--- schema: 1 type: video title: 'The Agent Works. Does Anyone Trust It? — Michele Grant, Leading at Scale' description: >- An agent's reports match the analysts' work during a pilot, yet the steering committee still has to decide whether it should operate in a core business process. Michele Grant uses an illustrative acquisition integration scenario to examine that decision. The question moves beyond technical accuracy to measurable value, accountable ownership and evidence for greater authority. A faster report is useful only if it changes a decision or releases capacity that the organization actually redeploys. Her framework starts with a baseline for reporting time, analyst capacity, operating cost and the detection of material issues, so an attractive demo does not become its own business case. Grant separates the people who operate the workflow, oversee its risk and provide independent assurance. A Human Works Standard records what the agent may do, where human judgment remains necessary and who must intervene when something goes wrong. An auditable decision log, a maintained inventory of agents and dependencies, and an autonomy scorecard make those boundaries reviewable. Financial calculations stay in deterministic code while the model helps with narrative and interpretation. Authority advances through shadow operation, reviewed drafts, exception handling with sampling and finally exception focused review supported by independent checks. At that last stage, missed issues matter as much as flagged ones. A falling override rate can reflect weaker review rather than better performance. The acquisition values and capacity examples explain the operating framework; they are not disclosed results from a completed deployment or proof that the agent created the projected synergies. Speaker info: - https://www.linkedin.com/in/michelegrantleadership/ Timestamps: 00:00 - From a working agent to organizational trust 2:34 - An illustrative acquisition integration scenario 4:47 - Measure value against a baseline 6:29 - Accountability across three lines of ownership 8:14 - The Human Works Standard 9:54 - Decision logs and the agent inventory 11:51 - Four stages of autonomy 12:50 - Independent sampling and missed issues 13:29 - What the framework lets leaders decide canonical: >- https://liquidgold.cc/channel/aidotengineer/explore/video/the-agent-works-does-anyone-trust-it-michele-grant-f1c09cf05adc markdown: >- https://liquidgold.cc/channel/aidotengineer/video/the-agent-works-does-anyone-trust-it-michele-grant-f1c09cf05adc.md llms: null parent: 'https://liquidgold.cc/channel/aidotengineer.md' channel: aidotengineer updated: '2026-10-23T22:18:46.817Z' language: null children: [] videoId: '-a2oTS9J67U' durationSeconds: 905 publishedAt: '2026-10-13T16:00:11.000Z' searchChunks: - ordinal: 0 text: null startSeconds: 2 endSeconds: 12 watch: 'https://liquidgold.cc/watch/-a2oTS9J67U?t=2' heldReason: rights - ordinal: 1 text: null startSeconds: 12 endSeconds: 158 watch: 'https://liquidgold.cc/watch/-a2oTS9J67U?t=12' heldReason: rights - ordinal: 2 text: null startSeconds: 12 endSeconds: 158 watch: 'https://liquidgold.cc/watch/-a2oTS9J67U?t=12' heldReason: rights - ordinal: 3 text: null startSeconds: 12 endSeconds: 158 watch: 'https://liquidgold.cc/watch/-a2oTS9J67U?t=12' heldReason: rights - ordinal: 4 text: null startSeconds: 12 endSeconds: 158 watch: 'https://liquidgold.cc/watch/-a2oTS9J67U?t=12' heldReason: rights - ordinal: 5 text: null startSeconds: 171 endSeconds: 293 watch: 'https://liquidgold.cc/watch/-a2oTS9J67U?t=171' heldReason: rights - ordinal: 6 text: null startSeconds: 171 endSeconds: 293 watch: 'https://liquidgold.cc/watch/-a2oTS9J67U?t=171' heldReason: rights - ordinal: 7 text: null startSeconds: 293 endSeconds: 305 watch: 'https://liquidgold.cc/watch/-a2oTS9J67U?t=293' heldReason: rights - ordinal: 8 text: null startSeconds: 305 endSeconds: 396 watch: 'https://liquidgold.cc/watch/-a2oTS9J67U?t=305' heldReason: rights - ordinal: 9 text: null startSeconds: 305 endSeconds: 396 watch: 'https://liquidgold.cc/watch/-a2oTS9J67U?t=305' heldReason: rights - ordinal: 10 text: null startSeconds: 305 endSeconds: 396 watch: 'https://liquidgold.cc/watch/-a2oTS9J67U?t=305' heldReason: rights - ordinal: 11 text: null startSeconds: 396 endSeconds: 408 watch: 'https://liquidgold.cc/watch/-a2oTS9J67U?t=396' heldReason: rights - ordinal: 12 text: null startSeconds: 408 endSeconds: 497 watch: 'https://liquidgold.cc/watch/-a2oTS9J67U?t=408' heldReason: rights - ordinal: 13 text: null startSeconds: 497 endSeconds: 508 watch: 'https://liquidgold.cc/watch/-a2oTS9J67U?t=497' heldReason: