Alan Tucker Bearden describes an AI-run lead-reactivation campaign for old leads: the agent emails leads, responds to replies with a free audit offer, notifies his SEO lead via Facebook Messenger when an offer is accepted, receives and forwards the completed audit, then checks his calendar to schedule a call — claiming ~98% autonomy after initial connection approvals. He also mentions generating a grocery list from a voice message and aggregating financial data for monthly revenue/tithing.
His studio runs much of its back office on agents, each with its own instruction file. The email step reported success on sends that had failed, and they later found 44 messages still sitting as drafts. Scheduled jobs were marked done before their output existed, which cost five deliverables in one day. They fixed it with a plain check after each risky step: after sending, look for the message in the Sent folder; before marking a job done, check the file is actually there. Now only a failed check needs a person's eyes.
This only partly worked. Worth knowing before you try it.
He had wanted an automatic provisioner for his boards for a long time but it was a whole project on its own, so he told the AI to write it. Some physical steps still need him, but each board now takes 2 to 3 minutes instead of a few stressful hours.
The AI goes through his sent emails to learn which sorts he typically responds to, pings him when it thinks he left something unanswered he'd want to answer, and repeats the exercise monthly to stay current.
Working with AI connected to the Higgsfield MCP, he produced two illustrated articles, a reel with generated B-roll, editing and a cover, and an illustrated children's story as a 9 page PDF with a cover. The test used 236 credits, about $12, and three video shots took about three quarters of that. He also shows the connection mistake he made at the start.
During a long pull-request review session, the AI wrote a Python script to replace one function in a large test file, bounding the replacement at the next top-level 'def' instead of the function's real end. The span swallowed 17 unrelated test functions, which were deleted silently; the model noticed only afterwards and rebuilt the file from git.
This only partly worked. Worth knowing before you try it.
Since July 2026 a Japanese creator has run a political clip channel with a pipeline the AI built: it watches source channels, fetches new videos, transcribes them locally, picks highlights from voice peaks and wording, fixes subtitle typos and names, makes thumbnails and titles, exports long videos and Shorts, uploads them unlisted, then collects view and retention data to feed the next round. In 3.5 months it shipped 319 videos for about 1.05 million total views and 1,080 subscribers, though the median long video reached only 268 views; he publishes the failures too, including the agent claiming 'fixed' when it was not.
He pointed the AI at his JavaScript project and asked for a full TypeScript migration without changing behavior. 236 commits and a full day later the project compiled clean, the linter passed, and 55 of 56 tests passed, with the one failure traced to a commit where a value was accidentally hardcoded.
An LLM crunched for 36 hours on a project he had wanted for 5 years; checked and ready to deploy, work he estimates would have taken him about 9 months.