Decide in the open https://craigstanley.work/podcast/ en-gb © 2026 Craig Stanley Short episodes that explain one idea from craigstanley.work at a time: how work really gets done, how to make repeated decisions better and cheaper, what Microsoft's AI tools can do, and how to keep risk and cost honest. Two hosts talk it through. One explains, the other asks the questions a sceptical head of IT or finance would ask. Scripts are written with Google Gemini from the site's own text and read by synthetic British voices. Short episodes that explain one idea from craigstanley.work at a time: how work really gets done, how to make repeated decisions better and cheaper, what Microsoft's AI tools can do, and how to keep risk and cost honest. Two hosts talk it through. One explains, the other asks the questions a sceptical head of IT or finance would ask. Scripts are written with Google Gemini from the site's own text and read by synthetic British voices. One idea from craigstanley.work, explained in under five minutes. Craig Stanley Craig Stanleymail@craigstanley.me https://craigstanley.work/media/podcast/cover.jpgDecide in the openhttps://craigstanley.work/podcast/ false episodic Sat, 10 Oct 2026 09:00:00 GMT Is Microsoft 365 Copilot worth it? Anna and Tom examine how to calculate the break-even threshold for Microsoft 365 Copilot seats. They discuss why repeated daily tasks pay back while broad rollouts usually fail. Anna and Tom examine how to calculate the break-even threshold for Microsoft 365 Copilot seats. They discuss why repeated daily tasks pay back while broad rollouts usually fail.

The page this episode explains: Is Microsoft 365 Copilot worth it?

Voices are synthetic, made with Google Gemini text-to-speech. Full transcript: https://craigstanley.work/media/podcast/is-microsoft-copilot-worth-it/transcript.txt

]]> https://craigstanley.work/podcast/#ep-30 craigstanley-work-podcast-is-microsoft-copilot-worth-it Sat, 10 Oct 2026 09:00:00 GMT 213 30 full false Tracking Microsoft roadmaps without the guesswork Anna and Tom examine the Microsoft 365 roadmap explorer and discuss how to verify supplier release targets before allocating budget. They also look at Craig Stanley's public site plan, covering upcoming tools for Foundry and threshold calculators. Anna and Tom examine the Microsoft 365 roadmap explorer and discuss how to verify supplier release targets before allocating budget. They also look at Craig Stanley's public site plan, covering upcoming tools for Foundry and threshold calculators.

The page this episode explains: Roadmap

Voices are synthetic, made with Google Gemini text-to-speech. Full transcript: https://craigstanley.work/media/podcast/roadmap/transcript.txt

]]> https://craigstanley.work/podcast/#ep-29 craigstanley-work-podcast-roadmap Sat, 10 Oct 2026 09:00:00 GMT 222 29 full false Set the budget weekly, move the underspend on Friday Anna and Tom discuss moving AI consumption allowances weekly, running a Friday review to reallocate underspend, and automating the process with Power Automate. Anna and Tom discuss moving AI consumption allowances weekly, running a Friday review to reallocate underspend, and automating the process with Power Automate.

The page this episode explains: Set the budget weekly, move the underspend on Friday

Voices are synthetic, made with Google Gemini text-to-speech. Full transcript: https://craigstanley.work/media/podcast/cost--underspend-pooling--set-the-budget-weekly-move-the-underspend-on-friday/transcript.txt

]]> https://craigstanley.work/podcast/#ep-28 craigstanley-work-podcast-cost--underspend-pooling--set-the-budget-weekly-move-the-underspend-on-friday Sat, 10 Oct 2026 09:00:00 GMT 198 28 full false The three costs of AI at work Anna and Tom look at the three costs of AI at work: licences, consumption, and people's time. They discuss measuring cost per active user, cost per decision, and how to track the total cost per decision supported using the page on craigstanley.work. Anna and Tom look at the three costs of AI at work: licences, consumption, and people's time. They discuss measuring cost per active user, cost per decision, and how to track the total cost per decision supported using the page on craigstanley.work.

