Honestify
← Back
AI Improvementsv1.2.0

AI Chat Improvements

AI Chat Improvements (2025-12-15): The profile AI chat now retains conversation context, cites source projects, and handles follow-up questions without repeating prior answers.

·4 min read·Honestify Engineering

Feature area: AI Chat

On this page

This release covers AI Chat Improvements Version 1.2.0, shipped 2025-12-15. Status: shipped. No breaking changes.

Summary

The profile AI chat now retains conversation context, cites source projects, and handles follow-up questions without repeating prior answers.

Early chat sessions felt stateless: users asked clarifying questions and received generic responses unrelated to their profile. We rebuilt the chat stack with session memory, retrieval-augmented context, and tighter system prompts scoped to user data.

What Changed

Session memory

New

Conversation history persisted per profile with token budget management.

Context retrieval

Improved

Relevant profile chunks injected before each response.

Citation labels

New

Responses link to source projects when claims derive from profile data.

Follow-up satisfaction

Before

61%

After

82%

Thumbs-up on second+ messages in session

  • Session memory — Conversation history persisted per profile with token budget management.
  • Context retrieval — Relevant profile chunks injected before each response.
  • Citation labels — Responses link to source projects when claims derive from profile data.

Why We Built It

Early chat sessions felt stateless: users asked clarifying questions and received generic responses unrelated to their profile.

We prioritized this work because metrics showed a measurable funnel leak we could fix in one sprint. The fix needed to be durable—not a patch—so we addressed root causes in AI Chat rather than symptoms alone.

Engineers, recruiters, and hiring managers all benefit when Honestify behaves predictably in production. This release reflects that bar.

User Impact

Follow-up question satisfaction (thumbs-up rate) increased 34% in beta cohort.

AudienceHow you benefit
EngineersFaster profile setup, clearer AI answers, less manual rework
RecruitersMore complete profiles and reliable share links when candidates use Honestify
Founders / hiring managersBetter signal on candidate preparation and skills alignment
Platform engineersInfrastructure patterns that reduce incident risk

Relevant skills: rag, prompt engineering, langchain, typescript. Target roles: ai engineer, backend engineer, full stack engineer.

Technical Highlights

  • Sliding window context with summarization for long threads
  • Embedding retrieval over profile sections
  • Prompt templates versioned in repo
  • Streaming responses with partial token display

Rollout used feature flags with staged percentage increase and automatic rollback on error budget burn.

Before

AI Chat Improvements: before vs after

Before

Each message was processed independently; the model could not reference earlier turns or specific profile sections.

After

Multi-turn conversations with explicit citations to projects, skills, and prior answers in the thread.

Users moving from the previous experience should notice multi-turn conversations with explicit citations to projects, skills, and prior answers in the thread.

Screenshots

Future Improvements

What we are building next

  • Voice input for mock interviews
  • Export conversation as prep notes
  • Interviewer persona modes

Known limitations

  • · Sessions expire after 24 hours of inactivity
  • · Citations require structured project entries

Feedback welcome: Reply via in-app feedback or support—especially if you hit edge cases we did not cover in this release.

This update connects to other Honestify work:

Create your own AI profile

Upload your resume, add expertise, and share a profile link beside LinkedIn so recruiters can ask follow-up questions before the interview.

Create your AI profileUpload your resume