Yearly running cost Assuming 12,000 sessions a year
By hand $390K / year 6,000 hours of work
Subscription n.a.
This workflow $2.5K / year Machine usage only; setup, hosting and review are extra. Do you want issue drafts or also connect your issue tracker? For example, turn a player's report of a stuck door into an issue draft linked to that moment in the recording.
One session includes up to 10 recorded minutes and 30 sampled evidence frames.
Analyze supplied sessions Review recorded sessions and prepare issue drafts $1,512 – 4,320 setup Connect approved issue tracking Send approved evidence to one issue tracker $1,944 – 5,520 setup
Which system should it connect to?Jira review project Existing issue tracker A QA reviewer approves shared tickets and severity. Use consented recordings and retain source timestamps. No automatic bug fixes, release approval, player profiling or claims that an observation proves a defect.
What is included Import consented footage and transcripts, group observations and prepare evidence-linked issue drafts.
Import consented recordings and transcripts, group related comments and link issue drafts to timestamps. Connect an issue tracker to create approved tickets and check for duplicates.
Transcript scope: up to 3,000 words.
Build scope: up to 1 game build per session.
Issue-tracker scope: up to 1 approved project.
Reference application CPU: 4 cores.
Reference application memory: 8 GB.
Standard independent test set: at least 30 sessions.
Strict independent test set: at least 60 sessions.
What standard do you need? Choose the acceptance targets your provider must demonstrate on your own test material.
Standard Test the defined workload and controls. Labeled issues covered: ≥85% Useful issue groups: ≥90% Included Strict Tighter targets and a larger independent test set. Labeled issues covered: ≥92% Useful issue groups: ≥95% Setup +$360 – 1,200 My own requirements Agree different targets and scope with your provider. Quote separately
How these standards are measured Use held-out sessions from distinct players and builds with repeated comments, ambiguous footage, missing audio, contradictory feedback and redaction cases. Independently label issue groups and evidence links; evaluate footage timestamps rather than plausible-sounding summaries. Agree the holdout before tuning. Report each metric with its numerator, denominator and exclusions. Count failed and rejected attempts, not just accepted outputs.
Targets for your selected standard What is checked Target Labeled issues coveredIndependently labeled actionable issues represented in the review queue divided by all labeled actionable issues. ≥85% Useful issue groupsReviewer-confirmed actionable groups divided by all generated issue groups, including duplicates and unsupported claims. ≥90% Correct evidence linksIssue links pointing to the correct session and supporting timestamp divided by all generated evidence links. ≥95% Consent and sharing controlsPassed consent, redaction and approved-sharing cases divided by all tested control cases; any unapproved sharing fails acceptance. ≥100%
Proposed targets apply to an independently labeled test set on the agreed workload, not every possible future input. Review each pilot result. Any unauthorized action or sensitive-data exposure fails acceptance.
How quickly do you need each session? Choose a processing-time target for the agreed reference workload; confirm feasibility with your provider.
Within 5 minutes Included Within 4 minutes Setup +$144 – 480 Within 3 minutes Setup +$288 – 840
Timing details Measure from an accepted recording to reviewable issue drafts, including audio processing, frame sampling, queueing and permitted retries. Recording length is the workload, not the time allowed for human review.
The target applies to at least 95% of agreed test runs, with 1 in progress at a time.
How much should each session cost? Choose a machine-cost cap for the agreed workload. A lower cap needs testing without relaxing the selected quality targets.
Up to $0.40 Included Up to $0.35 Setup +$144 – 480 Up to $0.30 Setup +$288 – 840
Cost details Includes transcription, bounded text and image-context analysis, retries and allocated hosting. Continuous video streaming, full-frame model analysis, capture SDK licenses and staff review are separate.
Reference machine cost: $0.24 – 0.27 per session at 1000 sessions a month. The selected cap is a target to test, not a replacement for this estimate.
Where do you want it to run? Run the workflow in your cloud or on your own machine. Choose separately whether it may use an external AI model.
Run it onYour cloud Local environment AI model accessApproved model API No external model calls
Data and access details Run in your controlled cloud environment with scoped access and retained review records.
Use an approved external model with the agreed data-processing terms and minimum necessary input.
Local hosting does not itself keep data offline. Approved business-system connections may still use the network. The reference input budget includes billed image tokens for sampled frames; it is not a promise to understand every moment of the recording. An existing authorized transcript removes speech usage but does not expand the video scope. Private running cost must be measured in the pilot.
How do you want to use it? Choose a browser interface, an MCP tool for your current agent, or a standalone agent.
Web page Inspect each session in a private browser interface. Included Your AI agent Call the workflow through MCP; the same access and approval rules apply. Setup +$72 – 360 Standalone agent Use a dedicated agent with the agreed triggers and controls. Setup +$144 – 480