The page this episode explains: The three costs of AI at work

Voices are synthetic, made with Google Gemini text-to-speech. Full transcript: https://craigstanley.work/media/podcast/cost--start-here--the-three-costs-of-ai-at-work/transcript.txt

]]> https://craigstanley.work/podcast/#ep-27 craigstanley-work-podcast-cost--start-here--the-three-costs-of-ai-at-work Sat, 10 Oct 2026 09:00:00 GMT 216 27 full false Counting the three costs of AI Anna and Tom discuss why tracking seat licences alone causes AI budgets to overspend. They examine consumption, people's time, weekly reviews, and moving unused allowances to people at their limit. Anna and Tom discuss why tracking seat licences alone causes AI budgets to overspend. They examine consumption, people's time, weekly reviews, and moving unused allowances to people at their limit.

The page this episode explains: Cost

Voices are synthetic, made with Google Gemini text-to-speech. Full transcript: https://craigstanley.work/media/podcast/cost/transcript.txt

]]> https://craigstanley.work/podcast/#ep-26 craigstanley-work-podcast-cost Sat, 10 Oct 2026 09:00:00 GMT 212 26 full false Five AI risk types for practical registers Anna and Tom look at categorising workplace AI risks into data, decision, people, supplier and cost. They examine the controls and ownership needed for each category. Anna and Tom look at categorising workplace AI risks into data, decision, people, supplier and cost. They examine the controls and ownership needed for each category.

The page this episode explains: Data, decision, people, supplier, cost: five risk types

Voices are synthetic, made with Google Gemini text-to-speech. Full transcript: https://craigstanley.work/media/podcast/risk--risk-map--data-decision-people-supplier-cost-five-risk-types/transcript.txt

]]> https://craigstanley.work/podcast/#ep-25 craigstanley-work-podcast-risk--risk-map--data-decision-people-supplier-cost-five-risk-types Sat, 10 Oct 2026 09:00:00 GMT 211 25 full false Why AI risk registers should list decisions, not models Generic model risks like hallucination and bias fail to prioritise action. Rebuilding risk registers around decisions shows who is accountable and what a wrong answer costs. Generic model risks like hallucination and bias fail to prioritise action. Rebuilding risk registers around decisions shows who is accountable and what a wrong answer costs.

The page this episode explains: Most AI risk registers list the model, not the decision

Voices are synthetic, made with Google Gemini text-to-speech. Full transcript: https://craigstanley.work/media/podcast/risk--start-here--most-ai-risk-registers-list-the-model-not-the-decision/transcript.txt

]]> https://craigstanley.work/podcast/#ep-24 craigstanley-work-podcast-risk--start-here--most-ai-risk-registers-list-the-model-not-the-decision Sat, 10 Oct 2026 09:00:00 GMT 202 24 full false Mapping AI risk to workplace decisions Most AI risk registers focus on the model rather than the decision. Anna and Tom discuss mapping risks to daily work, tracking reversibility, and using automated nudges. Most AI risk registers focus on the model rather than the decision. Anna and Tom discuss mapping risks to daily work, tracking reversibility, and using automated nudges.

The page this episode explains: Risk

Voices are synthetic, made with Google Gemini text-to-speech. Full transcript: https://craigstanley.work/media/podcast/risk/transcript.txt

]]> https://craigstanley.work/podcast/#ep-23 craigstanley-work-podcast-risk Sat, 10 Oct 2026 09:00:00 GMT 217 23 full false Mapping Microsoft capabilities to decisions Anna and Tom examine how Microsoft's AI tools sort into four jobs for decisions, from finding information to scoring set options. Anna and Tom examine how Microsoft's AI tools sort into four jobs for decisions, from finding information to scoring set options.

