The quiet truth about task apps in cross-border ops
Cross-border sellers don’t lose money because their to-do app lacks a Liquid Glass redesign. They lose money because a restock reminder lives in one person’s head, a supplier follow-up sits in a WhatsApp thread, and a VAT filing deadline hides in an email nobody owns. So when I see a productivity app ship an AI update, I don’t ask “is it pretty?” I ask “would this survive a Monday morning at a 12-person Amazon agency with three marketplaces and a TikTok Shop launch in flight?” That’s the lens I’m bringing to Doneit 3.0, which launched on September 19th, 2025 and is the fourth annual release from the same small team.
What Doneit 3.0 actually is, and the problem it’s chasing
Let’s be precise about what shipped, because the Product Hunt page is thin on marketing fluff and heavy on feature bullets. Doneit is an Apple-ecosystem task manager — iOS, iPadOS, macOS, watchOS — and version 3.0 brings the new Liquid Glass design language plus a reworked “Doneit Assist” planning feature. The maker, Daniils, describes the update as including support for the latest OS versions and a set of AI-driven planning capabilities.
The headline capability, per the maker’s launch comment, is multimodal prompt support: you can attach photos when planning tasks, so a handwritten list, a whiteboard, or a document becomes structured tasks. Doneit Assist also gained more customization at planning time — adding extra detail to sharpen a task, suggesting reminders and attributes, and breaking tasks into subtasks. It can generate a brief summary of a task list, highlight key tasks, and surface upcoming items. Tasks and lists are now exposed to the new Siri AI for onscreen awareness, you can add tasks via Siri, and the Shortcut for adding a task now accepts more configurable parameters. There’s also a new extra-large portrait widget option.
That’s the factual surface. Now the interesting part: what problem is this actually solving for someone who runs a store?
Why Amazon sellers should care more than Shopify ones
Here’s a distinction I keep coming back to. A Shopify DTC operator’s workday is mostly creative and campaign-shaped: briefs, creative reviews, email flows, ad iterations. Their task list is fuzzy and forgiving — if a Klaviyo flow review slips by a day, nothing breaks. An Amazon seller’s workday is deadline-shaped and unforgiving. Replenishment windows, removal-order cutoffs, case-log follow-ups, A+ content approvals, inventory reconciliation against Amazon Seller Central reports. Miss a restock PO and you’re out of stock for three weeks while your BSR collapses.
That asymmetry matters because it tells you who should actually evaluate a tool like Doneit. If your operational risk is concentrated in hard deadlines tied to platform mechanics, a task app that can ingest a photo of a supplier’s packing list or a whiteboard from your 3PL walkthrough and turn it into dated tasks is genuinely more valuable than one that just looks nice. If your risk is creative throughput, this is a nice-to-have.
How it differs from the incumbents you’re probably already paying for
I’ve run ops across Asana, ClickUp, Notion, Todoist, and Things. None of them are bad. The question is where Doneit 3.0 sits relative to them.
Against Todoist and Things: both are excellent at the “capture fast, organize later” loop, and both are cross-platform in ways Doneit isn’t. Todoist runs everywhere including Android and web; Things is Apple-only but has a decade of polish. Doneit’s differentiator isn’t the task model — it’s the Assist layer and the depth of OS integration, particularly the new Siri onscreen awareness and the expanded Shortcut parameters. If your team is Apple-only and you want tasks to be capturable by voice mid-warehouse-walk, that’s a real edge.
Against Asana and ClickUp: those are team tools with dependencies, timelines, and workload views. Doneit is not competing there. If you’re coordinating a 15-person cross-border team across sourcing, listing, ads, and CS, Doneit won’t replace your project management stack. It might replace the personal task layer on top of it.
Against Notion: Notion is a database with a task-shaped UI. Doneit is a task app with an AI planner. Different philosophies entirely.
The honest positioning: Doneit 3.0 is a personal and small-team task manager that leans hard into Apple’s AI surface area. It’s not an ops platform.
The multimodal photo-to-task feature is the one to actually test
Of everything in the changelog, the photo-to-task capability is the one with real cross-border applicability, and it’s also the one a commenter immediately stress-tested. Gal Dayan, a maker at Dial, raised the exact objection I would have raised: “handwriting recognition is where these things usually fall apart in practice — messy shorthand, arrows, crossed-out items.” He then asked the sharpest question on the page: when Doneit Assist misreads something, does it show you what it “saw” before creating tasks, or do you only find out after a wrong task shows up?
That’s not a nitpick. It’s the entire risk model for AI-assisted capture in an operations context. If a supplier’s handwritten packing list gets misread as 400 units instead of 4,000, and that number flows into a restock decision, you’ve automated your way into a stockout. The maker did not answer that question in the thread — it’s still open as of the scrape.
Where the math breaks
Let me be concrete about the failure modes I’d expect, because this is where most AI task tools quietly die inside a real business:
- Handwriting variance. Chinese, Vietnamese, and Turkish supplier documents have wildly different character sets and numeral conventions. A model tuned on English whiteboard scrawl will degrade fast.
- Ambiguity without confirmation UI. If there’s no “here’s what I read, confirm” step, the tool creates false confidence. False confidence is worse than no automation.
