Proto has released OpenAI Vision across its AI customer experience solutions, enabling enterprise and government clients to use image recognition for document analysis and visual query resolution without human intervention.
This integration allows AI assistants to interpret complex visuals, such as forms, error codes, and screenshots, with accuracy and speed.
Proto ensures this capability is applied only in cases where no personal identifiable information (PII) is present, maintaining its commitment to ethical AI practices and data privacy laws.
One standout example of this integration is Proto’s collaboration with Yinson GreenTech’s chargEV, Malaysia's largest electric vehicle (EV) charging network.
Visual driver support in Malaysia
chargEV is using image recognition within Proto’s AI assistant to improve support for drivers encountering charging difficulties. The AI assistant can now interpret charger error photos and app screenshots, a critical capability given that 15% of all support chats include images from drivers.
Previously, Proto enabled chargEV to automate 64% of customer queries across text-based messaging in Malay, Chinese, English and Manglish (the Malay-English mixed language). Resolving visual queries remained a challenge. Human agents were often required to interpret photos of damaged chargers, error codes, or unclear app interfaces — delaying responses and creating bottlenecks.
With the integration of image recognition, Proto’s AI assistant now delivers faster, more accurate support in critical scenarios by:
- Deciphering Charger Error Codes: Instantly recognising and diagnosing specific faults in charger images.
- Analysing App Screenshots: Understanding screenshots of chargEV’s mobile app to guide users through fixes.
- Reducing Reliance on Human Agents: Allowing human agents to focus on complex or nuanced cases, while the AI handles routine visual inquiries autonomously.
In the first week of the integration, 71% of chats with images were handled by solely by the AI assistant without human intervention.
For EV drivers, particularly those facing low battery levels, the ability to receive instant, accurate support is a stress reliever. The rapid image analysis reduces delays caused by manual interpretation, helping chargEV get drivers charged-up and back on the road faster than competing brands.
Proto’s integration of OpenAI Vision also highlights the critical role played by Proto’s team of prompt engineers, which assist enterprise clients in discovering opportunities for further automation. In the case of image recognition, this team is actively involved in fine-tuning the underlying prompts to adjust for image reflections and file quality.
