A personal audio digest delivered through Telegram
@mielke_jack uses Claude Code to combine personal context and live research into Telegram briefings, a Notion entry, an HTML reading page and topic-based audio.
x.com
Original prompt
Shared by the source author. See source notes for context.
Build me one of these, including the personal podcast I can listen to on walks or runs
Reach people exploring AI agents
See placements →Evidence at a glance
Sourced account- Source poster / publisher
- @mielke_jack
- Original account
- Read the source ↗
- Published
- Source link checked
- Reported result
- Reported by the poster: Claude Code delivers Telegram briefings at 6am and 5pm with roughly 35 minutes total, plus a Notion entry, HTML page and 12 audio tracks. These are reported delivery and listening details, not independently observed runs.
- Human involvement
- The poster supplied broad personal context and gives feedback. The public repository omits private account readers; connector access and safeguards must be configured separately, not assumed from the shared skill.
A linked account supports attribution. It does not mean we independently reproduced its result.
Report an issue with this case
Open correction emailOpens your email app with the case link included. Nothing is sent automatically.
How the agent did it
- The poster connected Claude Code to personal sources including messages, meeting transcripts, code activity, calendars and fitness data.
- The agent reads relevant context, reviews previous digests to avoid repeating news, and researches updates tailored to the poster's interests.
- It delivers a Telegram summary, a Notion digest entry, an HTML reading page and 12 audio tracks; the poster supplies feedback to improve later editions.
@mielke_jack describes a personalised briefing system built around Claude Code and Telegram. It combines context from conversations, meeting transcripts, code activity, calendars and fitness sources with live research about subjects the poster cares about. Previous digests are reviewed to reduce repeated news.
The outputs are more than a chat summary. The post names a Notion daily-digest entry, an HTML reading page, a texted summary and 12 audio tracks that can be selected by topic. Delivery is reported by the poster at 6am and 5pm, with roughly 35 minutes total. The duration is not presented as the length of each edition.
Feedback is part of the workflow. The poster says later editions can improve from replies, and the linked public repository describes reading feedback before writing the next digest and recording corrections in the skill. A same-author thread reply identifies access to messages and calendar events as particularly useful for catching buried information.
The repository offers the method, delivery scripts and an example, not turnkey access to someone else's accounts. It explicitly leaves out private account identifiers and the readers for the author's own accounts. The described outputs and benefits remain reported by the poster; this entry does not claim that Agents In Use reproduced the system or verified its overnight actions.
Prompt
Original prompt
Source: X · @mielke_jack
Frequently asked questions
How long is the personalised podcast?
The post says roughly 35 minutes total alongside delivery at 6am and 5pm. It does not say each edition lasts that long.
What does it deliver besides audio?
The poster names a Notion digest entry, an HTML reading page and a texted chat summary delivered through Telegram.
Does the shared repository include account access?
No. Its README excludes private account identifiers and the readers for the author's accounts. Your own connectors require separate setup.
More details
About this account
The agent selects and synthesises material across connected sources, researches relevant news and adapts subsequent digests using the poster's feedback.
Primary post by @mielke_jack: https://x.com/mielke_jack/status/2106460142699151477. Post, bio and available replies checked through public FxTwitter syndication; returned replies are not guaranteed exhaustive. Results are reported by the poster, not independently reproduced. Original prompt excerpt (not the full setup): quoted exactly from the primary post's request to recreate the linked daily-digest project. Linked public README checked: https://github.com/jackmielke/daily-digest. It documents the feedback loop and omits private account readers. Its current listening totals are not substituted for the post figures.
Credit: @mielke_jack on X
Source published
Added to Agents In Use
Source checked
Tools and topics
Claude Code · Telegram · Notion · Granola · Wispr Flow · GitHub · Garmin · Strava
personal digest · audio · Telegram
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.