CORNERSTONE PLAYBOOK

Qualify every lead in 60 seconds. Route the hot ones to a human. Recycle the cold ones to a sequence.

Inbound triage agent that qualifies, scores, and routes every lead within 60 seconds — over voice, SMS, chat, or form. Speed-to-lead is the single biggest predictor of close rate. Most teams take 3–8 hours. The trained agent takes 60 seconds — and only escalates the leads that are actually worth a sales rep's time.

The problem, precisely

Speed-to-lead studies are unambiguous: contacting a lead within 5 minutes makes you 9× more likely to close versus 30 minutes, and 21× more likely versus 24 hours. Yet the median B2B response time is between 42 and 47 hours. The reason isn't laziness — it's that leads come in 24/7 and humans don't, leads come in unqualified and reps don't want to waste calls, and the qualifying questions (budget, timeline, decision-maker, fit) are tedious to ask but indispensable to ask. The compromise most teams settle for is a sloppy version of the worst of both worlds: an auto-responder email ('thanks, we'll be in touch!'), a Slack ping to a sales rep, and a manual call back the next day with no qualification done. By then half the leads have bought from someone else.

How we build it

The qualification agent is built as a stateful conversation engine on Anthropic's API with three layers. The pre-check layer (Haiku, fast and cheap) does enrichment in parallel — Apollo or Clearbit for firmographic data, your CRM for dedup and history, optional LinkedIn data via Phantombuster, optional reverse-IP intent data via Albacross. By the time the conversation starts, the agent already knows the lead's company size, industry, tech stack, and whether they've been in your CRM before. The reasoning layer (Sonnet) runs the conversation — over voice (Deepgram + Cartesia), chat widget (Crisp / Intercom-compatible embed), SMS (Twilio), or form-fill (asynchronous follow-up email).

What you get

B2B SaaS case: lead-to-meeting conversion went from 11% to 28% by responding in 47 seconds instead of 5 hours.

A worked example

A Series-A vertical SaaS company moved from 11% to 28% lead-to-meeting conversion within 6 weeks. Their SDRs went from 80 dial-attempts/day on unqualified leads to 22 dials/day on pre-qualified hand-offs. Pipeline coverage doubled with zero sales hire.

Cost and timeline

Starts at $5,000 build. Ongoing: $599/mo + enrichment pass-through ($0.10–$0.40/lead depending on data sources). Ships in 21 days. Delivered as AI Lead Qualification Software, available across United States, Canada, UK, EU, AU.

How does it score leads?

We start with a default rubric (firmographic fit, BANT signals, behavioral signals like which pages they viewed) and then recalibrate against your closed-won/lost data from the last 90 days. After 30 days live we recalibrate again on the new data. Scoring is interpretable — every score comes with a 1-line explanation, so reps trust it instead of fighting it.

What if a hot lead refuses to talk to the agent and demands a human?

Immediate handoff. The agent recognizes the pattern ('let me talk to a real person', 'I don't want to chat with an AI', frustrated tone) and routes to whichever live channel you've configured — phone, Slack channel, on-call SDR rotation. We instrument escalation rate and review it weekly.

Can it qualify across multiple products / business lines?

Yes. You define the product taxonomy and the agent classifies inbound demand into the right line, applies the right qualification rubric, and routes to the right rep team. Useful for agencies and multi-product SaaS.