The page this episode explains: Capabilities

Voices are synthetic, made with Google Gemini text-to-speech. Full transcript: https://craigstanley.work/media/podcast/capabilities/transcript.txt

]]> https://craigstanley.work/podcast/#ep-22 craigstanley-work-podcast-capabilities Sat, 10 Oct 2026 09:00:00 GMT 235 22 full false The decision record schema Anna and Tom examine the decision record schema from craigstanley.work. They discuss using a short, queryable format to capture choices, measure calibration, and review decision quality without outcome bias. Anna and Tom examine the decision record schema from craigstanley.work. They discuss using a short, queryable format to capture choices, measure calibration, and review decision quality without outcome bias.

The page this episode explains: The decision record schema

Voices are synthetic, made with Google Gemini text-to-speech. Full transcript: https://craigstanley.work/media/podcast/decisions--decision-records--the-decision-record-schema/transcript.txt

]]> https://craigstanley.work/podcast/#ep-21 craigstanley-work-podcast-decisions--decision-records--the-decision-record-schema Sat, 10 Oct 2026 09:00:00 GMT 228 21 full false Act, ask a person, or stop Anna and Tom discuss the three routes for decision models: autonomous action above a calibrated threshold, human review in the uncertain band, and a hard stop when data is missing or rules fire. Anna and Tom discuss the three routes for decision models: autonomous action above a calibrated threshold, human review in the uncertain band, and a hard stop when data is missing or rules fire.

The page this episode explains: Act, ask a person, or stop

Voices are synthetic, made with Google Gemini text-to-speech. Full transcript: https://craigstanley.work/media/podcast/decisions--decision-models--act-ask-a-person-or-stop/transcript.txt

]]> https://craigstanley.work/podcast/#ep-20 craigstanley-work-podcast-decisions--decision-models--act-ask-a-person-or-stop Sat, 10 Oct 2026 09:00:00 GMT 196 20 full false What a decision model is, and isn't Anna and Tom discuss decision models, why fixed outputs make them cheap and auditable, and where structured scoring fits operational decisions. Anna and Tom discuss decision models, why fixed outputs make them cheap and auditable, and where structured scoring fits operational decisions.

The page this episode explains: What a decision model is, and isn't

Voices are synthetic, made with Google Gemini text-to-speech. Full transcript: https://craigstanley.work/media/podcast/decisions--decision-models--what-a-decision-model-is-and-isn-t/transcript.txt

]]> https://craigstanley.work/podcast/#ep-19 craigstanley-work-podcast-decisions--decision-models--what-a-decision-model-is-and-isn-t Sat, 10 Oct 2026 09:00:00 GMT 237 19 full false Anchoring in estimates Anna and Tom look at how anchoring distorts project, budget and AI-benefit estimates, and discuss the practical steps that help counter the bias. Anna and Tom look at how anchoring distorts project, budget and AI-benefit estimates, and discuss the practical steps that help counter the bias.

The page this episode explains: Anchoring in estimates

Voices are synthetic, made with Google Gemini text-to-speech. Full transcript: https://craigstanley.work/media/podcast/decisions--bias--anchoring-in-estimates/transcript.txt

]]> https://craigstanley.work/podcast/#ep-18 craigstanley-work-podcast-decisions--bias--anchoring-in-estimates Sat, 10 Oct 2026 09:00:00 GMT 232 18 full false Outcome bias and judging decisions by luck Anna and Tom discuss why judging decisions solely by results distorts reviews and damages organisations. They look at how recording reasoning beforehand helps evaluate choices fairly. Anna and Tom discuss why judging decisions solely by results distorts reviews and damages organisations. They look at how recording reasoning beforehand helps evaluate choices fairly.

The page this episode explains: Outcome bias: judging the decision by the luck

Voices are synthetic, made with Google Gemini text-to-speech. Full transcript: https://craigstanley.work/media/podcast/decisions--bias--outcome-bias-judging-the-decision-by-the-luck/transcript.txt

]]> https://craigstanley.work/podcast/#ep-17 craigstanley-work-podcast-decisions--bias--outcome-bias-judging-the-decision-by-the-luck Sat, 10 Oct 2026 09:00:00 GMT 164 17 full false Why everyone pads their estimates Anna and Tom look at why teams routinely add buffers to their project plans. They discuss how changing the incentives around pooled contingency and ranges can fix the problem. Anna and Tom look at why teams routinely add buffers to their project plans. They discuss how changing the incentives around pooled contingency and ranges can fix the problem.

