What Your Intern Can (and Can't) Do
Honesty builds trust. Here's exactly where they excel, where they need guidance, and what they can't do yet.
We'd rather be honest than impressive
Most AI product pages read like a list of superpowers. We think you deserve a clearer picture — especially if you're going to trust an assistant with real work.
Here's what they do well, where they need your help, and what they can't do yet.
Where they excel
Follow-ups and outreach. They track every open thread, draft contextual follow-up emails, and nudge you when something's been sitting too long. This is the single task our early users cite most — the one that used to fall through the cracks.
Meeting prep and summaries. Before your calls, they pull relevant context: past conversations with that contact, open action items, recent activity. After, they capture decisions and follow-ups. No more "what did we agree on?"
Research. Competitor analysis, market research, background checks on potential partners — they handle the digging so you get the synthesis. It's not just Googling; they cross-reference what they know about your business.
Recurring admin. Weekly reports, CRM updates, invoice follow-ups, status emails. The stuff you keep doing manually because no tool quite automates it. They handle the pattern once you show them the first time.
Email drafting. In your voice, with your context. They learn how you write and adapt. Early users report that their drafts need fewer edits after the first week than they did on day one.
Where they need guidance
Creative strategy. They can draft a newsletter or blog post, but they won't invent your brand voice from scratch. Give them examples of what you like, and they get remarkably close. But the creative direction still needs to come from you.
Ambiguous priorities. If you have 12 things to do and haven't told them which ones matter most, they'll ask. That's by design — we'd rather they ask than guess wrong.
Complex negotiations. They can prep your talking points and draft follow-up proposals, but they won't negotiate a deal on your behalf. Anything with stakes that high still needs your judgment.
What they can't do yet
Phone calls. They work in text — Slack, email, docs. Voice capabilities are on their roadmap but not available today.
Physical tasks. Obviously. But worth saying: they can't mail packages, attend events, or pick up your dry cleaning.
Deep domain expertise. They're a generalist operator. If you need specialized legal, medical, or financial analysis, you should still consult a professional. They can find and organize the information, but they shouldn't be your only source for high-stakes decisions.
Our philosophy
We believe the most trustworthy AI products are the ones that tell you what they can't do. They are useful today and getting sharper every week. But they're not magic, and we won't pretend they are.
If you're looking for a resident assistant who handles the recurring operational work that's important but not strategic, they are probably a good fit. If you need a chatbot that tells you it can do everything — there are plenty of those. We're building something different.
Next article
Why We Built Intern