Skip to content
Agents tracked: 467 Releases (7d): 716 Agent status: all upSince last limit reset: Claude 37d · Codex 1d Updated Oct 29, 2026
Case study SimplePractice

How SimplePractice reached 85% deflection with Decagon and Duet

SimplePractice, a practice management platform for mental health professionals, deployed Decagon's chat and email agents and used Duet to evaluate which agent procedures to build. As reported by Decagon, it passed its year-end deflection target within a month of launch.

The problem

SimplePractice serves over 250,000 practitioners who need quick, contextual help, and its self-service help library was not meeting that need. Hiring in step with growing volume was not sustainable. Because practitioners work under strict regulatory requirements, any solution also had to handle Protected Health Information correctly.

How they did it

  1. Ran a detailed vendor evaluation with prepared questions and sample scenarios.
  2. Had Decagon build redaction functionality tailored to SimplePractice and tested it by pushing PHI through the system.
  3. Trained the agents on SimplePractice's customer voice and knowledge base.
  4. Configured conversational handling that asks clarifying questions, for example on rejected insurance claims, instead of using fixed decision trees.
  5. Set up soft handoffs to human agents that keep context intact.
  6. Used Duet to query real ticket volume and check whether a proposed Agent Operating Procedure is worth building.

Results

  • As reported by Decagon, SimplePractice exceeded its end-of-year deflection target of 66% within one month of going live, reaching 85%.
  • PHI redaction happened within seconds during testing in the sales process.
  • Product leaders now use Duet to pressure-test ideas against actual ticket data, moving from quarterly decision cycles toward faster experimentation.
  • The agent handles conversations in Spanish, Korean and other languages and translates context for the support team.

As reported by the source (Decagon customer story); AgentGid did not measure these figures.

Takeaway. Test compliance-sensitive behavior such as PHI redaction with real data during vendor evaluation, and use ticket data to decide which agent workflows are worth building.
AgentGid's take

This fits support teams with high ticket volume, a documented knowledge base and a compliance owner who can test redaction with real data. Decagon has no public pricing and requires a sales conversation, and the 85% deflection figure is vendor-reported. If you want to prototype cheaper first, Rasa is open source and Free for one bot, though its free tier is capped at 1,000 external conversations a month.

The agent used here

Similar use cases

Guides