The page this episode explains: Why everyone pads their estimates

Voices are synthetic, made with Google Gemini text-to-speech. Full transcript: https://craigstanley.work/media/podcast/decisions--game-theory--why-everyone-pads-their-estimates/transcript.txt

]]> https://craigstanley.work/podcast/#ep-16 craigstanley-work-podcast-decisions--game-theory--why-everyone-pads-their-estimates Sat, 10 Oct 2026 09:00:00 GMT 209 16 full false Calibrating confidence in models and teams Anna and Tom explain why probability scores must match actual hit rates before thresholds and decisions can be trusted. They look at simple ways to test calibration and how to correct both models and human estimates. Anna and Tom explain why probability scores must match actual hit rates before thresholds and decisions can be trusted. They look at simple ways to test calibration and how to correct both models and human estimates.

The page this episode explains: Calibration: are your 80% calls right 80% of the time?

Voices are synthetic, made with Google Gemini text-to-speech. Full transcript: https://craigstanley.work/media/podcast/decisions--decision-theory--calibration-are-your-80-calls-right-80-of-the-time/transcript.txt

]]> https://craigstanley.work/podcast/#ep-15 craigstanley-work-podcast-decisions--decision-theory--calibration-are-your-80-calls-right-80-of-the-time Sat, 10 Oct 2026 09:00:00 GMT 227 15 full false Setting a threshold you can defend How to set an action threshold where the cost of a false positive equals that of a false negative, adjusted for review capacity. How to set an action threshold where the cost of a false positive equals that of a false negative, adjusted for review capacity.

The page this episode explains: Setting a threshold you can defend

Voices are synthetic, made with Google Gemini text-to-speech. Full transcript: https://craigstanley.work/media/podcast/decisions--decision-theory--setting-a-threshold-you-can-defend/transcript.txt

]]> https://craigstanley.work/podcast/#ep-14 craigstanley-work-podcast-decisions--decision-theory--setting-a-threshold-you-can-defend Sat, 10 Oct 2026 09:00:00 GMT 198 14 full false When is more information worth paying for? Anna and Tom look at the expected value of information in decision theory. Using a software licence renewal example, they explain when analysis is worth paying for and when to act immediately. Anna and Tom look at the expected value of information in decision theory. Using a software licence renewal example, they explain when analysis is worth paying for and when to act immediately.

The page this episode explains: When is more information worth paying for?

Voices are synthetic, made with Google Gemini text-to-speech. Full transcript: https://craigstanley.work/media/podcast/decisions--decision-theory--when-is-more-information-worth-paying-for/transcript.txt

]]> https://craigstanley.work/podcast/#ep-13 craigstanley-work-podcast-decisions--decision-theory--when-is-more-information-worth-paying-for Sat, 10 Oct 2026 09:00:00 GMT 206 13 full false Expected value on one page Anna and Tom examine expected value using a worked service desk example, and look at the two places where the calculation can mislead. Anna and Tom examine expected value using a worked service desk example, and look at the two places where the calculation can mislead.

The page this episode explains: Expected value on one page

Voices are synthetic, made with Google Gemini text-to-speech. Full transcript: https://craigstanley.work/media/podcast/decisions--decision-theory--expected-value-on-one-page/transcript.txt

]]> https://craigstanley.work/podcast/#ep-12 craigstanley-work-podcast-decisions--decision-theory--expected-value-on-one-page Sat, 10 Oct 2026 09:00:00 GMT 224 12 full false Three goals for every work decision Anna and Tom look at the three goals for improving work decisions: saving money, making money, and looking after people, and how to measure trade-offs between them. Anna and Tom look at the three goals for improving work decisions: saving money, making money, and looking after people, and how to measure trade-offs between them.

