Stitch has launched Stitch AI, building AI agents into its banking operating system so institutions can deploy them within existing controls.
Saudi Arabia-based Stitch provides an operating system for banks and financial institutions covering lending, cards, payments, and wallets. Stitch AI is built natively into the platform and offers three components: a copilot for bank staff, ready-made Stitch Agents for everyday operations, and an Agent Harness framework for building custom agents, all operating within the limits each bank sets and escalating for approval beyond them.
Copilot, agents, and custom builds
Using the Copilot, staff can retrieve account and customer data, set up products such as accounts, cards, and loans, including limits, pricing, fees, and eligibility, and update workflows using plain-language instructions. Stitch Agents handle tasks including onboarding, credit decisioning, and collections within the bank’s existing workflows, permissions, and rules, while the Agent Harness lets banks build, test, and deploy their own agents with permissions, approvals, guardrails, and audit trails.
According to Stitch, most AI in banking sits on top of legacy systems and works on incomplete copies of data, unable to execute tasks or enforce approvals. Stitch AI agents instead act directly on the core system through the same services the bank’s products use, without copying data or maintaining separate connectors. Each agent has its own identity, scoped permissions, hard limits, maker-checker controls, and a full audit trail.
Data protection and model choice
Sensitive data, including ID numbers, account numbers, phone numbers, and addresses, is masked before any AI model processes it, and every question, action, and approval is logged with its reasoning for risk and compliance teams. Banks can choose where Stitch AI runs, including in-country or in a private cloud, select their preferred model, and switch models without rebuilding, while keeping control of their data.
Mohamed Oueida, CEO of Stitch, stated that most AI projects in banks stall at the proof-of-concept stage because underlying systems cannot support them, and because regulated institutions will not let an agent act without an owner and an approval. According to Oueida, Stitch addresses both by building AI into its operating system under the bank’s controls. Founded in 2022, Stitch is backed by investors including Andreessen Horowitz.