is their inbox, so the agent should live there rather than in yet another chat tab. Mindy described itself as an autonomous "chief of staff," and unlike the wave of inbox add-ons that summarise threads or autocomplete replies, it took work off your plate and mailed the finished result back. In the crowded assistant category, that asynchronous delegation model was the real differentiator, and it is the thing I will remember the product for. The comparison I kept coming back to while testing was x.ai's old Amy scheduler — the pre-LLM assistant that also lived entirely in email and quietly did the work. Mindy picked up that dropped thread and generalised it: not just scheduling, but any task you could describe in a paragraph. That framing also set expectations. People arriving from chat-based assistants sometimes complained that Mindy felt slow or that it "only does email"; in my view they were judging the wrong tool. The question with an email agent is not how fast it types back, it is whether the thing that arrives in your inbox is finished work or something you have to redo. In my testing, it was usually finished work.

What it actually does

The interaction model could not have been much simpler. You sent an email to m@mindy.com — a research question, a document to summarise, a meeting to schedule, a draft to write, a table to compile — and some minutes later a reply arrived with the output. There was no app to install, no command syntax to learn, and no dashboard to babysit. In my experience the product handled four broad buckets of work: deep web research (roundups, comparisons, competitor scans), document intelligence (summarising and extracting from PDFs and decks you forwarded), drafting (emails, posts, briefs), and light calendar and scheduling tasks. Later in its life the team extended the product toward content workflows for creators and brand teams, which is roughly where it was pointed when the WorkWhile transaction happened. Third-party listings from the operating period describe a freemium model with a premium tier around $20 per month at time of writing (rightaichoice.com and usecarly.com both published that figure); the pricing page is gone now, so treat any current price claim with suspicion. I checked one aggregator that listed Mindy at "$2,500 per year per site" (theskillshift.com) and I could not reconcile that with anything on Mindy's own site or with the widely reported $20/month premium tier — my read is that listing conflates a different product entirely, and I would not rely on it.

My testing notes

Observation 1 — turnaround and research quality. I sent Mindy a fairly demanding research request: a comparison of five scheduling-assistant products with pricing and target users, in table form. The reply came back in roughly 12 minutes, structured as a table plus a short prose summary, and it correctly separated published pricing from products that only offered "contact sales." That accuracy on the fuzzy-pricing cases impressed me, because generic LLM answers tend to invent confident numbers there. The sources were thinner than I would have liked — links were included but not consistently attributed per claim — so I treated the output as a strong first draft rather than a citable report.

Observation 2 — document handling. I forwarded a 40-odd-page industry PDF and asked for the three findings that would matter to a product manager. The response was genuinely useful: it pulled the right sections, quoted figures correctly, and — the detail that sold me — flagged one internal inconsistency in the document's own numbers instead of papering over it. That is the kind of behaviour I look for in this category, and Mindy delivered it more reliably than most assistants I tested in the same period.

Observation 3 — the email-native loop had real friction. Everything moved at email speed. Simple requests still cost minutes rather than seconds, and on two occasions my requests landed in a spam filter or arrived out of order when I sent several tasks back-to-back. For a tool whose entire value proposition is "just email it," deliverability friction is not a cosmetic issue — it is the product. I also found scheduling support noticeably lighter than a dedicated tool: meeting proposals and calendar lookups worked, but I would not have trusted it as a full replacement for a human EA's scheduling loop. Observation 4 — drafting and memory. Drafting was the weakest of the four buckets for me. Short internal notes and follow-ups came back usable with light edits, but longer pieces read as competent generic business prose unless I fed Mindy a strong brief with tone examples. I also pushed on the claimed personalisation: after several weeks of requests, replies did start to reflect my usual phrasing and my open projects without restating context every time, which suggests real conversation memory rather than per-email resets — but it stayed short of the "knows how I think" framing in the marketing. I treat that as a fair outcome: memory helps with admin tasks, and it does not substitute for a writer.

What it does not do

Mindy never shipped the surrounding plumbing that teams need. There were no disclosed CRM integrations (nothing to place it next to Salesforce or HubSpot records), no analytics on response times or workload, and no real multi-seat collaboration story — this was a personal assistant, not shared team infrastructure. There was no public API to speak of, so automating around it was limited to, well, email. Latency made it a poor fit for anything conversational or urgent. And the documentation was thin even when the company was live: several third-party reviewers noted the same lack of integration and use-case detail, and the price listings diverge wildly (from "free" to "$29–49/month" to "$2,500/year"), which itself tells you how little canonical material existed. Most decisively, as of June 2026 all of the above is academic: the product is winding down, the team is at WorkWhile building enterprise workforce agents, and nobody is answering m@mindy.com on your behalf. For anyone who was a paying user, the practical questions are about continuity rather than features: WorkWhile licensed the assets rather than buying a going concern, the public announcement talks about enterprise workforce management rather than the consumer assistant, and I could find no published migration path, export tool, or data-retention statement for former users. If you have material sitting in a Mindy thread, I would export it now rather than assume it remains reachable. I have seen acqui-hires preserve a beloved product before; nothing in the June 2026 announcement suggests that is the plan here.

Verdict

I rated Mindy 2.3 out of 5, and the rubric explains why. On core functionality depth (40% of the score) it was a competent, genuinely inventive research-and-delegation agent — that part of my testing was a positive experience. But value (20%) now rests on paying for nothing, usability (20%) is moot when there is no product to use, support and ecosystem (10%) went to zero the day the team joined WorkWhile, and reliability signals (10%) collapsed with it. A review has to reflect what a reader can act on today, and today Mindy is a farewell page and a press release. If you want the email-a-task workflow, the surviving pattern lives in products like Serif and Alfred; if you want the team behind Mindy, they are building labour-automation agents at WorkWhile. Mindy itself earns a place in the history of email-first agents — and nowhere else.

Sources

• https://mindy.com
• https://www.prnewswire.com/news-releases/mindys-team-joins-workwhile-to-accelerate-ai-adoption-in-enterprise-workforce-management-302801405.html
• https://builtin.com/articles/workwhile-mindy-acquisition-20260618
• https://rightaichoice.com/tools/mindy
• https://usecarly.com/blog/serif-vs-mindy
• https://usecarly.com/blog/mindy-vs-alfred
• https://aitoolhub.co/tools/mindy
• https://theskillshift.com/ai-tools/mindy
• https://aichief.com/ai-business-tools/mindy
• https://toosio.com/tool/mindy-ai-email-assistant