The page this episode explains: Save money, make money, look after people: the three goals

Voices are synthetic, made with Google Gemini text-to-speech. Full transcript: https://craigstanley.work/media/podcast/decisions--start-here--save-money-make-money-look-after-people-the-three-goals/transcript.txt

]]> https://craigstanley.work/podcast/#ep-11 craigstanley-work-podcast-decisions--start-here--save-money-make-money-look-after-people-the-three-goals Sat, 10 Oct 2026 09:00:00 GMT 208 11 full false Five questions before any decision Anna and Tom discuss framing decisions before analysing them using five practical questions. They examine how agreeing options and methods up front stops bad decisions at work. Anna and Tom discuss framing decisions before analysing them using five practical questions. They examine how agreeing options and methods up front stops bad decisions at work.

The page this episode explains: Five questions before any decision

Voices are synthetic, made with Google Gemini text-to-speech. Full transcript: https://craigstanley.work/media/podcast/decisions--start-here--five-questions-before-any-decision/transcript.txt

]]> https://craigstanley.work/podcast/#ep-10 craigstanley-work-podcast-decisions--start-here--five-questions-before-any-decision Sat, 10 Oct 2026 09:00:00 GMT 191 10 full false Deciding better, cheaply and in the open Anna and Tom discuss using small models to score routine workplace choices, applying practical ideas from decision theory, and publishing methods so colleagues can check and reuse them. Anna and Tom discuss using small models to score routine workplace choices, applying practical ideas from decision theory, and publishing methods so colleagues can check and reuse them.

The page this episode explains: Decisions

Voices are synthetic, made with Google Gemini text-to-speech. Full transcript: https://craigstanley.work/media/podcast/decisions/transcript.txt

]]> https://craigstanley.work/podcast/#ep-9 craigstanley-work-podcast-decisions Sat, 10 Oct 2026 09:00:00 GMT 235 9 full false The decision inventory template Anna and Tom look at the decision inventory template, explaining how frequency, handling time, stakes, and reversibility identify where decision support belongs. Anna and Tom look at the decision inventory template, explaining how frequency, handling time, stakes, and reversibility identify where decision support belongs.

The page this episode explains: The decision inventory template

Voices are synthetic, made with Google Gemini text-to-speech. Full transcript: https://craigstanley.work/media/podcast/work--tasks-to-decisions--the-decision-inventory-template/transcript.txt

]]> https://craigstanley.work/podcast/#ep-8 craigstanley-work-podcast-work--tasks-to-decisions--the-decision-inventory-template Sat, 10 Oct 2026 09:00:00 GMT 219 8 full false Crosswalking ESCO to O*NET Anna and Tom discuss using the European Commission crosswalk to connect ESCO skills with O*NET tasks, resolving one-to-many matches, and checking profiles against local job reality. Anna and Tom discuss using the European Commission crosswalk to connect ESCO skills with O*NET tasks, resolving one-to-many matches, and checking profiles against local job reality.

The page this episode explains: Crosswalking ESCO to O*NET

Voices are synthetic, made with Google Gemini text-to-speech. Full transcript: https://craigstanley.work/media/podcast/work--esco--crosswalking-esco-to-o-net/transcript.txt

]]> https://craigstanley.work/podcast/#ep-7 craigstanley-work-podcast-work--esco--crosswalking-esco-to-o-net Sat, 10 Oct 2026 09:00:00 GMT 209 7 full false Using ESCO occupations and skills in practice Anna and Tom look at how ESCO connects occupations to skills and knowledge concepts, comparing it with O*NET and applying it to UK job descriptions. Anna and Tom look at how ESCO connects occupations to skills and knowledge concepts, comparing it with O*NET and applying it to UK job descriptions.

