GoTo: An Engineer Sets His Rules for Using AI at Work

Key Takeaways

1Brandon Pedersen, principal engineer at GoTo, leads a team dedicated to AI activation

2He applies a two-second rule to decide whether to automate a task and recommends always verifying AI results

3Tools like Claude Code and GitHub Copilot are used, but he maintains control over sensitive documents

Why it matters : Brandon Pedersen's experience illustrates how to structure AI adoption in business while preserving quality and human accountability.

At GoTo, Brandon Pedersen oversees the activation of AI and outlines a very concrete usage framework. Balancing tasks to automate with critical work that should remain human, he combines systematic checks, the two-second rule, and prompt training. He anticipates improvements in tools while maintaining a demand for control.

Quickly Solve Without Losing Control, Up to 40 Minutes

Brandon Pedersen notes that users sometimes manage to resolve an issue in 30 to 40 minutes by clearly describing their goal to an AI tool. Despite errors that he considers less frequent in small cases, he still observes them in more complex situations. He expects systems to gain in capability, speed, and intelligence, likely making fewer mistakes. However, he believes that users' judgment, critical thinking, and values will remain essential.

Brandon Pedersen Enforces Systematic Verification of AI Results

To avoid pitfalls, he prohibits blind trust and recommends asking models for sources, following links, and verifying logic. His two-second rule leads him not to automate an action that can be done manually just as quickly, a point reinforced by tasks that colleagues attempt to automate even though they take two seconds. He advises against using a model for personal exchanges. During a training session relayed by the financial group, the same advice on traceability and verification was emphasized. For sensitive content, such as the internal governance policy of AI for which he was responsible, he discarded a generated draft to be able to defend every word.

Le brief IA que lisent les pros

Le brief IA que les pros lisent chaque soir

Les 7 actus IA du jour, décryptées en 5 min. Gratuit.

Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.

Choisissez votre rythme

Gratuit · Pas de spam · Désabonnement en 1 clic

Organization: A Dedicated Team for AI Activation

At GoTo, he leads an AI activation team that maintains platforms, designs new tools, and provides training and consulting within the company. The group gathers automation ideas, assesses what will have the most impact, and selects what deserves to be automated. On a daily basis, he trains teams to better formulate their prompts and implements automations through AI skills or custom applications. Brandon Pedersen is a principal engineer at the company, where he started in 2013.

The Engineer Adjusts His Prompts to Improve Response Quality

He approaches each model output with the idea that there is likely a point to improve, clarify, or correct, noting that small variations in prompts can completely change the result. He states that he has always reviewed what the AI produces, even for a simple line of code, due to errors. Having worked intensively on these uses for three years, he claims to trust tools like Claude Code and GitHub Copilot for significant work and indicates that they allow him to build solutions on a larger scale.