David McCandless helps businesses adopt AI, covering tool selection, data governance, and hands-on team training. McCandless's posts draw on real AI rollout examples, including enterprise procurement and workforce training.
David McCandless is most useful for grounded explanations of enterprise AI adoption across procurement, data governance, and hands-on team training.
Best for
A data platform or enterprise AI enablement brand running a thought-leadership series on AI procurement, governance, and training.
Not for
A security vendor seeking technical launch content with benchmarks, architecture walkthroughs, live demos, and code-level detail.
Key takeaways
What companies call AI risk is roughly 80 percent procurement friction and 20 percent actual risk.
The chat assistant is the small layer; governed business data is the real layer.
State workforce money can help pay to train a team on AI.
Check first: Confirm audience fit and partnership terms, as no rates, deliverables, or sponsorship format are stated.
Main ideas
What companies call AI "risk" is usually procurement work · The chat assistant is the small layer; the governed data layer is the real one · AI training is a trust problem before it is a technology problem
David McCandless writes about practical business adoption of AI, from choosing tools and governing company data to training people to use them.
In the reviewed LinkedIn posts, the recurring thread is that successful adoption depends on more than picking a chatbot: companies must handle procurement, connect data responsibly, and help people learn.
The posts also show a developing series about business lessons from family, grounded in specific people and work stories.
McCandless makes abstract AI adoption questions concrete.
One post reframes objections about AI risk by separating familiar procurement work from the human responsibility to check AI output.
Another follows an account manager’s quarterly review to show why joined-up, permission-aware business data matters beyond the choice of chat assistant.
The rollout posts add practical detail often missing from tool comparisons: a Louisiana workforce program that can help fund training, and the deliberate use of cookies and conversation to build trust in a four-hour session.
The family-business series takes a more personal direction, using remembered sayings and specific working lives to show that the work extends beyond AI advice.
Follow McCandless for grounded explanations of enterprise AI adoption and examples of how people, process, data, and training fit together.
David McCandless writes about enterprise AI adoption from a consistent starting point: most of what stalls it is ordinary work, not danger.
Read more
Their recurring arguments are that procurement friction gets mislabelled as risk, the governed data layer beneath the chat assistant is the durable investment, adoption succeeds or fails on trust and trained people, and public workforce money can pay for that training. A newer thread makes the case for specific human detail against generic AI-written content.
What companies call AI "risk" is usually procurement work
When a business refuses a tool such as Claude, the honest label is mostly paperwork: another data processing agreement, another security review, another afternoon wiring up single sign-on.
Read more
Microsoft-heavy companies reach for Copilot and Google-heavy companies for Gemini because the contracts, security review and identity configuration are already done, and the vendor relationship already exists—not because those tools won a comparison on safety.
Businesses scope and staff a CRM or accounting migration without calling it a philosophical debate, yet adding AI to the same stack gets filed under scary.
They acknowledge a new attack surface, then argue that companies can point their existing controls at one more vendor.
The costly failures they cite come from people not checking output, something no procurement process prevents.
Their stated role sits upstream of the paperwork: discovery and tool comparisons that surface value slipping away.
The failure stories are a lawyer who filed a brief full of cases the AI invented without reading it, and Air Canada's chatbot promising a refund policy that did not exist—a promise a tribunal made the airline honour. For an everyday comparison, they cite a move from HubSpot to Salesforce, which gets called a project rather than a risk.
A related post condenses the argument to a single line, shifting the focus from procurement to the data layer beneath the tools.
The chat assistant is the small layer; the governed data layer is the real one
The Copilot-versus-Claude-versus-ChatGPT comparison is real but shallow, they argue.
Read more
The bigger problem is data trapped in systems that do not talk to each other; the tool for that layer is a data platform, not an assistant.
They name Databricks and make three claims: it runs inside the cloud a company has already standardised on and under contracts already signed, so nothing new has to be onboarded; it gives business users joined-up data while IT keeps the entitlements enforced by source systems, so a salesperson sees only their own accounts, down to the row; and, because it sits underneath, switching assistants later requires no rebuilding.
The stated payoff is that both camps get what they want, and that the move is safe either way because whichever assistant wins still needs someone to mind the data.