The page this episode explains: ESCO occupations and skills in practice

Voices are synthetic, made with Google Gemini text-to-speech. Full transcript: https://craigstanley.work/media/podcast/work--esco--esco-occupations-and-skills-in-practice/transcript.txt

]]> https://craigstanley.work/podcast/#ep-6 craigstanley-work-podcast-work--esco--esco-occupations-and-skills-in-practice Sat, 10 Oct 2026 09:00:00 GMT 221 6 full false Mapping judgement and stakes with O*NET ratings Anna and Tom examine how two O*NET ratings, Freedom to Make Decisions and Consequence of Error, build a judgement-versus-stakes grid to identify where automation fits. Anna and Tom examine how two O*NET ratings, Freedom to Make Decisions and Consequence of Error, build a judgement-versus-stakes grid to identify where automation fits.

The page this episode explains: Work context: freedom to make decisions, consequence of error

Voices are synthetic, made with Google Gemini text-to-speech. Full transcript: https://craigstanley.work/media/podcast/work--onet--work-context-freedom-to-make-decisions-consequence-of-error/transcript.txt

]]> https://craigstanley.work/podcast/#ep-5 craigstanley-work-podcast-work--onet--work-context-freedom-to-make-decisions-consequence-of-error Sat, 10 Oct 2026 09:00:00 GMT 185 5 full false Reading an O*NET occupation profile Anna and Tom discuss how O*NET occupation profiles can be used to map tasks, detailed work activities, and decision context across roles. Anna and Tom discuss how O*NET occupation profiles can be used to map tasks, detailed work activities, and decision context across roles.

The page this episode explains: Reading an O*NET occupation

Voices are synthetic, made with Google Gemini text-to-speech. Full transcript: https://craigstanley.work/media/podcast/work--onet--reading-an-o-net-occupation/transcript.txt

]]> https://craigstanley.work/podcast/#ep-4 craigstanley-work-podcast-work--onet--reading-an-o-net-occupation Sat, 10 Oct 2026 09:00:00 GMT 248 4 full false Map the work, then the decisions Anna and Tom discuss why mapping what people actually do should come before choosing AI tools. They walk through a five-step way to use public frameworks to build task and decision inventories. Anna and Tom discuss why mapping what people actually do should come before choosing AI tools. They walk through a five-step way to use public frameworks to build task and decision inventories.

The page this episode explains: Map the work, then the decisions

Voices are synthetic, made with Google Gemini text-to-speech. Full transcript: https://craigstanley.work/media/podcast/work--start-here--map-the-work-then-the-decisions/transcript.txt

]]> https://craigstanley.work/podcast/#ep-3 craigstanley-work-podcast-work--start-here--map-the-work-then-the-decisions Sat, 10 Oct 2026 09:00:00 GMT 214 3 full false Mapping work before buying tools Anna and Tom explain why mapping everyday tasks comes before any AI rollout. They look at public job frameworks and how to find real decision points. Anna and Tom explain why mapping everyday tasks comes before any AI rollout. They look at public job frameworks and how to find real decision points.

The page this episode explains: Work

Voices are synthetic, made with Google Gemini text-to-speech. Full transcript: https://craigstanley.work/media/podcast/work/transcript.txt

]]> https://craigstanley.work/podcast/#ep-2 craigstanley-work-podcast-work Sat, 10 Oct 2026 09:00:00 GMT 208 2 full false Decide in the open on the Microsoft stack Anna and Tom explain Craig Stanley's method for putting small, cheap AI models behind routine workplace decisions. They look at how to map actual tasks and keep people on the close calls. Anna and Tom explain Craig Stanley's method for putting small, cheap AI models behind routine workplace decisions. They look at how to map actual tasks and keep people on the close calls.

The page this episode explains: Craig Stanley · Work: Decide in the open.

Voices are synthetic, made with Google Gemini text-to-speech. Full transcript: https://craigstanley.work/media/podcast/home/transcript.txt

]]> https://craigstanley.work/podcast/#ep-1 craigstanley-work-podcast-home Sat, 10 Oct 2026 09:00:00 GMT 228 1 full false