customer support · pattern

Knowledge base self-service

Help-center articles and KB-grounded answers that let customers solve their own questions without touching the queue.

Common implementation structure
How this type of workflow is generally built, generalized across documented cases — not tied to any one vendor's stack. Click any stage to read what happens there. Specific products that implement these stages appear in “Tools commonly seen” below.
Stage 1 · Knowledge ingestion & indexing
Help-center articles, internal wikis, and resolved-ticket history are indexed for retrieval; updates flow as content changes so answers stay current.
What fails first / common problems

Recurring first-deployment failures from the matching workflows'what_failednotes. First sentence of each, attributed to the source case.

Epos Now's IVR system was pre-configured around scripted routing and failed to deliver the quality of experience they wanted, with customers sometimes ending up with the wrong agent and experiencing longer wait times.
from: Epos Now saves 60,000 human labor hours per month and achieves 70% automated resolution rate with Ada's AI agent Sidekick
The existing IVR system was outdated, rarely updated, and unable to retain callers in self-service, sending the majority to expensive outsourced live agents.
from: Catholic Health achieves 57% call containment and zero wait times with Notable Voice AI
YAZIO had minimal success with an internal AI search tool, and other vendors they evaluated offered only rule-based chatbots built on large decision trees requiring regular maintenance—not true agentic AI.
from: YAZIO deflects 80% of support tickets without lowering CSAT in 6 months with Forethought Solve
Before Forethought, Kickfin had no automation tools for customer self-service, forcing the team to staff overnight human shifts that were chronically difficult to fill and cover.
from: Forethought AI enables 72% self-serve rate and 96% sentiment accuracy for Kickfin customer support
The previous chatbot provider gave users inaccurate responses, required manual keyword entry for every workflow, and produced thousands of duplicated, incorrect workflows that became too complex to manage.
from: Forethought AI helps Upwork achieve a 75% chat self-serve rate and 50% faster ticket resolution
Tools commonly seen
ragforethoughtlangchainworkflow builderforethought solvegpt-4langgraphlangsmithllm judgesolveadaamazon bedrock
Representative outcomes

Real metrics from selected cases — verbatim from each workflow'snumberspanel. Click any title to open the full case.

Epos Now saves 60,000 human labor hours per month and achieves 70% automated resolution rate with Ada's AI agent Sidekick
Time saved60,000
Volume30%
Cost40%
Ask Julie intelligent virtual assistant delivers 8x ROI and 32% containment increase for Amtrak
Time savedlower call volumes and handle times
Volume8x
Cost30%
Leading life insurance company achieves 97.9% IVA containment rate with Verint Intelligent Virtual Assistant
Time savedsix months after the IVA's launch
Volume97.9%
Catholic Health achieves 57% call containment and zero wait times with Notable Voice AI
Time saveddropped to zero
Volume57%
Cost$350,000 a year
Forethought Solve helps Spordle deflect 21,000 chat inquiries at an 86% self-serve rate
Time saved142%
Volume21,000
Example workflows

Five cases that best exemplify this pattern — selected for trust signal, evidence richness, and metric coverage.

Knowledge base self-service
Epos Now saves 60,000 human labor hours per month and achieves 70% automated resolution rate with Ada's AI agent Sidekick
Ada → Sidekick → IVR
Sidekick now automates 70% of support demand across all channels, saves 60,000 human labor hours per month, increased CSAT by 3….
View case →
Knowledge base self-service
Catholic Health achieves 57% call containment and zero wait times with Notable Voice AI
Notable AI Voice Agent
After a five-week launch, Notable's AI Voice Agent managed over 25,000 calls, raised containment from 30% to 57%, delivered a 5….
View case →
Knowledge base self-service
DoorDash 2025 Summer Interns Build In-House LLM for Never-Delivered Order Feature Extraction and RAG-Based Chatbot Service
Meta Llama 3 → DistilBertForSequenceClassification → Kafka → Cadence
The in-house fine-tuned DistilBertForSequenceClassification model achieved an F1 score of 0.
View case →
Knowledge base self-service
Ask Julie intelligent virtual assistant delivers 8x ROI and 32% containment increase for Amtrak
Ask Julie → Verint IVA
Ask Julie delivered an 8x return on chatbot investment, a 32% increase in containment, and 30% more revenue per booking, while ….
View case →
Knowledge base self-service
Lime achieves 77% reduction in response time with Forethought AI triage and automated support
Forethought Triage → Forethought Solve → Workflow Builder → Robotic Process Automation
Forethought automated 27% of email and web cases and tagged 98% of support tickets automatically out of more than 1.
View case →