Built by a founder, for founders.
Overwise was built by Tobias Duelli, the founder of Bookingbird and Gleap, to do his own outbound. Every feature exists because he needed it on a real Tuesday night between shipping product. Trust-first, founder-priced, no demo gate, no per-seat tax.
I built Overwise because I do my own outbound.
I'm Tobias. I run Bookingbird and Gleap, and I do outbound for both myself — late at night, between shipping product. Apollo couldn't find the leads I needed. Lemlist would send them, but didn't know who they were. Clay could glue it all together, but cost me a weekend per campaign.
So I built the agent I wanted. It finds leads in segments where everyone else's database is empty. It writes in my voice. It stops on reply and only pings me when someone really wants to talk. It's the tool I trust to send under my name — to my own prospects, on my own domain.
If something doesn't feel right, email me directly. Same address shipped on every Overwise account.
Four principles, everything else flows from these.
If a feature doesn't strengthen a trust loop, it's noise. Buyers won't churn on missing features they don't know they want — they'll churn on a single bad email going out under their name.
Review-each-send is the default for new mailboxes. Mailbox warmup forces conservative caps for the first 7 days. Bounce auto-pause kicks in at 8%. None of these are negotiable.
Listed prices, self-serve signup, card on file. No 'contact sales' gating. The buyer is a founder doing their own outbound — they need the friction-free path.
On every /vs/* page, there's a section telling you when the competitor is the better choice. Long-term, honesty wins; trash-talking competitors looks weak and triggers their brand searches.
The agent, not the wrapper.
Overwise's outreach loop is an agentic system, not a fixed sequence. The lead agent (Sonnet) decides per-lead what to send and on which channel; an advisor (Opus) reviews edge cases. The MessageVerifier discards drafts whose claims can't be cited from the lead's actual signals. This isn't ChatGPT-with-a-prompt — it ships with cost budgets, dry-run gates, and the agentic decision log you can inspect per lead.
- Discovery: Apify scraping (Google Maps, Instagram, business websites) with absence-signal probes you configure per project. Apollo for known-on-LinkedIn segments.
- Drafting: Claude Sonnet 4.6 with prompt-cache, per-mailbox brand-voice extraction from your sent folder, MessageVerifier cite-or-discard, EU AI-Act footer.
- Sending: Postmark for transactional, BYO Gmail/Outlook via OAuth for outreach (no password storage). SPF/DKIM/DMARC checks block sending from broken domains.
- Reply triage: ReplyClassifier per inbound — positive → AI Inbox approval; question → AI-drafted answer queued; negative → suppression + polite acknowledgement.
- Storage: MongoDB Atlas EU-Central (Frankfurt). AES-256 at rest, TLS 1.3 in transit. Data wipe within 30 days of cancellation.
Your first leads, in 5 minutes.
14-day trial. Card on file, hard-stop at your lead cap (no surprise bills). All trust artifacts standard. Cancel any time without an email.
Start 14-day trial — first leads in 5 min → Card on file · No demo gate · Cancel any time