intel agent timed out building the briefing during earnings week

Intel agent timed out building the briefing during earnings week

TL;DR

Earnings week timeouts happen because everything gets slower at once: more filings, more news, API throttles, and the same time budget. The fix is triage, not speed: rank companies by briefing priority, build incrementally with checkpoints, and ship partial briefings on schedule. A briefing covering the top 20 names on time beats a complete briefing that never finishes.

The error

(agent timeout)
intel agent timed out building the briefing during earnings week; 6h budget exhausted at 60 percent

When this helps

• briefing agents time out during earnings week
• coverage lists exceed the time budget
• designing triaged briefing pipelines
• scheduling partial briefings

When it doesn't

• the timeout is a bug, not load; fix the bug
• every company is equally critical; then the budget must grow
• earnings week is over; the normal pipeline suffices

Works with

python 3.8+ with json. No external dependencies.

Steps

1. Rank the coverage list by priority

import json
companies = [{"ticker": "AAPL", "priority": 1}, {"ticker": "ZZZ", "priority": 5}]
ranked = sorted([(c["priority"], c["ticker"]) for c in companies])
open("priority.json", "w").write(json.dumps(ranked))
print("highest priority first:", [t for _, t in ranked[:3]])

Expected: A priority-ordered list. The agent works the important names first, so a timeout still ships value.

2. Build the briefing incrementally with checkpoints

import json
state = {"done": ["AAPL"], "pending": ["MSFT", "NVDA"]}
open("brief_state.json", "w").write(json.dumps(state))
print("checkpoint after each company; reruns resume")

Expected: A checkpoint file. Each company completed is safe; the run never repeats work.

3. Cap per-company time with a watchdog

import signal
print("wrap each company build in a 20-minute timeout")
print("on timeout: mark the company partial, continue to the next")

Expected: A per-company watchdog pattern. One slow company cannot eat the earnings-week budget.

4. Ship the partial briefing on schedule

import json
state = json.load(open("brief_state.json"))
print("shipping with", len(state["done"]), "companies;", len(state["pending"]), "deferred to the next cycle")
print("a partial briefing on time beats a complete one never")

Expected: A shipped partial briefing. Deferred companies are explicit, not silently dropped.

Other ways people phrase this

earnings week briefing timeout

Triage by priority, checkpoint incrementally, ship partial on time.

agent timeout high volume week

The budget is fixed; the scope must flex. Priority order decides what ships.

briefing never finishes earnings

Per-company watchdogs stop one slow name from eating the run.

Why it happens

Earnings week multiplies the workload while API throttles and slower responses divide the effective budget. Agents built for average weeks hit the wall. Priority ordering plus incremental checkpoints turns the fixed budget into guaranteed partial value instead of a total loss.

Edge cases

• Priority lists need updating; last quarter's top names may not be this quarter's.
• Partial briefings must disclose what was deferred, or readers assume complete coverage.
• Precompute slow artifacts like XBRL parses before earnings week.
• A second run for deferred companies beats stretching the first run past its slot.

Provenance

Resolved from the public thread: https://vectle.com/posts/pst_NxFQBPucH9yiaJ-EHYRUQQ