Notes on outbound infrastructure.
How we think about sequencing, deliverability, multi-channel outbound, and the plays we build with growth teams.
We benchmarked agent fanout against our SDK pipeline. The SDK ran 1,165% faster.
Same task, same SDK calls, same AI model. One finished in 31 seconds with zero retries. The other took 392 seconds and needed 4 retries across 10 subagents.
GTM as code: stop rebuilding outbound lists by hand
Instead of rebuilding filters and CSVs every week, define the audience once, run it as a play, and let Claude Code adjust it when the market changes.
Why the Verbiflow SDK comes before a Verbiflow agent
Customers ask when we're shipping a Verbiflow agent. We built the MCP and SDK first so Claude Code, Cursor, Codex, and eventually our own agent can build outbound and custom plays on top of the same Verbiflow infrastructure.
The hard part starts after enrichment
Apollo and Clay can give you a list. The hard part is making it safe to send: CRM suppression, sequence dedupe, healthy mailboxes, one reply inbox, and clean reporting.
Outbound is an infrastructure problem, not a tool problem
Every growth team has the same five-tool outbound stack glued together by Zaps. The problem isn't any one tool. It's that outbound has become infrastructure, and everyone is still buying it like software.
Outbound at pre-seed: it's not a volume game, it's a customer discovery game
If you're a pre-seed founder, the goal of cold outbound isn't pipeline. It's learning whether your product solves a real problem. The math is different, the metric is different, the setup is different.
Verbiflow vs Nooks: when the parallel dialer hits its ceiling
Nooks made parallel dialing the default for high-volume SDR teams. In 2026, the connect-rate decay and the missing email + LinkedIn layer are what teams notice. Here's where it breaks down.
Outbound reporting breaks when your CRM is missing the work
Most sales leaders inherit a stack where the sequencer and the CRM disagree. Here's what changes when every send, reply, and call attempt is written back to the CRM.
Verbiflow vs Lemlist: when the personalization category stops paying off
Lemlist invented dynamic-image personalization. In 2026, that same trick is the one most likely to land your first email in spam. Here's where the numbers actually land.
Verbiflow vs Instantly: when your team needs visibility, control, and CRM sync
Instantly is a strong email sender. Once your team grows and the CRM has to hold the full history, the gaps show up: who sent what, who replied, and where it lives.
How we found every regional security-guard firm in the US (and surfaced the ones losing customers)
A customer selling into physical-security tech needed a list of regional guard firms. No vendor sells one. We built it from Census data, Google Maps, and Claude, then surfaced which firms had the most negative reviews so the customer sequenced the right ones first.
We pulled 1,600 SaaS companies from a funding DB and sorted them by what they actually sell
A walkthrough of one of our customer plays: crawling B2B SaaS sites, extracting their actual customers and category, ranking them by ICP fit, and sequencing the survivors.
Trust Atlas: how we read every YC company's trust page to find compliance gaps
A customer selling compliance tooling needed to find startups that hadn't yet hit SOC 2. We read the trust pages of every company in the last few YC batches, mapped certifications and gaps, and ranked the misses by buying-signal strength.
Verbiflow vs HeyReach: when one-channel LinkedIn breaks down
HeyReach is a great LinkedIn-only sequencer. The problem is that LinkedIn-only is the wrong unit of outbound. Here's where it breaks down.