- No audit trail. In ops, you need to know who created a task, from what source, and when. If Assist silently generates tasks, your accountability chain breaks.
- Language coverage. Not disclosed on the page. For cross-border teams, that’s the first spec I’d want.
What cross-border sellers can borrow from this launch
Even if you never install Doneit, there are three patterns here worth stealing for your own stack.
1. Treat capture as the bottleneck, not organization. Most ops teams over-invest in Notion templates and under-invest in making it trivially easy to get a task out of a photo, a voice note, or a Slack thread and into a system. Doneit’s bet is that multimodal capture is the unlock. I think that bet is directionally right, even if the execution is unproven.
2. Push tasks into the OS layer your team already lives in. The Siri onscreen awareness and expanded Shortcut parameters are the unsexy parts of this release, and they’re probably the most durable. If your ops lead can add a task by voice while walking a 3PL floor, that’s a workflow change, not a feature.
3. Ask vendors the confirmation question first. Any AI tool that writes into your operational system of record should be able to show you its interpretation before committing. Make that a procurement checkbox. If a vendor can’t answer it, that’s your answer.
A note on the release cadence
One thing I noticed scanning the full launch history: this team ships roughly once a year, in mid-to-late September. Doneit 2.8 landed September 20th, 2024 with 273 upvotes. Doneit 2.5 landed September 22nd, 2023 with 155. Doneit 2.2 landed September 14th, 2022 with 7.
That’s a small, consistent indie operation. Which cuts two ways. On one hand, you get focused, opinionated software with a real point of view. On the other, you get no enterprise roadmap, no SOC 2 chatter, no SLA, and no obvious answer to “what happens to my task data if this person stops shipping?” For a solo operator, fine. For a 30-person cross-border brand, that’s a real consideration before you let it touch supplier data.
Where my judgment says Doneit 3.0 falls short
Three things, in order of how much they’d matter to a cross-border operator.
First, the unanswered misread question. Gal Dayan’s comment sits there without a response. Until the maker explains the confirmation flow — or until someone posts a real test with a genuinely messy handwritten list — the headline feature is a demo, not a tool. I’d want to see a side-by-side: photo in, tasks out, with the intermediate interpretation visible.
Second, the ecosystem lock. Doneit is Apple-only by design. That’s a legitimate product choice, and it’s also a hard ceiling for cross-border teams. Your sourcing agent in Shenzhen is on Windows. Your 3PL contact is on Android. Your VA in the Philippines is on a Chromebook. If your task system can’t reach them, it becomes your personal task system, which is a much smaller value proposition. Compare that to Todoist, which runs everywhere, or ClickUp, which is aggressively cross-platform.
Third, no pricing or data-handling details on the page. Not disclosed. For a tool that wants to ingest photos of supplier documents — which may contain pricing, SKUs, and quantities — that’s a gap. If you’re evaluating this for anything beyond personal use, ask about data retention, model training on your inputs, and where inference happens. Those questions matter more in cross-border than in most contexts because your supplier documents are commercially sensitive.
The comparison I’d actually make
If you’re an Amazon FBA brand owner with an Apple-only team and a genuine problem with tasks falling through the cracks between your 3PL, your supplier, and your ads agency, Doneit 3.0 is worth a two-week trial. Run it against your current setup — probably Todoist or a Notion database — and specifically test the photo-to-task flow with real documents, not clean ones.
If you’re a Shopify DTC operator with a distributed team, this is almost certainly not your tool. Your work lives in Klaviyo, Shopify, and your ad platforms. Your task layer should be shared, cross-platform, and integrated with those. Doneit isn’t built for that.
If you’re a marketplace account manager juggling multiple seller accounts, the Siri and Shortcut improvements are the interesting bits, but you’d be better served by something that integrates directly with Amazon Seller Central workflows than by a general-purpose task app.
What I’d watch / test next
Three concrete things I’d do this week if I were evaluating Doneit 3.0 for a cross-border operation.
One: run the handwriting stress test yourself. Grab a real supplier packing list — ideally with crossed-out items, arrows, and non-Latin numerals — photograph it, and see what Doneit Assist produces. Then do it again with a whiteboard photo from your 3PL. Document where it misreads. If the tool shows you its interpretation before creating tasks, that’s a green flag. If it silently commits, that’s a red flag for anything tied to inventory or money.
Two: ask the maker the confirmation question directly. Gal Dayan’s comment is still unanswered. Post a follow-up on the launch thread asking specifically about the review-before-commit flow, language coverage, and whether inputs are used for model training. A responsive answer tells you something about the team. Silence tells you something too.
Three: pressure-test the Apple-only constraint against your actual team. List every person who touches your operational tasks — sourcing, 3PL, CS, ads, finance — and mark their device. If more than one is off-Apple, Doneit is a personal tool for you, not an ops tool for the business. Price that accordingly. Also check the full launch history to see whether the annual cadence matches your tolerance for a small-team dependency.
The broader lesson: AI task capture is coming to every productivity tool, and the ones that win in cross-border won’t be the prettiest. They’ll be the ones that show their work, handle messy multilingual inputs, and reach every device your team actually uses. Doneit 3.0 is a credible step in that direction. It’s not yet a system I’d let near my replenishment calendar.