The tension is deliberate: the assistant market is contested, but they argue the choice matters less if this layer is right.
An account manager preparing a quarterly client review: instead of logging into a web app, pulling four reports, filtering them correctly and pasting them into a chat, they can ask in one sentence for the assistant to use Databricks to assemble the review over the official Microsoft connector.
This post builds on the earlier procurement argument, treating it as established and moving to the layer beneath it.
AI training is a trust problem before it is a technology problem
Across the field-notes series, the method is to make the room human before the technology arrives.
Read more
They bring homemade cookies to every session because four hours of new technology goes more easily when people have broken bread with the person teaching them.
AI is the topic; trust is the medium.
People learn better from someone they have eaten with.
The posts also present the training as a family undertaking: baking gives their wife and daughters a way to contribute to the business.
Read alongside the funding post, this makes delivery as much a priority as content: the format and the person in the room are part of what is being sold.
The sixth installment opens with a joke about oatmeal rAIsin cookies, then explains the real choice behind bringing them to a four-hour Copilot training.
The field-notes posts are numbered installments, showing the pattern across multiple posts rather than asserting it in a single piece.
State workforce money should be paying for AI upskilling
The same public programmes that have long funded safety certifications, leadership courses and skilled trades can fund AI training, they argue, and too many owners leave the money unclaimed.
Read more
The programme is funded by a small social charge folded into the unemployment-insurance taxes employers already pay every quarter.
A company that has paid in for three years has effectively pre-funded its team's training, and the programme lets it draw that money back down.
They add that other states likely run similar schemes.
The post credits training provider Elite Training's chief executive, Erin Lasyone, and HiTech for knowing to ask; those endorsements are attributed to the named parties, not presented as neutral coverage.
Louisiana's Incumbent Worker Training Program can fund Copilot training for a team; the post explains how employers pay into it before drawing on it.
Specific human detail is the thing generic AI content cannot copy
The reason for launching a series on business lessons from McCandless's family is that, as feeds fill with AI-generated material, what a model cannot invent becomes more valuable: a real person, a real number, an adage actually said on a factory floor at six in the morning.
Read more
The material is deliberately concrete, and the cast is named: a father described through LinkedIn recommendations as a CIO, two grandfathers with decades in sales and paper-mill management, and one who was a Marine and a serial entrepreneur.
The post asks named former colleagues of their father to contribute the sayings they remember, turning the audience into a source for the series.
Treat it as announced intent; ask whether installments have been published and how the series has landed.
Their father was nicknamed the "Leonidas of IT" after a job change during the financial crisis; their grandfather managed a paper mill and flew as far as Australia to fix production problems.
The series is announced as a plan, not a record of published installments.
This is the clearest statement of the argument David McCandless keeps returning to, and the fastest way for a newcomer to see how they think.
Read more
If you open only one post, make it this one.
David McCandless argues that when a business refuses an AI tool, the stated reason is “risk,” but the real mix is mostly paperwork.
Microsoft-heavy companies reach for Copilot and Google-heavy companies reach for Gemini not because of a safety bake-off, but because the data processing agreement, security review and single sign-on are already in place, and the vendor relationship already exists.
The contrast they draw is with core systems: businesses replace accounting platforms and CRMs as a scoped project, yet adding AI to the same stack becomes a philosophical debate.
On the security objection, they list controls a buyer already owns — single sign-on, data-loss prevention, scoped data access, audit logs, a signed agreement and a switch to turn off training on your data.
They locate the real danger in costly AI failures caused by someone not checking the output, such as a lawyer filing a brief built on invented cases or Air Canada's chatbot promising a refund policy that did not exist.
No procurement process prevents that.
For brands planning a campaign
A strong opening or anchor piece for a co-authored series: it reframes the buyer's objection rather than arguing against it. A data platform or AI enablement partner could take the “procurement as change management” half and supply the concrete security checklist the post leaves as a list.
Answers the question buyers face after selecting a tool: what sits underneath the assistant, and who governs it?
Read more
It also shows the house style — one worked scenario instead of a feature list.
This post starts with the Copilot-versus-Claude-versus-ChatGPT bake-off and calls it the smallest layer of the problem.
