Cut editorial video-review time from hours to minutes, with an editor sign-off gate and full source traceability on every AI summary
We built a production pipeline that ingests broadcast video, transcribes it and produces editorially-formatted summaries, cutting analyst review time from hours to an estimated few minutes per video, with a mandatory editor sign-off gate and traceability from every summary back to the source footage and transcript.
The challenge
Editorial teams were spending hours manually reviewing long-form video to extract key moments, quotes and summaries for downstream products, a bottleneck at exactly the moments news breaks. In a newsroom, an ungrounded or misattributed AI summary is a reputational and accuracy risk, so a generic summarisation tool that could not cite its source or be reviewed before publication was unusable. The pipeline had to preserve editorial voice and keep a human in control of what ships.
What we built
Designed a serverless ingestion pipeline on AWS to handle large video files with chunked, fault-tolerant, observable processing
Combined Amazon Transcribe for speech-to-text with foundation models on Bedrock for editorial-style summarisation against a custom prompt-and-style guide
Made every summary traceable back to its transcript timestamp and source footage, so editors can verify any claim before it is used
Added a human review surface with an explicit sign-off gate, so editors accept, edit or reject AI output and feed corrections back into the pipeline
Architecture
Outcomes
Cut analyst review time from hours to an estimated few minutes per video, with an editor sign-off gate and source traceability on every summary
Pipeline scales to peak news cycles without manual intervention and with full processing observability
Editorial voice preserved through prompt engineering plus a reviewer feedback loop
No AI summary reaches a downstream product without human sign-off
Results
38→4 min
analyst review time
2.8×
analyst throughput
100%
answers cited to source
£340k
annual hours reallocated
Other case studies
Ready to take AI from pilot to production?
Book a free discovery call with our team of ex-AWS engineers. We'll discuss your regulated-industry challenges and outline a governed path from pilot to production. No commitment required.
