Locus

What do they actually do

Locus provides a policy engine and tooling that lets autonomous agents access and spend funds with guardrails. Teams give an agent an identity, attach rules like budgets, per‑transaction caps, and required approvals or justifications, and Locus enforces those rules while recording who paid what and why. They are in a closed beta with a live developer sandbox; the initial live rail is USDC on Base, with non‑custodial wallet flows and direct‑to‑wallet/email payouts supported today (homepage, YC company page).

They’ve published a reference implementation and demo code for the Agentic Commerce Protocol (ACP), and showcase workflows like a micro‑task platform where an agent evaluates submissions and automatically pays contributors within preset limits while maintaining an audit trail (GitHub demo, YC company page).

Who are their target customer(s)

• Internal automation teams (IT/ops) building bots for procurement, reimbursements, or invoice approvals.: They can’t safely give bots corporate cards or bank access; mistakes or runaway scripts can overspend, and finance needs clear attribution for who approved each payment.
• SaaS platforms adding agent features (tasking apps, no‑code agent builders).: They need an embeddable way for agents to pay users or vendors without owning full payments risk or forcing manual approvals that slow user experience.
• Marketplaces and micro‑task platforms with automatic contributor payouts.: Manual payout queues limit scale; errors or duplicates cause disputes and erode seller trust. They need controlled, near‑real‑time payouts with auditability.
• Finance and compliance teams at growing companies evaluating agent automation.: They require enforceable controls (spending limits, separation of duties) and audit logs that reconcile agent‑driven payments with policies and ledgers.
• Startups building agent‑first products that must add payments.: Building secure payment flows, permissions, and audit from scratch is a distraction and introduces legal and operational risk.

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

• First 10: Convert active builders from the demo repo, Discord, hackathons, and YC/Web3 networks with small USDC credits and hands‑on onboarding; offer 1–2 hour integration support and turn each pilot into a short case study.
• First 50: Run a second wave of themed hackathons, publish ACP/OpenAI‑style integration guides, and do targeted outreach to product leads at tasking platforms and internal automation teams with low‑friction pilot contracts; add one sales/partnership hire to close early platform deals.
• First 100: Form channel partnerships with no‑code agent builders, payroll/payout platforms, and one custody/fiat partner for ACH/wires and white‑label embeds; run 30–90 day enterprise pilots with onboarding support and clear success metrics to convert to paid contracts.

What is the rough total addressable market

Top-down context:

Locus maps to embedded finance and finance automation/spend control rather than total payment volume. Recent estimates put embedded finance around USD 108.6B in 2024 and AP automation around USD 3.1B; broader payments revenue pools are ~USD trillions (context only) (IMARC, Grand View, McKinsey).

Bottom-up calculation:

Summing embedded finance (~$108.6B) and AP automation (~$3.1B) yields an addressable software/platform pool of roughly ~$111.7B. Illustratively, 0.1% capture ≈ ~$112M ARR and 1% ≈ ~$1.12B, assuming platform/subscription plus per‑transaction fees (IMARC, Grand View).

Assumptions:

• TAM is based on software/platform revenue pools (embedded finance + AP automation), not total payment volume or total fee pools.
• Monetization blends subscription/platform fees with per‑payout fees; capture scenarios are illustrative, not forecasts.
• Near‑term SOM is smaller: agent builders and early platform adopters while fiat rails and enterprise integrations mature.

Who are some of their notable competitors

• Stripe: A full payments platform (Connect, Issuing, Treasury) for accounts, payouts, and cards; broad coverage but not purpose‑built for agent‑centric policy/audit layers (Issuing/Connect docs, Financial Accounts).
• Marqeta: Card‑issuing APIs with real‑time card controls and authorization; strong for guarded card programs but focused on card rails rather than cross‑rail agent identity and justification/audit workflows (Docs).
• Ramp: Spend‑management with virtual cards, limits, and APIs; optimized for human/team spend and finance workflows, not autonomous agent decisioning with an embedded policy engine (Developer docs).
• Safe (formerly Gnosis Safe): On‑chain smart‑contract wallet with modules and transaction services for policy/multisig; useful for locking down on‑chain funds but lacks fiat rails and higher‑level approval/justification/audit flows Locus targets (Docs).
• Unit: Banking‑as‑a‑service APIs for accounts, ACH/wires, and cards; solves fiat custody/rails but does not provide an out‑of‑the‑box agent policy engine with budgets, approvals, identities, and audits (Docs).