About This Offer

Cognition is a U.S. AI company and the maker of Devin, the world's first autonomous AI software engineer. Devin works inside the U.S. government and hundreds of enterprises, among them Goldman Sachs, Citi, Santander, Itaú, HSBC and Nubank, where it writes, tests, migrates and maintains production software as a member of the engineering team. Cognition's contribution to Trade Over Aid is not a product license or compute. It is a workforce capacity expanding program: upskilling government and industry teams in participating countries to work with AI agents as everyday collaborators. The constraint on digital modernization in most economies is not access to AI models. It is skilled capacity. National workforces need to scale faster than traditional training can produce graduates, curricula lag the pace of change in the industry, and reliance on external contractors and consultants leaves governments and companies without the ability to maintain what was built for them. Cognition's program addresses this directly by making participants agent-native: engineers who direct autonomous AI agents the way a lead engineer directs a team, rather than typing every line themselves. The curriculum is practical and hands-on, delivered by Cognition's forward-deployed engineering team, and covers:

• Delegating real work to AI agents. Scoping a task so an autonomous agent can complete it end to end, from a legacy migration to a security fix to a new citizen-facing service, and knowing which work to hand off and which to keep.
• Reviewing and governing agent output. Reading, testing and approving code an agent produced, setting guardrails, and keeping humans accountable for every change that ships.
• Turning one-off tasks into repeatable workflows. Building reusable playbooks so that recurring work (vulnerability backlogs, documentation of undocumented systems, data pipeline maintenance, reporting) runs with minimal supervision.
• Leading an agent-native team. For technology managers and ministry CIOs: how to plan, staff and measure engineering work when part of the team is AI.

This is the same enablement Cognition runs for its enterprise customers. Every deployment starts with Cognition's forward-deployed engineers working alongside the customer's team on their own codebase: picking a first use case, teaching engineers to scope and review agent work, and codifying what they learn into playbooks the whole organization reuses. Teams at Nubank, Ramp and other large organizations went from first pilot to agents contributing production code at scale this way, and that operating knowledge is what the program transfers to participating countries. Participants practice on their own systems, not toy examples, so the program produces working results for the organization during training rather than after it. Cognition's existing enablement program, including live sessions and office hours with the forward-deployed team, forms the backbone. The economic case is direct. Cognition's enterprise customers report 6 to 12x increases in engineering capacity from the same headcount. Spread across a national workforce, that changes what a country can build and how fast. Governments spend less on outside contractors and deliver trade-enabling systems such as customs, business registration, licensing and payments faster. Local companies can take on software work for international clients that was previously out of reach. Graduates become job-ready in months rather than years. That is how skill contributions turn into higher economic output for member nations: more capability per person, owned locally, and compounding after the program ends.

Limitations

None