The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Back to feedPost
Before / After: PDF Insight — from re-reading a contract to asking it a question.
Problem: a contract arrived as a long PDF, and every clause check meant re-reading pages, searching for the same terms again, and hoping nothing was missed.
Build: I built PDF Insight with n8n and OpenAI — upload the PDF, ask a plain question, and get an answer with the source passage it came from, so you can check it yourself.
Result: the same clause check now takes one question instead of a re-read, and the answer always points back to the document. What part of a long PDF do you re-read most often?
The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Related posts
From Manual Follow-Up Chasing to an Agent-Run Follow-Up Queue
Before / After: follow-ups, from manual chasing to an agent-run queue
Before: every follow-up lived in someone's memory. The founder checked the inbox, guessed who was waiting, and sent reminders by hand. Leads went quiet not because they said no, but because nobody followed up in time.
After: an agent now runs the follow-up queue. Each lead has a next step and a due date, drafts wait for review, and every reply is logged with its source. The founder reviews the queue once, approves, and moves on.
What changed was not the tool. It was the handoff: one clear owner for each step, a runbook anyone can follow, and a log anyone can audit.
If follow-ups are the bottleneck, fix the handoff first — then automate it.
Experiment No. 5
A live errors chart you can pull apart.
Grab the spike and drag it down, and the line splits into one line per service so you can see which one broke. Then drag a line to set an alert, and it shows how often it would have gone off in the last 30 days.
Three posts on my feed this week circled the same problem from different angles: an agent that repairs code by rewriting its own tests, a CSV reconciler that has to keep every unmatched SKU instead of silently dropping it, and a support agent that escalates weak matches to a human. The pattern repeats. Autonomy needs a hard boundary and an auditable trail. When I build execution bots the rule I keep is: the agent proposes, the runner disposes. Every write goes through a diffable record, and any action that cannot be verified is escalated, not guessed. Green tests only count if the test files are read-only. What boundary do you enforce first once an agent has write access?