The larger layer, by their account, is data trapped in systems that do not talk to each other; the tool for that layer is a data platform rather than a chat assistant.
They name Databricks and make three concrete claims: it runs inside whichever cloud a company has already standardised on, under contracts already signed, so there is nothing new to onboard; it lets business users query joined-up business data while IT preserves the entitlements of source systems, down to the row, so a salesperson sees only their own accounts; and, because it sits underneath, switching chat assistants later does not mean rebuilding.
The illustration contrasts an account manager logging into a web app, pulling four reports and pasting them into a chat to prepare a quarterly review with asking the assistant in one sentence to assemble the review using the platform, served over an official Microsoft connector.
For brands planning a campaign
A template for co-authored explainer content that positions the partner as neutral infrastructure in a contested assistant market. This post names a specific vendor and reads as a recommendation; decide whether to retain that specificity or generalise the layer for any partnership use.
Answers a budget question most AI content ignores: who pays for upskilling, and how does a business draw that money down?
Read more
It offers a rare, concrete funding example.
Part 7 of a running field-notes series.
David McCandless describes engaging a training provider to tap Louisiana's Incumbent Worker Training Program, so state workforce-development dollars fund Copilot training for a team.
They argue that the same programme that has funded safety certifications, leadership courses and skilled trades also fits AI upskilling, and that many owners leave the money unclaimed.
The mechanics: the programme is funded by a small charge folded into unemployment-insurance taxes Louisiana employers already pay quarterly, so a company that has paid in for several years has effectively pre-funded its own training and can draw that money back through the programme.
They add that other states likely run similar schemes.
The post credits the training provider and its chief executive, as well as the employer for asking, and includes a promotional link, a discount for clients of selected managed-service partners and the series hashtag.
For brands planning a campaign
Useful for a workforce-training angle in a campaign brief, especially for regional or mid-market employers. Treat the partner names as evidence of who was involved in this rollout, not as evidence of any paid relationship.
Shows the editorial direction being built and why: original, specific, personal material set against generic AI-written content.
Read more
Read it to understand the upcoming series' tone before pitching.
David McCandless announces a LinkedIn series about business lessons learned from family, prompted by their father's retirement announcement.
The roster includes a father called the “Leonidas of IT” after a job change during the financial crisis, a grandfather with thirty-plus years in sales, a grandfather who managed a paper mill and flew as far as Australia to fix production problems, and a great-grandfather who was a Marine and serial entrepreneur.
The reason for doing it now: as feeds fill with AI-generated content, things a model cannot invent — a real person, a real number, an adage actually said on a factory floor at six in the morning — become more valuable.
The post asks named former colleagues of their father to share the sayings they remember.
For brands planning a campaign
This is the clearest evidence of a series format in progress, making a multi-part partnership a natural ask. The subject is family business rather than enterprise AI; a co-authored pitch should build a parallel series or connect the two through training and adoption.
A short read on the delivery method behind the training work: how a four-hour session is run.
Read more
Useful for judging whether a live or workshop format would fit a campaign.
Part 6 of the field-notes series, built around a joke about oatmeal rAIsin cookies.
The substance is a deliberate choice to bring homemade cookies to every training session, because four hours of new technology goes more smoothly when the room feels human first.
The stated principle is that AI is the topic but trust is the medium, and that people learn better from someone they have shared food with.
David McCandless notes that baking gives their wife and daughters a way to contribute to the family business.
The post repeats the promotional link and managed-service discount footnote used in surrounding instalments.
For brands planning a campaign
Evidence for a workshop or live-session component, not just written content. Keep the domestic detail and recurring discount footer in the relationship section rather than repeating them in every campaign plan.
In a post on why businesses refuse tools such as Claude, David McCandless argues that the objection labelled “risk” is roughly 80% procurement friction and 20% actual risk: another DPA, another security review, another afternoon wiring up SSO.
Read more
They observe that Microsoft shops reach for Copilot and Google shops for Gemini because the paperwork and identity configuration are already done, not because those tools won a safety bake-off.
The post then identifies the risk that actually bites: an AI-written brief filed with invented case citations, or Air Canada’s chatbot promising a refund policy that did not exist.
