Customer Service Automation
Ada needed to move past containment-rate chatbots and build a customer-service system that actually resolves issues well, supports voice use cases, and reduces hallucinations at production scale.
Reported impact
Ada doubled automatic resolution rates for migrated customers, kept containment roughly similar, and created a measurable framework for high-quality support automation that improves ROI, satisfaction, retention, and signup performance for its customers.
Ada rebuilt its AI-native customer-service platform around OpenAI models so it could optimize for resolution rate instead of containment alone. The company first created an evaluation framework that scores whether conversations are resolved with relevant, accurate, and safe replies, reaching 80-90% agreement with human reviewers. It then combined GPT-4, GPT-4o, OpenAI's API, and the fine-tuning API into a multi-agent configuration with a central planning agent and a set of subagents that understand the request, reflect on it, call tools, and keep gathering knowledge until an answer is ready. GPT-4o improved low-latency reasoning for voice use cases, while fine-tuning is used to estimate hallucination risk and feed that score into the rest of the toolchain. This let Ada preserve roughly similar containment rates while doubling resolution performance, and gave enterprises new coaching, transparency, and control levers over LLM-based agents.
Customer requests enter the AI Agent, where a planning agent and subagents use GPT-4 or GPT-4o to understand the issue, call tools, retrieve knowledge, and generate an answer. Ada's evaluation layer scores resolution quality, and fine-tuning outputs hallucination-confidence signals that are used to suppress risky responses before escalation or delivery.
"Ada's AI Agent uses a multi-agent configuration, with a central planning agent and a collection of subagents"
"For real-time reasoning, we've seen a big leap in latency and overall quality with GPT-4o"
"Initially using GPT-4 ... Ada built a new evaluation framework"
"they have the similar containment results, but typically see a resolution rate of up to 60%"
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
The Organization JSON-LD lists addressLocality 'Toronto', addressCountry 'CA', and numberOfEmployees value 421.
Tracxn reports Ada Support (ada.cx) has 803 employees as of Aug 31, 2026.
Related
OpenAI Sample Sep 22, 2026 Ringg Ringg needed to help enterprises automate large volumes of customer calls and chats with reliable tool use, low latency, multilingual quality, and unit economics that improved rather than worsened at scale. 4.8 average customer CSAT GPT-4.1 · GPT-5.6 Luna · GPT-5.6 Sol OpenAI Sample Aug 3, 2026 Circles Circles needed to turn rich telco signals such as usage, billing, network activity, and location into real-time support and monetization actions instead of reactive, siloed customer journeys. 22% increase in ARPU in Singapore from AI-driven personalization AI Concierge · CareX · Codex OpenAI Sample Jun 12, 2026 Preply Preply wanted to personalize language learning more deeply and reduce repetitive tutor work without replacing the human relationship that makes one-to-one teaching effective. 4.7/5 satisfaction rating Brand Voice GPT · ChatGPT Enterprise · Codex 