The Generative AI Design Process
Common Questions about the Generative AI (GAI) Design & Delivery Process:
Q: "How will our AI delivery process change with Generative AI?"
A: It mostly stays the same, but initial design and development time to "first draft" will decrease. More time will be needed to test for potential hallucinations. As a designer you should expect to update the prompt in Prompt Library and test, retest, and test some more.
Q: "How do I know when I've tested my prompt enough?!"
A: Test an answer / flow at least 10 times. If you don't see any hallucinations or errors, you're good to go!
The Generative AI Design & Dev Delivery Process
- Prompt Design & CX Testing:
- Refine use cases and intents.
- Diagram GAI handoffs from non-GAI ai (conversation builder flows).
- Write detailed prompts and guardrails.
- Collaborate with Tuner for larger projects.
- Design > Dev Handoff:
- Share prompt and flow with Developers.
- Set up handoff calls and involve QA for test planning.
- Prompt Development & Architecture:
- Developers create or enhance prompts, focusing on adding in additional variables, context, logic, and integrations.
- Test flow again with designer.
- Add custom events if needed
- Final QA/UAT Before Launch:
- QA Engineers test the full build, confirm GAI flow integrity, and collaborate with Designers and Tuner.
- Secure stakeholder sign-off before launch.
Key Reminder:
Changing the prompt means changing the design.
Was this article helpful?
Be the first one to vote!
This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service
Loading