Intern Pipeline

Total Active

—

all departments

Week 0

—

onboarding

Probation

—

W1 + W2

Intern

—

regular

Name	Candidate ID	Department	Stage	Status	Start Date	Misses	Actions
Loading…

Probation Week 1 — Observe

FT observes without intervening. Bot sends the FT a daily focus brief each morning and signal questions each evening. Signals logged: proactive start, thinking before building, debugging posture, comms rhythm, depth under questioning.

Probation Week 2 — Hold

FT holds the intern to commitments made on Day 6. Specific deliverables with specific dates. Bot tracks Wednesday and Friday deadlines. Miss a deadline without flagging it first → formal warning.

Intern — Ongoing

Regular interns still submit EOD reports. Same miss logic applies — miss one gets a warning, miss two and HR is notified. No elimination recommendation — HR follows up directly.

📋

EOD Report Format

▼ expand

Every intern posts their EOD report in their department Slack channel. The message must start with EOD: or REPORT: — otherwise the bot ignores it.

EOD:
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
EOD REPORT — Your Name — DD Mon YYYY
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

TASKS COMPLETED
- Task 1 with PR/ticket ID if any
- Task 2

TASKS IN PROGRESS
- What you started but didn't finish
- Expected completion: tomorrow / specific time

BLOCKERS
- What is stopping you / None

WHAT I LEARNED TODAY
- Concept or tool you understood today

AI USAGE TODAY
Tool used: Claude / ChatGPT / Copilot
Chat link: https://...

PLAN FOR TOMORROW
- Task 1 with expected output

CONFIDENCE LEVEL TODAY
> [x] On track  [ ] Crushing it  [ ] Need help  [ ] Stuck
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

The bot uses Claude to extract all fields from the report. If any required field is missing, it replies with exactly which fields are missing and the correct format.

                The AI chat link is required. Interns must share the actual conversation link so their thinking can be verified. If they used AI but have no link, that is flagged as a quality issue.
              

⚠️

Miss and Warning Logic

▼ expand

The bot checks for missed reports 30 minutes after each report window closes. Before firing a warning, it checks the intern's Slack status — if they have set a leave-related status, the miss is skipped.

Leave keywords detected: leave, sick, away, ooo, out of office, on leave, medical, emergency, holiday, unavailable, off today.

Stage	Miss 1	Miss 2
Week 0	Warning posted in channel. References commitment. States consequence.	Elimination recommendation sent to HR. HR confirms before outcome is posted.
Probation W1 / W2	Warning posted. Logged as a signal flag in the evaluation.	Formal warning posted. FT and HR notified via DM. Reflected in end-of-week eval.
Intern	Warning posted in channel. Softer tone.	HR notified via DM. No elimination — HR follows up directly.

Sundays are skipped. No warnings fire on Sunday. Day numbers also skip Sundays — so a week is 6 working days, not 7.

📊

Evaluations — What the Bot Reads

▼ expand

At the end of each stage, the bot generates a full evaluation using Claude. It reads all reports, all FT signal answers, and prior eval outputs — and produces a recommendation.

Week 0 signals (4):

Report cadence

Did they submit on time without being chased?

Technical engagement

Are reports showing real AI tool usage or surface-level activity?

Honesty

Do they admit blockers? Any signs of fabricated progress?

Friction response

When tools broke or APIs failed — what did they do?

Probation W1 signals (5):

Proactive start

Clarifying questions Day 1, architecture Day 2 without being asked.

Thinking before building

Documented plan before code. Trade-offs named. Failure cases considered.

Debugging posture

Blockers surfaced with specific detail and prior attempts.

Comms rhythm

Daily reports on cadence showing real reflection, not just activity.

Depth under questioning

Day 4 check — explained decisions, held reasoning under pushback, knew where system fails.

Probation W2 signals (6): delivery reliability, proactive comms, ownership of hard parts, depth under questioning, end-of-probation document quality, and two-week trajectory.

                HR confirms before anything is communicated. Every evaluation recommendation — offer, no offer, extend, end relationship — goes to HR first. The bot never announces an outcome without HR approval.
              

⏰

Automated Schedule — What Fires When

▼ expand

Time (IST)	What happens
9:00 AM	FT receives daily focus brief for each probation intern they manage
12:30 PM	Missed report check — noon window closes
5:30 PM	Missed report check — 5pm window closes
6:00 PM	FT receives signal questions for today's evaluation
6:30 PM (Wed)	Wednesday deliverable deadline check — W2 interns
12:00 PM (Fri)	Friday deliverable deadline check — W2 interns
8:00 PM	End-of-stage evaluation generated on Day 5 and Day 10
10:30 PM	Missed report check — 10pm window closes

All jobs skip Sunday automatically. Day numbers also skip Sunday in the count.

👤

FT Responsibilities in This System

▼ expand

The bot handles tracking and logistics. The FT handles judgment and accountability. Here is what the FT does and does not do.

Morning brief (9 AM): The bot sends the FT a focus note for the day — what to observe, what not to do. The FT reads it and acts accordingly.

Evening signal questions (6 PM): The bot asks 2–3 specific questions about today's intern behaviour. FT replies in the DM. Bot logs the answers as signals.

Day 4 depth check: FT runs a 15-minute check-in with the intern. Bot pre-generates the questions based on the architecture the intern submitted on Day 2.

Day 6 commitments: FT sets specific deliverables with specific dates. Bot logs them. Bot enforces them on Wednesday EOD and Friday noon.

Do NOT coach report quality mid-week. The bot notes quality. The FT does not comment on it. That removes the signal the report quality provides.

Do NOT chase missed reports. The bot handles warnings. The FT chasing defeats the purpose of the discipline filter.

🏢

Adding a New Department

▼ expand

The system supports multiple departments. Each department gets its own Slack channel. All interns across all departments are tracked in the same Google Sheet.

Create a Slack channel

e.g. #sales-eod or #hr-eod. Invite the intern-management-agent bot to it.

Get the channel ID

Right-click the channel in Slack → View channel details → Copy ID at the bottom (starts with C).

Add interns via this dashboard

Use the Add Intern button. Set department to the new department name. Set channel_id to the channel ID from step 2. The bot will start listening to that channel immediately.

🗄️

What Is Stored in Google Sheets

▼ expand

All data lives in a Google Sheet with 7 tabs. Nothing is deleted — only marked as eliminated or cleared. This gives you a full audit trail.

candidates — every intern's full record, stage, status, miss count.

reports_log — every EOD report submitted, with timestamp, quality score, and flags.

signals_log — every FT signal answer logged per day per intern.

warnings_log — every warning issued, when, what triggered it, exact message sent.

commitments — Week 2 deliverables, due dates, whether delivered on time.

evaluations — end-of-stage evaluations with signals, recommendations, HR confirmation status.

decisions_log — every final decision made — offer, elimination, extension — with who confirmed it and when.

Intern Pipeline

Intern Pipeline

All Interns

EOD Report Format

Miss and Warning Logic

Evaluations — What the Bot Reads

Automated Schedule — What Fires When

FT Responsibilities in This System

Adding a New Department

What Is Stored in Google Sheets

Add Intern

Remove Intern