Teamday
← People

Aman Singh

AI Platform Lead at Arca

agentsenterprisefinancearca

About Aman Singh

Aman Singh leads the AI platform team at Arca, an AI-native wealth management firm that offers financial advice, planning, tax, insurance and estate planning. Arca raised $64 million across seed and Series A rounds. Singh's team builds the model-agnostic harness and capabilities Arca uses to run its business with AI.

In a Latent Space interview at AI Engineer New York, Singh describes how agents handle the work around human advisors. Agents analyze client finances, track multi-day processes, bring in the right person at the right step, and draft plans that humans approve before any money moves.

Career Highlights

  • Leads the AI platform team at Arca
  • Speaker at AI Engineer New York (2026) on agents for long-horizon financial work

Notable Positions

On Being an AI-Native Firm, Not a SaaS Vendor

Singh argues that owning the business lets Arca standardize how work is done and require staff feedback on agent output. A firm with AI tools waits for people to prompt agents. At Arca, agents prompt people. See Human-in-the-Loop.

On Owning the Harness

Arca uses frontier models, with Claude as its main model, but keeps its own harness instead of a lab's managed agents product. Singh's reason is that the memory and operating knowledge agents build up are what make the business different. See Agent Harness.

On Deterministic Guardrails

Safety rules are deterministic and placed "at the edges of capability": LLM checks on every database write, prompt injection detection on incoming data, PII and exfiltration checks on outgoing data, and compliance checks on client documents. See Prompt Injection.

Key Quotes

  • "Agents prompt people more than people prompt agents." (on Arca's operating model)
  • "We're actually not a SaaS company. We are the firm ourselves." (on AI-native services)
  • "All of our safety restrictions are fully deterministic." (on guardrails)

Related People

noyan tokgozoglujohn feminella