rights - ordinal: 14 text: null startSeconds: 508 endSeconds: 600 watch: 'https://liquidgold.cc/watch/-a2oTS9J67U?t=508' heldReason: rights - ordinal: 15 text: null startSeconds: 611 endSeconds: 722 watch: 'https://liquidgold.cc/watch/-a2oTS9J67U?t=611' heldReason: rights - ordinal: 16 text: null startSeconds: 611 endSeconds: 722 watch: 'https://liquidgold.cc/watch/-a2oTS9J67U?t=611' heldReason: rights - ordinal: 17 text: null startSeconds: 611 endSeconds: 722 watch: 'https://liquidgold.cc/watch/-a2oTS9J67U?t=611' heldReason: rights - ordinal: 18 text: null startSeconds: 722 endSeconds: 734 watch: 'https://liquidgold.cc/watch/-a2oTS9J67U?t=722' heldReason: rights - ordinal: 19 text: null startSeconds: 778 endSeconds: 790 watch: 'https://liquidgold.cc/watch/-a2oTS9J67U?t=778' heldReason: rights - ordinal: 20 text: null startSeconds: 790 endSeconds: 811 watch: 'https://liquidgold.cc/watch/-a2oTS9J67U?t=790' heldReason: rights - ordinal: 21 text: null startSeconds: 811 endSeconds: 823 watch: 'https://liquidgold.cc/watch/-a2oTS9J67U?t=811' heldReason: rights --- # The Agent Works. Does Anyone Trust It? — Michele Grant, Leading at Scale An agent's reports match the analysts' work during a pilot, yet the steering committee still has to decide whether it should operate in a core business process. Michele Grant uses an illustrative acquisition integration scenario to examine that decision. The question moves beyond technical accuracy to measurable value, accountable ownership and evidence for greater authority. A faster report is useful only if it changes a decision or releases capacity that the organization actually redeploys. Her framework starts with a baseline for reporting time, analyst capacity, operating cost and the detection of material issues, so an attractive demo does not become its own business case. Grant separates the people who operate the workflow, oversee its risk and provide independent assurance. A Human Works Standard records what the agent may do, where human judgment remains necessary and who must intervene when something goes wrong. An auditable decision log, a maintained inventory of agents and dependencies, and an autonomy scorecard make those boundaries reviewable. Financial calculations stay in deterministic code while the model helps with narrative and interpretation. Authority advances through shadow operation, reviewed drafts, exception handling with sampling and finally exception focused review supported by independent checks. At that last stage, missed issues matter as much as flagged ones. A falling override rate can reflect weaker review rather than better performance. The acquisition values and capacity examples explain the operating framework; they are not disclosed results from a completed deployment or proof that the agent created the projected synergies. Speaker info: - https://www.linkedin.com/in/michelegrantleadership/ Timestamps: 00:00 - From a working agent to organizational trust 2:34 - An illustrative acquisition integration scenario 4:47 - Measure value against a baseline 6:29 - Accountability across three lines of ownership 8:14 - The Human Works Standard 9:54 - Decision logs and the agent inventory 11:51 - Four stages of autonomy 12:50 - Independent sampling and missed issues 13:29 - What the framework lets leaders decide Canonical: https://liquidgold.cc/channel/aidotengineer/explore/video/the-agent-works-does-anyone-trust-it-michele-grant-f1c09cf05adc · Channel: [AI Engineer](https://liquidgold.cc/channel/aidotengineer.md) > Search chunk text is held until public-source provenance is verified. ## Document chunks --- #### Chunk 0: Document excerpt (0:02 to 2:38) > Held: transcript rights do not permit this chunk's text. --- #### Chunk 1: Document excerpt (2:38 to 4:53) > Held: transcript rights do not permit this chunk's text. --- #### Chunk 2: Document excerpt (4:53 to 6:36) > Held: transcript rights do not permit this chunk's text. --- #### Chunk 3: Document excerpt (6:36 to 8:17) > Held: transcript rights do not permit this chunk's text. --- #### Chunk 4: Document excerpt (8:17 to 10:00) > Held: transcript rights do not permit this chunk's text. --- #### Chunk 5: Document excerpt (10:00 to 12:02) > Held: transcript rights do not permit this chunk's text. --- #### Chunk 6: Document excerpt (12:02 to 12:58) > Held: transcript rights do not permit this chunk's text. --- #### Chunk 7: Document excerpt (12:58 to 13:31) > Held: transcript rights do not permit this chunk's text. --- #### Chunk 8: Document excerpt (13:31 to 15:05) > Held: transcript rights do not permit this chunk's text.