HiBob turned its experiment with GPTs into a machine that speeds up adoption, product development, and internal decisions. Their formula? Give the right people the tools, measure concrete results, and reuse what works. (openai.com)
GPTs in every team: from idea to daily tool
It's not magic or a secret AI department. HiBob made GPTs available to sales, product, CS, and operations employees to solve real tasks: from preparing meetings to spotting upsell opportunities. Teams use these GPTs as assistants that condense context, data, and actionable steps. (openai.com)
Concrete examples they mention from their case:
- Meeting Prep GPT: pulls CRM data and notes to prepare briefings.
- Upsell GPT: analyzes usage patterns to prioritize accounts.
- VBO Project Manager Assistant: summarizes onboarding calls and creates task lists.
- SEO Assistant: connects web analytics to recommend keywords.
- Roadmap GPT: turns roadmap data into actionable insights.
These examples show something simple: when a GPT is integrated into the workflow and has an owner, it stops being an experiment and becomes a coworker. (openai.com)
How they turned GPTs into reusable agents
HiBob didn't just allow GPT creation: they put in place a repeatable process to turn ideas into agents other teams can use.
- Idea and proof of concept: proposals anchored to a specific problem.
- Build: engineers create secure agents with ChatGPT Enterprise and internal systems.
- Adoption and enablement: documentation, training, and a responsible owner.
- Maintenance: feedback loops to refine them.
- Scale: successful agents enter an internal directory to be reused.
That process is what converts hundreds of prototypes into dozens of productive tools. (openai.com)
Measurable results and clear lessons
HiBob shares concrete metrics that are useful if you're thinking of bringing AI into your team:
- Over 90% of their employees actively use ChatGPT Enterprise. (openai.com)
- They built more than 2,500 experimental GPTs and deployed 200 into internal flows. (openai.com)
- They integrated features that use models like GPT‑4o into their customer platform. (openai.com)
"AI isn't going to take people's jobs. Humans who know how to work with AI will." — Ronni Zehavi, cofounder and CEO. (openai.com)
That line sums up the bet: it's not about replacement, it's about amplifying capabilities and measuring impact in hours saved, revenue, and satisfaction.
What you can apply today, without a million-dollar budget
- Start with a case that really hurts: cut prep hours or improve onboarding.
- Define an owner and metrics from day one: who is responsible for the GPT and how success is measured.
- Prototype in a closed environment (for example ChatGPT Enterprise) before integrating into production.
- Document and publish: an internal directory prevents duplicated effort.
- Measure economic or time impact: that turns experiments into investments.
Conclusion
HiBob shows that the big difference isn't the technology itself, but the discipline to turn experiments into repeatable processes. If you give people tools, structure, and metrics, AI stops being a luxury and becomes a systematic way to improve work and product.
If you want to read the original piece, you can find more on HiBob's site or in OpenAI's post about their case. HiBob · OpenAI: HiBob
