You’re able to initiate training in the product yourself. Here’s how it works:
3.1 Baseline training (during implementation)
- What it is: One-time training using documents imported from your EMR and a cleaned-up set of document types.
- Goal: Give your model a strong, relevant starting point so it recognizes your common faxes immediately.
- Who does it: You can do this yourself or reach out to your Phelix Implementation Manager
Good fit scenarios
- Fewer than 50 document types → often “ready out of the box.”
- 50–500 types, well-organized and evenly used → baseline training can deliver strong results quickly.
Caveat scenarios
- 50–500 types but messy/uneven usage → baseline training helps, but some types will need usage-based learning to catch up.
- 500+ types → baseline alone may not move the needle much; expect gradual improvement plus consolidation work.
3.2 Usage-based training (automatic, after go-live)
- What it is: Inbox Agent learns from day-to-day filing and corrections by your staff.
- What improves: High-volume categories first (e.g., Referral, Labs); rare categories take longer.
- What you do: Just work in Inbox Agent. The more you use it, the faster it learns.
Example:
If your clinic files dozens of Referrals daily, you’ll see that category stabilize quickly. If you get only a couple of Consult Notes per week, that category improves more slowly unless we add extra examples.
3.3 Manual/batch uploads (to accelerate learning)
- What it is: Adding sets of example documents for a chosen category (e.g., 50 Consult Notes) so the model learns faster.
- When it’s used now: You can do this during implementation or post-go-live to boost specific categories.
- Best practice: Provide ≥30 examples per document type, with variety (different layouts/providers/templates).
Putting it together — what this means for your clinic
- If you have <50 document types:
You can often use Inbox Agent out of the box with minimal training. Expect quick recognition of common faxes. - If you have 50–500 clean document types:
Our initial training based on your EMR documents typically gets you good recognition right after setup. Early weeks of usage make it even better. - If you have 50–500 messy document types:
Baseline training helps, but full coverage needs daily usage and (optionally) batch uploads for the rarer categories. We’ll guide you. - If you have >500 document types:
Plan for consolidation and a longer runway. We’ll partner with you on grouping types and supplement with usage learning and targeted uploads.
FAQs
Can we change our document type list after training?
Please avoid removing or reordering types. That can harm accuracy. You can add new types, but they’ll only be recognized after retraining or with enough usage.
How do we speed things up for a tricky category?
Provide at least 30 example documents for that category. We can upload them and plan a retraining round.
Will we be able to train ourselves?
Yes — with Phelix’s self-serve tools, you’re able to upload examples and request retraining directly from the Document Training page.
Related next steps
- After training you can enable automation (Auto-Save, Auto-Referral, Auto-Chart). See: Automating Document Filing with Inbox Agent.
- If you’re starting with a very large/complex list, ask us for a consolidation working session before go-live.