Skip to content
AI.info

The Pulse

Synthflow Launches Vector, an AI Agent That Tracks Lending Accounts

Synthflow launched Vector, an AI agent that follows consumer lending accounts across calls, texts and email. The company says a four-week collections trial reached borrowers more often than the lender’s human team, while keeping underwritin

Synthflow Launches Vector, an AI Agent That Tracks Lending Accounts

AI.info Team · Written with AI · fact-checked ·

Synthflow says its new lending agent should keep working after a borrower hangs up; its own product page, however, draws a firm boundary around decisions that remain with staff. The Berlin company launched Vector on October 28 as an agent that follows consumer loan accounts across calls, text messages and email, while routing underwriting judgments and hardship cases to people.

That split defines the product: Vector is meant to carry routine account work forward, not to make every lending decision itself. “What lenders actually need is an agent that owns the outcome,” Synthflow co-founder and CEO Hakob Astabatsyan said in the launch announcement. Astabatsyan, as quoted in the Business Wire release carried by VentureBeat

Vector follows accounts across days and channels

Synthflow describes Vector as a “long horizon” agent: a lender assigns an account and a target outcome, and the system carries forward the borrower’s history, intent and next steps between interactions. It can call, text or email, schedule follow-ups and resume work where an earlier contact ended. The product page presents collections as live with lenders, while applications, refinancing and servicing appear as additional use cases to come.

For applications, the company says Vector can request missing documents and prepare a file for a decision, but underwriting decisions and manual reviews go to staff. In collections, it can pursue a payment or plan within lender-configured rules; the company says hardship cases are paused and handed to a team rather than assessed by the agent. Vector also discloses that it is AI at the start of each conversation, according to the product page.

A lender trial supplies the headline comparison

Synthflow reports that in a four-week deployment with a North American consumer lender, Vector reached the right borrower on 4.9% of dialing attempts, compared with 3.5% for the lender’s collections team. The company describes that as a 40% improvement and says the human benchmark came from the lender’s top-performing collectors working the same accounts. On a separate measure, promises to pay per account were 3.9% for Vector and 2.1% for the team.

The launch announcement adds that 49% of conversations ended in a promise to pay and that those commitments totaled nearly $1 million. Those figures are company-reported results from one deployment, not an independently published evaluation. Synthflow’s product page also presents results from a Latin American bank, including a 23.5% rate of verified borrowers committing to pay the full overdue amount within seven days; the company says those were full-balance commitments, not partial plans.

Rules constrain the agent’s next move

Synthflow says Vector’s policy engine checks actions before calls or messages go out, including contact windows, attempt limits, consent and permissions. The company says it runs around 70 checks before an action and blocks or escalates when a required check fails. It also says each decision is recorded, giving lenders a mission record of the action, applicable policy and checks.

The product page lists SOC 2 Type II and ISO/IEC 27001:2022, and describes US and EU regional deployments, with German data residency available. It says pricing is tied to agreed outcomes rather than minutes. Vector’s launch announcement says the agent is available to consumer lenders in the US, South America, the UK and the EU.

Human judgment remains the dividing line

Synthflow is selling continuity across routine tasks, but its own product description stops short of handing the full account lifecycle to software. Staff retain decisions on underwriting, exceptions and hardship, while borrowers can ask for a person and issues outside configured permissions are escalated. The initial product evidence centers on collections: Vector is live there, and the company’s published US comparison reports 4.9% right-party contact per dial against 3.5% for the lender’s collectors.

Sources

Explore

More articles