References tracker + recommendation letter drafts #96

New issue

Closed

opened 2026-04-13 16:58:45 -07:00 by pyr0ball · 0 comments

pyr0ball commented

2026-04-13 16:58:45 -07:00

Owner

Summary

Add a references management system to Peregrine, covering three surfaces:

References tracker — store name, relationship, company, phone/email for each reference. Persists across applications.
Reference prep email generator — for a given application, draft a short email to send each reference: what the role is, which accomplishments to highlight, what the hiring manager is likely to ask about. Most candidates skip this entirely; it directly improves close rates.
Recommendation letter drafts — user provides recommender relationship + key talking points → AI drafts a letter the recommender can edit and send on their own letterhead. AI is the draft, human signs off.

Bonus: Reference relevance matching

For a given job, surface which stored references are the strongest match (e.g. technical lead for IC engineering roles, former manager for leadership-track positions). Could be a simple tag system (technical / managerial / character) with a match hint on the apply view.

Tier

Reference storage and prep email: Free (drives signups)
AI-drafted recommendation letter: Paid (or local LLM BYOK)

Fits existing pattern

AI Assist → Human Approval → Execute — AI drafts the letter or prep email, user reviews, recommender sends. No AI output goes anywhere without human review.

UX placement

New "References" section in the profile/settings area
On the Apply view: "Prep your references" step after cover letter, before submit
Recommendation letter flow: triggered from the Apply view or standalone from References

## Summary Add a references management system to Peregrine, covering three surfaces: 1. **References tracker** — store name, relationship, company, phone/email for each reference. Persists across applications. 2. **Reference prep email generator** — for a given application, draft a short email to send each reference: what the role is, which accomplishments to highlight, what the hiring manager is likely to ask about. Most candidates skip this entirely; it directly improves close rates. 3. **Recommendation letter drafts** — user provides recommender relationship + key talking points → AI drafts a letter the recommender can edit and send on their own letterhead. AI is the draft, human signs off. ## Bonus: Reference relevance matching For a given job, surface which stored references are the strongest match (e.g. technical lead for IC engineering roles, former manager for leadership-track positions). Could be a simple tag system (technical / managerial / character) with a match hint on the apply view. ## Tier - Reference storage and prep email: **Free** (drives signups) - AI-drafted recommendation letter: **Paid** (or local LLM BYOK) ## Fits existing pattern `AI Assist → Human Approval → Execute` — AI drafts the letter or prep email, user reviews, recommender sends. No AI output goes anywhere without human review. ## UX placement - New "References" section in the profile/settings area - On the Apply view: "Prep your references" step after cover letter, before submit - Recommendation letter flow: triggered from the Apply view or standalone from References