August 28, 2026

Hey Storylaners! This release is mostly about the training page — the part of RepX you spend the most time in when setting up an agent. It got faster, easier to scan, and better at handling the messy.

🎓 Training page, rebuilt

Updates to webpage training experience

• When you add URLs for RepX to learn from, you can enter multiple URLs one by one and RepX will crawl all those pages to learn from the content on those pages. You can also check the resync box.

• When you add a URL and want RepX to crawl all the pages linked from that URL, RepX lists all discovered URLs grouped by directories.

  • If you don't want RepX to learn from your /blogs, for instance, you can uncheck the entire directory.

• You can now see tabs in the URL section to filter URLs by the ones that have been queued for training, which ones have been processed, which ones have failed to get ingested. You

• Previously, if URLs had fragments or anchors (anything after #), the part after # used to get truncated. This no longer happens.

---

📄 More file types in transcript uploads

Sales call transcripts can now be uploaded as .docx and .txt, not just as .vtt. FAQs generated based on uploaded transcripts are used by RepX when generating answers.

---

📊 Marketo integration

RepX now writes conversation data to Marketo - leads, activities, and the RepX fields on each record. You can set it up from Settings → Integrations.

---

🌍 Multilingual support (in voice and text responses)

RepX can now understand and respond in different languages, including Italian, French, Spanish, German, etc. Visitors can switch languages mid-conversation and RepX will respond in the same language the visitor talks or sends messages in.