These are failures of humans not checking output, which no procurement gauntlet prevents.
They say they are not needed for tactical signature-chasing, but can run a tool bake-off during discovery and surface value slipping away in the queue.
Reframes a buyer objection with a rough quantitative split, then contrasts AI adoption with enterprise system replacements that are scoped and staffed as projects rather than debated as risks.
For brands planning a campaign
The 80/20 split and comparison with routine CRM and accounting migrations give a co-authored series a ready-made hook: AI adoption becomes an ordinary project, not a philosophical debate. A procurement, governance or enablement sponsor could build the first installment around that reframe and use the bake-off and discovery offer as the reader’s next step.
The governed data layer underneath the chat assistant
In a post about Databricks, David McCandless treats the Copilot-versus-Claude-versus-ChatGPT debate as the visible tip of the iceberg: real questions, but about the smallest layer.
Read more
The bulk of the problem is data trapped in a dozen systems that do not talk to each other, and a chat assistant is not the tool for that layer.
They argue that a governed platform inside the cloud a company has already standardized on runs under paper already signed, preserves row-level entitlements from source systems (Johnny in sales sees only Johnny’s accounts), and survives a change of chat assistant without ripping anything out.
The worked example contrasts an account manager preparing a quarterly client review by pulling four reports and pasting them into a chat with asking Copilot in one sentence to use Databricks over the official Microsoft connector.
Iceberg framing, a worked before-and-after scenario with named characters, and a no-regret argument for buying the platform whichever chat assistant wins.
For brands planning a campaign
For a data platform or governance vendor, this offers a full narrative arc: assistant choice is reversible, the governed layer underneath is the no-regret investment, and switching assistants later costs nothing. The before-and-after account-manager scenario is concrete enough to adapt directly into a co-authored piece.
Field Notes from a Real AI Rollout, Part 7, describes HiTech working with Elite Training to tap Louisiana’s Incumbent Worker Training Program to pay for Copilot training.
Read more
David McCandless explains that IWTP is funded by a small social charge folded into unemployment-insurance taxes employers already pay, so employers who have paid in for three years have effectively pre-funded their own team’s training. They note that other states likely run similar programs, credit Elite Training and its CEO Erin Lasyone, thank HiTech for asking, and offer $1,000 off for clients of select MSP partners.
First-person field report with named partners and programs, insider detail on how the funding mechanism works, and a repeated call to action.
The post carries a commercial offer—$1,000 off for clients of select MSP partners—so assess the training claim separately from the promotion.
For brands planning a campaign
A training, workforce or enablement partner could co-create content on funding pathways for AI upskilling, where the state-program mechanics are specific and hard to replicate elsewhere. Treat the MSP discount as a promotion attached to the post, not neutral third-party coverage, and identify it as such in campaign planning.
Training is a trust problem before a technology problem
Field Notes from a Real AI Rollout, Part 6, is built around bringing homemade cookies to every AI training.
Read more
The stated reason: four hours of new technology goes easier when the room feels human first, because trust is the medium and people learn better from someone they have broken bread with.
The post also says the cookies give their wife and daughters a way to contribute to the family business.
It offers a deliberate counterweight to purely technical rollout content.
Anecdote first, practical takeaway second, with a recurring series hashtag and a standing offer at the end of each installment.
For brands planning a campaign
Signals an approach a sponsor can lean into: adoption content that leads with the room and the people in it rather than the tool. A training or change-management sponsor could use this human-first angle as a softer entry point to a series that later gets technical.
Business lessons from family, and a case against AI slop
In a personal post, David McCandless announces a forthcoming LinkedIn series on business wisdom from their family: their father Michael McCandless, described via his LinkedIn recommendations as a CIO, plus two grandfathers with decades in sales and paper-mill management, one of whom was a Marine and serial entrepreneur.
Read more
They say they are publishing now because, as feeds fill with AI-generated material, what a model cannot invent—a real person, a real number, an adage actually said on a factory floor at 6am—becomes more valuable. The post asks former colleagues of their father to contribute sayings in the comments.
Personal narrative as a hook, crowd-sourced detail drawn from a comment network, and an explicit position against generic AI content.
