Ford Prior

Ford Prior is a senior staff engineer in GEICO’s Claims Technology organization, where he works on AI-assisted engineering and claims modernization. The organization builds systems intended to help customers move from a loss to resolution with less disruption. His work spans developer productivity, product ideation, and the coordination needed to change consequential software systems.

Earlier in his career, Prior worked on delivery experience, inner-sourcing, and CI/CD pipelines as a principal DevOps engineer. In a 2021 Capital One article, he explained how Ansible playbooks and Python scripts can automate AWS deployments, distinguishing software deployment procedures from infrastructure provisioning. He also worked at AWS. He joined GEICO as an engineer and evangelist to help run an innovation lab, improve engineering productivity with AI, and share the team’s work.

From ideas to engineering decisions

Prior’s ideation experiment explored whether AI could generate claims product ideas that were both novel and feasible. He built a knowledge base from 147 meeting notes, then used three AI roles: a dreamer generated unconstrained possibilities, an interpreter turned them into proposals, and a critic assessed them against organizational and technical constraints. Prior made the final judgment himself. A plain-language claims explainer was among the promising ideas; a blockchain claims ledger failed his feasibility test. One idea advanced into the Claims Lab incubation pipeline.

At GEICO, Prior has also worked on what he calls the outer loop of software delivery: the system context, ownership, dependencies, review, deployment, and operations surrounding the code itself. In the team’s AI-native war room, decision makers and subject-matter experts work with a driver operating AI tools. FleetKit, an agent harness built on a framework supplied by Anthropic, divides a goal into research and planning threads. Agents and people review plans before builder agents produce changes; shared worksheets and diagrams help the team surface questions about ownership, approvals, and edge cases.

A one-day claims modernization pilot produced an agent-callable API into a legacy claims system and demonstrated it in a user acceptance testing environment. Production rollout and writes were outside its scope. For Prior, faster code generation still depends on engineers’ judgment: people must recognize when a team is optimizing for the wrong goal and decide how a change fits the wider system.

Building and organizing beyond GEICO

Prior also experiments with AI in personal workflows. His voice-memo transcription project uses Whisper, macOS Automator, and scripts to transcribe recordings locally and back up transcripts to Amazon S3. He documents practical limitations, including silent failures and the absence of automatic retries.

In Richmond, he helped create Hack for RVA, a volunteer civic-innovation initiative that brings technologists together with city and nonprofit stakeholders around defined problems. He has also organized the Richmond Cloud Wranglers user group and co-chaired Code & Cloud Con.