Straightline

The Problem

In real estate, the fastest agent wins. The average agent takes 90 minutes to respond to an inbound lead — despite research showing that waiting just 5 to 10 minutes causes a 400% decrease in the odds of qualifying that lead (Oldroyd & Elkington, Harvard Business Review). 74% of buyers hire the first agent who responds. Our primary research across 51 agent interviews found that slow follow-up costs the average agent an estimated $600,000 per year in lost commissions.

Agents know this. They just can't fix it. They're at showings, driving between appointments, coaching their kid's soccer game. By the time they get back to a lead, it's gone.

What We Built

Straightline is an AI lead response system for real estate agents. A lead arrives — from Zillow, MLS, a personal website, anywhere — and Straightline drafts a personalized reply in under 60 seconds. The agent gets a push notification, reads the draft, and approves it with one tap. The reply sends as a text message, seamlessly, from the agent's number. The buyer never knows AI was involved.

The system learns over time. Every approval and edit teaches Straightline the agent's tone, phrasing, and style — not a generic voice, that specific agent's voice. Approval rates improve from roughly 60% on day one to 95% over time.

This isn't a CRM. It doesn't replace the agent. It inserts itself into the critical 60-second window that determines whether a lead converts — and it does it invisibly.

How We Built It

The core stack is intentionally lean:

• Claude API (Anthropic) — draft generation. Each draft is personalized to the specific property, the buyer's inquiry text, and the agent's accumulated style profile
• Twilio SMS — lead ingestion and reply delivery. Leads text a dedicated number; Twilio routes them to Straightline; approved replies send back as texts
• One-tap approval interface — mobile web app, built for speed. The entire approval flow takes under 15 seconds
• Deployed in under 2 weeks with infrastructure costs under $200/month

What We Learned

The biggest insight wasn't technical — it was behavioral. In 51 interviews with real estate agents across 8 states, the single most consistent finding was this: 38 out of 46 agents who expressed interest required human approval before anything sends. Not because they don't trust AI, but because their professional reputation is on the line with every message. The product only works if agents trust it enough to build a daily habit around it.

That shaped every product decision. The approval step isn't a limitation — it's the feature. It's what separates Straightline from the AI auto-bots that buyers have already learned to ignore.

We also learned that the $4,000/month ISA (Inside Sales Agent) is the real competitive frame. Agents aren't comparing us to software — they're comparing us to the human they can't afford to hire.

Challenges

The hardest challenge was prompt engineering for voice consistency. A real estate agent's communication style is deeply personal — some are warm and conversational, some are crisp and professional, some lead with neighborhood knowledge. Getting the AI to draft in a voice that feels like that agent rather than a generic assistant required significant iteration on how we structure the style profile from approval history.

The second challenge was trust. Agents have been burned by automation tools that sent embarrassing messages to buyers. Building conviction that Straightline would never send anything without explicit approval — and making that viscerally clear in the UX — was as important as anything technical.

What's Next

We're testing three assumptions: whether 10%+ of trial users convert to paid, whether $249/month behavioral willingness-to-pay holds when a credit card is required, and whether the daily approval habit forms (signaled by agents proactively forwarding leads without being prompted).

Our target: $20M ARR by Year 6 at 80% gross margin, with a likely exit to Zillow, kvCORE, or HubSpot at a 6× revenue multiple — consistent with Zillow's $400M acquisition of Follow Up Boss in December 2023.

Built With

• claude
• twilio