The posts announce the series but show no installments. Ask for published examples and audience-response data before committing to a campaign.
For brands planning a campaign
Shows an appetite for human, story-led content and names a recurring cast a sponsor could be written into. David McCandless’s posts announce the series but do not show installments; ask for published examples before building a campaign around it.
David McCandless's brand relationships center on hands-on enterprise AI adoption.
Read more
Their strongest tie is HiTech: they document its rollout in a running field-notes series and advertise a discount on AI enablement for its clients.
They also recommend Databricks as the governed data layer beneath whichever AI assistant a company chooses.
Ask them to clarify the specifics of their relationships with HiTech and Databricks.
Copilot, Claude and Gemini appear as tools being compared or examples in their arguments, not as relationships.
HiTech
Disclosed partnership
McCandless writes a numbered series, “Field Notes from a Real AI Rollout,” under the #hiTechAIRolloutFieldNotes tag.
Read more
They credit HiTech with requesting AI training, note that staff at select MSP partners including HiTech get $1,000 off AI enablement, and link to the offer. Ask whether they’re employed by HiTech, contracted by it, or documenting a rollout as a client.
McCandless recommends Databricks as the governed data layer beneath whichever chat assistant a company picks, citing the official Microsoft–Databricks connector and MCP support that also serves Claude.
Read more
They offer to help teams work out where the assistant ends and the data platform begins.
Original disclosure
No commercial relationship is disclosed in the post.
In the same field-notes series, McCandless credits Elite Training and its CEO Erin Lasyone for helping tap Louisiana's Incumbent Worker Training Program to fund Copilot training for a team.
Read more
HiTech arranged the training, so the post describes Elite Training's work with HiTech, not a relationship between Elite Training and McCandless.
David McCandless writes about enterprise AI adoption as an operations and people problem rather than a technology debate.
Read more
Start with their argument that what companies call AI "risk" is mostly procurement paperwork (another data processing agreement, another security review, another identity integration), then read the companion post on the governed data platform beneath whichever chat assistant a company picks.
The Field Notes series offers the most human entry point, with numbered installments about funding Copilot training through a state workforce program and why the training room matters as much as the tooling.
Their newest thread, a planned series of business lessons from their family, argues for specific human detail over generic AI-written content.
Read the numbered installments in order; they build on each other, and the cookie installment is a deliberate counterweight to purely technical rollout writing.
Three comparators overlap on different parts of the subject and take distinctly different approaches.
Works with the same Microsoft data stack from the delivery side.
Read more
Where David McCandless argues about which layer deserves investment, this creator writes hands-on about building the governed Fabric and Power BI platform, including a six-week delivery story covering customer data reconciliation, incremental refresh, and UK and EU data protection controls designed in at discovery. Their training offer covering Fabric, DAX, GDPR, and the EU AI Act overlaps directly with enablement and governance.
More vendor- and delivery-specific.
The posts describe consultancy capabilities and services, with detailed build outcomes and certification prep.
David McCandless frames adoption as an organizational and procurement argument and rarely covers implementation mechanics.
Covers the same governance gap from inside the risk function.
Read more
The argument that a two-year-old SOC 2 never assessed a model's training data, bias testing, or human oversight gives a sharp example of David McCandless's point that costly AI failures trace back to people and process rather than vendor onboarding. Both offer practical, checklist-shaped guidance on policy and procedure, repeatable risk assessment, vendor AI review, and incident response playbooks.
Anchored in regulations and frameworks, citing NIST AI RMF and examiner expectations, and aimed squarely at risk and compliance teams. David McCandless takes a more commercial approach, leading with procurement friction, data platform arguments, and a training offer rather than benchmarking against a named framework.
Overlaps on Microsoft Copilot adoption and governance, including a detailed account of a 400-plus-seat deployment where licenses were live but value was inconsistent, and a 90-day AI Center of Excellence model built on data security review, Purview sensitivity labels, usage policy, role-based training, and measured outcomes.
Read more
Both treat adoption as a people and governance problem rather than a licensing problem.
Program- and metrics-driven, with a phased framework, automation counts, and time-saved figures, and a career-and-community voice rooted in the Microsoft partner ecosystem. David McCandless is more argumentative and conversational, using reframing and everyday analogies and covering funding and training-room practice rather than a structured rollout methodology.
