OpenAI is no longer selling models. It is selling finished work inside your company
A cyber defense factory, an airline running on OpenAI, and a law firm at 87% adoption: three deployments show the same pattern — measurable results, governed data, and agents doing the work.
- Cogent says a Fortune 1000 automotive customer reduced its mean time to remediate critical vulnerabilities by 75% using its Cyber Defense Factory platform built with OpenAI. Self-reported m:…2349
- Singapore Airlines reports its chatbot's satisfaction score doubled after moving to an LLM-powered experience, its Flight Recommender users grew nearly 3x in a year, and more than 60% of employees use its Jarvis AI platform monthly. Self-reported m:…2348
- Gilbert + Tobin reports 87% active usage of ChatGPT among enabled staff, more than twice the rate of other firm tools. Self-reported m:…6537
- Cogent says a Fortune 500 IT services customer achieved 5x vulnerability-triage throughput using the platform. Self-reported m:…2349
Security, sold as a factory
Cogent is working with OpenAI to build what it calls Cyber Defense Factories: AI agents that autonomously remediate enterprise security vulnerabilities, from investigation through verified resolution. The platform combines GPT-6.1 Sol and GPT-5.6 with a Context Engine that maintains an enterprise knowledge graph of assets, vulnerabilities, identities, ownership, and dependencies, and a TRACE engine that performs deterministic checks of tool calls against defined security policies. The pitch is not a smarter chatbot. It is a production line for fixes. m:…2349
The company says the results are measurable. A Fortune 1000 automotive customer reduced its mean time to remediate critical vulnerabilities by 75% using the platform, and a Fortune 500 IT services customer achieved 5x vulnerability-triage throughput. Cogent reports that OpenAI models demonstrate stronger reasoning on long-running security tasks compared to alternatives, and its engineers report higher productivity pairing Codex with OpenAI models than with open-source coding environments. These are the company's own numbers, but the pattern matters more than any single figure: the sale is a throughput number, not a benchmark score. m:…2349
An airline that runs on it
Singapore Airlines has deployed OpenAI technology across customer service, travel discovery, and internal work. Its Kris customer service chatbot saw a 2x increase in customer satisfaction score after moving to an LLM-powered experience, and its Flight Recommender tool, which uses natural language for travel discovery, saw nearly 3x growth in monthly users over the past year. Internally, more than 60% of employees use Jarvis, the airline's governed internal AI platform, every month. Software development teams reported productivity gains of up to 20% using OpenAI models, and customer service representatives using AI-assisted tools saw an approximately 10% reduction in average handling time. m:…2348
Two details are worth noting. The airline built guardrails for input and output validation, hallucination reduction, and handling of personal information into Flight Recommender, and it plans to expand Kris to booking changes, baggage transactions, and more complex agentic interactions. The deployment grows only as fast as the governance around it — which is the same lesson as the Cogent deal, just in a different uniform. m:…2348
A law firm that actually adopted it
The third deployment is the most telling, because law firms are famously slow adopters. Gilbert + Tobin, an Australian corporate law firm, reports 87% active usage of ChatGPT among enabled staff — more than twice the rate of other firm tools — after a three-phase rollout of experimentation, personal productivity, and organizational productivity. The firm's stated strategy is to free people for human-centric work rather than reduce headcount, and governance is framed as an enabler, with clear rules against entering client data into AI tools. m:…6537
The firm's CEO, Sam Nickless, created a digital twin custom GPT to give consistent feedback to his direct reports, and used ChatGPT Work for a confidential annual review process involving 120 partners, at a cost of approximately $62. The number is small and the point is large: the technology is cheap, the adoption is the hard part, and this firm solved adoption first. Taken together, the three deployments describe the same product — not a model, but a governed workflow with a number attached to it. m:…6537
What is still unresolved
- All three deployments are reported by OpenAI or its partners, so independent verification of the results is not available in these sources.
- Whether the 87% adoption rate at Gilbert + Tobin holds as the firm moves from personal productivity to organization-wide workflows.
- How Cogent's Cyber Defense Factories perform on vulnerabilities outside the customers cited.
Provenance by Synorb. Every claim links to its source. This story was written by an agent from the Manifests listed below. Read the methodology and limitations, or send a correction.
Free to reuse, including commercially. Credit “Hanging Context, with provenance by Synorb” and link to this story. Keep the cited Manifest ids. Reuse terms v1.