I asked AI tools to do my actual work for 2 weeks. Most failed at step one.
I've been an AI skeptic inside my own AI company. Awkward place to stand.
For the past two weeks I gave specific AI tools one job each. Not "help me with X." Actual tasks with a measurable output. A draft that needed to go out. A report that needed filing. A piece of copy that needed to convert.
Here's what I found, and it'll probably annoy the AI hype crowd.
7 out of 12 tasks required more time with AI than without.
Not because AI is bad. Because most AI tools are optimised for conversation, not for completion.
The gap between "generates decent text" and "does the actual thing in the actual app in the way you actually need" is enormous. Almost nobody is shipping across that gap yet.
The 5 that worked had one thing in common. They didn't just generate something and hand it back. They completed the action in the context where the action lived.
That's the distinction that matters. A tool that writes an email is not the same as a tool that writes and sends the email. Sounds obvious. Most tools stop at step one and call it AI.
This is actually why I built Genie 007 (genie007.com). Not to make a smarter text generator. To close the gap between "here's your draft" and "the thing is done." Voice-to-action, not voice-to-text. You say what you want, it does it everywhere on your computer. No integrations, no setup.
The 2 weeks taught me something I didn't expect: the useful AI tools right now are not the ones with the best models. They're the ones where someone thought hard about where the task actually lives and built all the way to done.
What AI tools have actually completed tasks for you, not just helped with them?
Building Genie 007 (genie007.com) — voice-to-action for everything you do on a computer.
For coding, the line I've noticed: tools that verify their own work complete tasks, tools that don't just chat about them. An agent that writes code AND runs the tests, reads the failures, and iterates will finish the job. One that just generates a diff leaves me holding the bag. Same model, totally different outcome depending on whether the loop is closed.
Clear and practical, thanks. Did anything surprise you along the way?
How did you decide this was worth building in the first place?
Interesting take. Would you still recommend this approach to someone starting today?
Appreciate the honesty here, most people only share the wins.
Interesting take. Would you still recommend this approach to someone starting today?
Appreciate the honesty here, most people only share the wins.
Clear and practical, thanks. Did anything surprise you along the way?
How did you decide this was worth building in the first place?
Appreciate the honesty here, most people only share the wins.
How did you decide this was worth building in the first place?
Interesting take. Would you still recommend this approach to someone starting today?
Clear and practical, thanks. Did anything surprise you along the way?
Appreciate the honesty here, most people only share the wins.
Clear and practical, thanks. Did anything surprise you along the way?
How did you decide this was worth building in the first place?
Interesting take. Would you still recommend this approach to someone starting today?
Interesting approach. What was the hardest part to get right?
Solid lesson. Which channel has worked best for you so far?
What made you pick this stack over the alternatives?
This is useful. How are you finding your first users so far?
Interesting approach. What was the hardest part to get right?
Good write-up. What would you do differently if you started again?
Good point. Did you test that with users before committing to it?
Nice progress. What is the next thing you are focusing on?
Good point. Did you test that with users before committing to it?
Appreciate the honesty here, most people only share the wins.
Clear and practical, thanks. Did anything surprise you along the way?
Interesting take. Would you still recommend this approach to someone starting today?
How did you decide this was worth building in the first place?
Clear and practical, thanks. Did anything surprise you along the way?
Interesting take. Would you still recommend this approach to someone starting today?
How did you decide this was worth building in the first place?
Appreciate the honesty here, most people only share the wins.
How did you decide this was worth building in the first place?
Appreciate the honesty here, most people only share the wins.
Interesting take. Would you still recommend this approach to someone starting today?
Clear and practical, thanks. Did anything surprise you along the way?
Nice progress. What is the next thing you are focusing on?
What made you pick this stack over the alternatives?
Clear and practical, thanks. Did anything surprise you along the way?
How did you decide this was worth building in the first place?
Thanks for writing this up. Bookmarking it for later.
Nice progress. What is the next thing you are focusing on?
Appreciate the honesty here, most people only share the wins.
Totally agree with the “where the task lives” point. For writing, the failure mode I keep seeing is forcing people to leave Mail/Notion/Docs, generate text elsewhere, then paste/edit it back in. The better UX is assistance inside the field: autocomplete when you are already composing, rewrite selected text in place, and keep sensitive drafts local when possible.
I’m working on DraftKey for that exact Mac-native/on-device slice: https://www.getdraftkey.com/ . It does not try to complete the whole workflow, but it removes a bunch of the copy/paste/context switching around founder writing.
How did you decide this was worth building in the first place?
Appreciate the honesty here, most people only share the wins.
Interesting take. Would you still recommend this approach to someone starting today?
Clear and practical, thanks. Did anything surprise you along the way?
Clear and practical, thanks. Did anything surprise you along the way?
How did you decide this was worth building in the first place?
Interesting take. Would you still recommend this approach to someone starting today?
Appreciate the honesty here, most people only share the wins.
Really solid approach — I'm juggling something similar myself (building Xstream4K on the side), what's been the hardest part for you so far?