The LinkedIn profile that published these posts had 5,496 followers when the count was recorded.
Read more
The posts don’t describe their geography, industry, job function, seniority, or company size.
For an enterprise AI enablement or data platform partnership, ask whether the audience includes the procurement, IT, and data decision-makers you need.
The writing is aimed at readers who already negotiate DPAs, set up SSO and SCIM, and work in a Microsoft or Google tenant.
The data-platform post addresses the person deciding where the governed data layer sits, along with IT and InfoSec approvers and business users.
Two AI rollout field notes speak to employers, owners, and managed-service partners, including a mid-market example about state workforce-training dollars funding Copilot training.
The family-business post reads as a personal, feed-native piece rather than buyer content.
Together, the posts suggest a practitioner and decision-influencer readership on LinkedIn, based on the knowledge they assume rather than audience measurement.
Before committing, ask for recent results on comparable topics, depth, and formats—for example, reach and engagement on the enterprise AI procurement and data governance posts versus the training field notes.
Also ask whether an audience survey or platform demographics are available for geography, job function, and seniority.
Platform demographic data generally cover geography, age, and gender only; an audience survey is the better way to verify employer, occupation, and job seniority.
LinkedIn followers
5,496
Follower count recorded on the LinkedIn account that published the reviewed posts · 2026-10-11
David McCandless’s posts show commercial signals, but no stated partnership or sponsorship format.
Read more
They offer to help companies run discovery and tool bake-offs for AI adoption, promote AI enablement training and a $1,000 discount for clients of select MSP partners including HiTech, and write approvingly about Databricks running inside existing cloud environments.
The posts state no payment, sponsorship terms, rates, deliverables, exclusivity, or partnership inquiry route.
A brand considering a thought-leadership partnership should ask directly whether the vendor mentions are commercial and what terms apply.
Strong fit for the argument-led role in that series. The creator supplies the reframe, narrative arc and editorial voice; the partner's staff should supply the technical detail the posts stop short of.
Read more
Brief
A data platform or enterprise AI enablement company wants a named thought-leadership partnership: a co-authored series on enterprise AI procurement, governance and workforce training for the people who sign off on new tools.
Reasons
The procurement argument is already shaped like a campaign opener. In the Claude and Copilot post, David McCandless reframes the objection: what a company calls "risk" is roughly 80% procurement friction and 20% actual risk. Microsoft-heavy or Google-heavy shops reach for the incumbent tool because the data processing agreement, security review and single sign-on are already done. The post contrasts AI with accounting platforms and CRMs, which businesses scope, staff and replace as projects. It gives a sponsor a first installment to build on rather than argue with.
The data-platform post supplies the second act. It treats the Copilot-versus-Claude-versus-ChatGPT bake-off as the smallest layer, puts the bulk of the problem in data trapped across systems, and argues that the governed layer inside a company's existing cloud is a no-regret purchase because switching assistants later requires no rebuild. It includes one worked scenario: an account manager pulling four reports and pasting them into a chat versus asking the assistant to assemble the review from governed data. The partner can extend that example with its own connector and entitlement detail.
The workforce-training installment supplies a budget angle most AI content skips. It names Louisiana's Incumbent Worker Training Program, explains that employers effectively pre-fund their own training through a small charge in unemployment-insurance taxes, and notes that other states likely run similar schemes. That is specific, checkable and hard for a competitor to copy, making it a natural third installment or standalone co-authored piece.
The series format is already in use. The field notes run as numbered installments with a recurring hashtag, a recurring offer and a consistent voice, so a multi-part partnership is not new behaviour for this creator. The training-method installment shows the delivery style: bringing homemade cookies to every session, on the argument that four hours of new technology goes easier when the room feels human first and people learn better from someone they have shared food with.
The writing assumes a buyer who already negotiates agreements, wires up identity systems and works inside a Microsoft or Google tenant. For a governance or enablement sponsor, that assumed knowledge is the point: the content does not spend its first half explaining what a security review is.
