Meridian
An AI research analyst that interviews HR software buyers by voice or chat and matches them to verified vendors.
4 sources: GitHub (groverpaul/Meridian), Claude Code, Replit export, live link
What they built
Meridian is the buyer-facing AI analyst for Kyle & Co, an independent HR technology research platform. It runs a structured interview — by voice or text — to learn what a buyer actually needs, then matches them against vendors scored on the Category Compass framework and hands back a buyer brief. Analysts get an admin suite for vendor data, and vendors get a self-service portal to correct their own profiles. It started as a Replit prototype and was then rebuilt and extended in Claude Code.
-
Buyer interview
A LiveKit voice agent (Deepgram STT, Claude, ElevenLabs TTS) or an in-browser text chat runs a guided requirements interview.
See it -
Post-processing
When the interview wraps, the transcript is turned into structured buyer requirements.
See it -
Objective matching
Hard gates plus a scored rubric against the Category Compass bearings, producing tiers rather than vanity rankings.
-
Buyer brief
A generated brief with matches and reasoning, emailed to the buyer on consent.
See it -
Vendor research agent
A Claude-powered CLI and background enrichment queue that builds a structured vendor profile from just a URL.
See it -
Vendor intake portal
HMAC-secured links let vendors review and correct their AI-drafted profile before it goes live.
-
Compare workspace
Side-by-side vendor comparison with a streaming AI analyst to answer questions about the shortlist.
-
Analyst admin
Vendors, sessions, transcripts, leads, prompt editor, ingestion pipeline, and vendor analytics in one dashboard.
Core features
4 of 8 features on screen. Each one is a thing Meridian does.
Buyer interview
A LiveKit voice agent (Deepgram STT, Claude, ElevenLabs TTS) or an in-browser text chat runs a guided requirements interview.
Post-processing
When the interview wraps, the transcript is turned into structured buyer requirements.
Buyer brief
A generated brief with matches and reasoning, emailed to the buyer on consent.
Vendor research agent
A Claude-powered CLI and background enrichment queue that builds a structured vendor profile from just a URL.
How it was built
Meridian took three runs. The first was on Replit: 15 weeks from late March to mid July 2026, with 2,192 commits along the way. In July the code moved to GitHub and Replit took over. That second run was 8 weeks from mid July to early September 2026, with 37 commits along the way. In July Claude Code took over. That third run was 3 sittings over 5 days, 8 prompts in all, and 232 commits. One person built it.
28 March to 13 July 2026. 2,192 commits across 16 active weeks. The busiest was the week of 29 March, with 640. 1,635 files of Python, JavaScript and TypeScript. Replit keeps the chat, so this run's conversations are not on this page.
Read from the Replit export the builder uploaded.13 July to 4 September 2026. 37 commits. 1,686 files of Python, JavaScript and HTML. Replit keeps the chat, so this run's conversations are not on this page.
Read from GitHub.13 July to 14 October 2026. 232 commits across 12 active weeks. The busiest was the week of 13 September, with 78. 3 sessions, 8 prompts. 1,954 files of Python, Markdown and JavaScript.
Read from Claude Code's own session files.- Week of 12 Jul: 3 commits
- Week of 26 Jul: 1 commit
- Week of 2 Aug: 1 commit
- Week of 9 Aug: 9 commits
- Week of 16 Aug: 20 commits
- Week of 30 Aug: 4 commits
- Week of 6 Sep: 1 commit
- Week of 13 Sep: 78 commits
- Week of 20 Sep: 18 commits
- Week of 27 Sep: 34 commits
- Week of 4 Oct: 51 commits
- Week of 11 Oct: 12 commits
- 27 weeks
- active weeks
- 7 days ago
- last push
- 3 sessions
- Claude Code sessions
- 8 prompts
- typed in Claude Code
In their own words
Answered by the builder, by voice or by hand.I built Meridian to create a way for us to monetize our research. We had done an AI interviewer research project called the Category Compass, where I went really deep on every AI interviewer product across 12 different solutions and vendors. I realized how valuable that research was to prospective…
I built Meridian to create a way for us to monetize our research. We had done an AI interviewer research project called the Category Compass, where I went really deep on every AI interviewer product across 12 different solutions and vendors. I realized how valuable that research was to prospective buyers, because they are not very educated on the category, yet they are trying to move fast. Meridian was a way to bridge that gap between buyers and vendors by monetizing on the vendor side, doing the research on their behalf, and giving them a verified vendor profile.
The biggest challenge was fine-tuning the latency in the voice component during a buyer brief interview. Sometimes the voice would jump in too quickly, and sometimes it would wait too long. Getting that to feel natural and flow naturally was definitely a challenge. It took a lot of testing, and…
The biggest challenge was fine-tuning the latency in the voice component during a buyer brief interview. Sometimes the voice would jump in too quickly, and sometimes it would wait too long. Getting that to feel natural and flow naturally was definitely a challenge. It took a lot of testing, and doing these interviews, which take five minutes each, turned out to be quite time-consuming. But ultimately we got it to a great place.
My least favorite part was dealing with so many third-party integrations. Managing them, manually entering secret keys, and creating API keys was time-consuming and just a pain in the butt. It was really annoying and frustrating, and it felt highly inefficient. I wish somebody would build a solution to fully automate that.