janyl

A line drawing of two colleagues seated side by side at a table, leaning over the same open document. One of them points at a page, the other writes on it. A desk lamp stands at the end of the table, and a single horizontal line runs behind them from edge to edge.

Forward Deployed Engineering

The first step in any janyl deployment is one of our engineers embedding herself or himself in your company starts the agent off in its learning process. This is called Forward Deployment Engineering (FDE).

First, a bit about the history of the role, as it will help you understand why it’s essential to our process. The role of Forward Deployed Engineer was invented by Palantir in the mid-2000s to put its engineers inside the customer’s building, writing software against the customer’s own data instead of shipping a product over the wall and hoping it fit. Since then, in the AI-era, companies like OpenAI, Anthropic, Microsoft and Amazon have all created FDE arms. These engineers embed themselves in organizations to find where the work actually is, redesign the workflow around it, and leave behind something that keeps running after they have gone. Some well-known collaborations have included Airbus, whose Skywise platform grew out of a Palantir deployment on the A350 production line; LSEG, Unilever and Novo Nordisk, named as customers of Microsoft’s deployment arm; and the NFL, the NBA and Southwest Airlines, named as customers of Amazon’s.

Our janyl process takes many cues from these established players, such as the embed itself, which puts an engineer in your office instead of a ticket queue somewhere else; a fixed delivery window, which for us is twenty-four days; and a short working session before anybody builds anything. What we don’t share, however, is their price. None of the four AI companies above publishes one. Palantir does, on the framework it sells to the British government through, and its May 2024 list reads: £150,000 per person per quarter for implementation and engineering, £40,000 for a two-week design sprint, and £6,000 to £20,000 a month afterwards to keep a single use case alive. The software those engineers deploy is £3,000,000 a year for one organization. Anthropic publishes the other half of the picture on its own job board, where a forward deployed engineer is advertised at $280,000 to $320,000 a year. It is a price for a company that can put a seven-figure line in its budget for one project.

So how do we keep costs low? At janyl, a large part of the FDE process is offloaded to our own internal AI agents. For example, reading everything your company has already written down, the handbooks, the shared drives, the three years of group chat, and coming back with a list of the jobs an agent could take. Drafting the first version of your agent’s instructions, which our engineer then corrects rather than composes. Running your agent against the week that has just finished, so that its mistakes happen where they cost nothing, before it ever touches the week that is coming.

The advantage of this is twofold. Our AI agents verifiably excel beyond human engineers in the tasks that are confided to them, providing superior quality in addition to saving cost. And the same agentic stack that we use to build our own internal agents is the one we are deploying into your business as well.

When you work with a janyl FDE, you’ll feel like you have a new teammate that helps you smoothly transition into the agentic economy. You can get started for free with the process in just a few steps.