Marr Labs

What do they actually do

Marr Labs runs Vox, a managed AI voice‑agent service that makes and receives real phone calls for mortgage lenders. They configure the agent to a lender’s scripts, integrate with LOS/CRM/telephony, and operate it in production as a call‑center augmentation or replacement, with live deployments rather than demos (site, POC overview).

The product focuses on sub‑second speed‑to‑lead, warm transfers to human loan officers, and mortgage‑specific guardrails with a closed‑loop data setup so recordings and customer data stay in compliant environments; they also reference SOC 2 and lender‑grade observability (dashboards, recordings, KPI reports) (features, process/design post).

Commercially, Marr sells a paid POC ($3,000) that runs live calls—up to 1,000 during the trial—and then moves successful pilots into production with a setup fee and minimum monthly service. They cite production‑scale usage (e.g., “over 1,000,000 calls made”) and announce partnerships like Figure as evidence of enterprise engagement (POC details, homepage claims, Figure partnership).

Who are their target customer(s)

• Retail loan officers and small origination teams: They lose deals when leads aren’t contacted immediately and spend time manually qualifying callers instead of closing. They want instant engagement and warm transfers so they handle only qualified conversations (POC, site).
• Mortgage servicing teams (collections and borrower support): They run high‑volume, routine outreach under strict compliance and audit rules; manual calling is expensive and inconsistent. They need compliant scripts, recordings, and reliable execution (features).
• Enterprise lenders and mortgage operations leaders: They require SLA‑backed solutions that integrate with LOS/CRM/telephony and scale from pilot to production without heavy internal engineering (POC, news/partnerships).
• Marketing and growth teams managing lead pipelines: They struggle to re‑engage dormant leads and prove ROI because contact rates and response times vary. They need consistent speed‑to‑lead and conversion reporting from live trials (POC).
• Contact center and operations managers: They face staffing costs, turnover, and script adherence challenges. They want a managed service that runs calls with consistent qualification, recordings, and KPI dashboards (site).

How would they acquire their first 10, 50, and 100 customers

• First 10: Leverage founder/network intros to run tightly scoped, paid POCs ($3,000) where Marr handles the one‑hour kickoff, integrations, and first live calls to produce recordings and a KPI report quickly (POC). Offer light discounts or case‑study commitments to secure fast feedback and publishable results.
• First 50: Standardize the POC offer and build a small outbound SDR/AE motion targeting retail LO teams, regional banks, and servicers; run cohort POCs that emphasize speed‑to‑lead and conversion metrics. Pair with co‑sell partnerships through mortgage‑tech vendors to add enterprise credibility (POC, example partnership).
• First 100: Productize onboarding (templated integrations, configurable compliance, standardized reporting) to shrink setup time, and scale via channel partners (LOS/CRM/telephony vendors, BPOs, fintech resellers). Use a small enterprise team with SLA commitments and referenceable ROI to win larger lenders and expand into servicing workflows (features, news).

What is the rough total addressable market

Top-down context:

Initial focus is the US mortgage market—originators and servicers that rely on phone‑based lead engagement, qualification, scheduling, and borrower support. These organizations already staff or outsource call operations, creating a clear budget line for production voice automation.

Bottom-up calculation:

Approximate TAM for mortgage: ~1,800 banks/independent mortgage banks/large servicers at ~$100k ARR each (~$180M) plus ~10,000 retail branches/teams at ~$12k ARR each (~$120M), totaling roughly ~$300M in annual spend addressable by managed AI voice agents in US mortgage.

Assumptions:

• Counts: ~1,800 enterprise/mid‑market lenders and large servicers; ~10,000 retail branches/teams that could buy smaller footprints.
• Average ARR: ~$100k for larger lenders (multi‑workflow deployments); ~$12k for smaller teams (lead engagement/qualification).
• Scope limited to US mortgage workflows; excludes adjacent verticals and non‑phone channels.

Who are some of their notable competitors

• Replicant: Autonomous contact‑center voice AI for enterprises; notable for production deployments across industries and focus on fully handled calls.
• PolyAI: Enterprise voice assistants for customer service that handle natural, open‑ended calls; strong emphasis on call containment and brand‑specific tuning.
• Kore.ai: Conversational AI platform with voice and digital virtual assistants used by large banks and contact centers; broad tooling and enterprise integrations.
• Skit.ai: Voice AI for collections and customer support in BFSI; relevant to servicing and payment reminder workflows.
• Retell AI: API‑first real‑time AI phone agents; developer‑oriented platform competing on speed and call quality for programmable voice agents.