Watch outs
Vendor names appear in the posts without stated terms. Databricks is named approvingly, and the training installments carry a discount for clients of selected managed-service partners. Neither is described as paid. Before a co-authored series runs, agree in writing how vendor mentions appear and what is disclosed to readers.
Only a follower count is available, with no geography, job function, seniority or company size. The campaign's promise to reach procurement, IT or data decision-makers would rest on the topics the posts are written for, not audience measurement. Ask for per-post performance on the procurement, governance and training pieces before committing budget.
The posts leave the tactical half as a list: single sign-on, data-loss prevention, scoped data access, audit logs, a signed agreement and a switch to turn off training on your data. The partner's own technical staff should supply the completed checklist and receive credit.
When to choose a specialist
Poor fit.
Read more
Brief
A security vendor launching an AI-augmented threat detection product wants a technical launch campaign: benchmark results, architecture walkthroughs, live demos and code-level detail for security operations and detection engineering teams.
Reasons
David McCandless's posts offer no technical proof: no benchmark numbers, architecture diagrams, code, demo or performance data. The most concrete technical statement is a one-sentence description of an official connector and the claim that source-system entitlements hold down to the row for a named salesperson. That is not material a detection engineering audience can act on.
The central argument cuts against a tool-led risk pitch. The procurement post concludes that failures that actually cost money—a lawyer filing a brief built on invented case citations, a chatbot promising a refund policy that did not exist—trace back to a human not checking output, and that no procurement gauntlet prevents that. The position favours governance, training and process content over a product asked to catch the risk on the buyer's behalf.
David McCandless's posts show no security credentials, certifications, engineering background or security-operations readership. They assume familiarity with procurement and identity configuration, a different kind of expertise from threat detection.
The only audience figure available is a raw follower count, with no industry, function or seniority breakdown. Ask for audience data before promising reach to security operations teams.
A second poor fit: any brief whose value depends on audience size, consumer reach or follower-count-driven placement. This creator's evidence supports argument and teaching, not reach delivery.
A worked campaign example
Brief
A data governance or AI enablement platform wants a three-part co-authored series on where enterprise AI actually gets stuck, for the buyers who approve new tools.
Instalments
Content
Open with the 80/20 split and the comparison to CRM and accounting migrations that get scoped and staffed rather than debated. Walk through the controls a buyer already owns. Close on the harder point: the failures that cost real money come from a person not checking output. The sponsor adds the pre-sales security checklist as a downloadable companion piece.
Move from the assistant bake-off to the data underneath it. Use the account manager scenario in full, then extend it with the sponsor's own connector and row-level entitlement detail, including a plain description of what an administrator configures and how long it takes. Close on the no-regret argument: assistant choice can change; the governed layer does not have to be rebuilt.
Combine the funding mechanics with the delivery method. Explain how a state workforce program is funded through employer tax contributions and how an employer draws it back down; note that comparable schemes exist elsewhere; and show what a four-hour session looks like when the room is treated as human first. The sponsor supplies a short worksheet for checking eligibility in a reader's own state.
Before commissioning, ask for per-post performance on the procurement and data-platform posts as well as the training installments. The audience gap is a measurement problem, not an interest problem.
Confirm the terms of the existing Databricks mention and the managed-service discount, and agree in writing how sponsor mentions are disclosed.
Set the acceptance test for the series: after one installment, a reader unfamiliar with the creator should be able to explain where enterprise AI gets stuck and what to do next.
Premise
Run the series as three consecutive installments, each ending with the same next step. The creator writes the argument and scenarios; the sponsor's staff supply verified technical and financial detail to earn reader trust. Edit each installment so readers who see only one still get a complete point.
Reader handoff
End each installment with the same step: book a discovery session to map where the reader's data actually sits and run a tool bake-off for their own organisation. The offer already appears in the creator's material, so the series needs no new call to action.
Title
The 90% Below the Waterline
What the sponsor supplies
The completed security and configuration checklist that the first post leaves as a list.
Technical review of the connector paragraph in the second installment, with a named engineer credited.
A verified, dated eligibility summary for the workforce-training funding in the third installment, plus the reader worksheet.
Editorial review of every claim about the sponsor's own product, kept separate from the creator's argument.
Choose Clive Lemmon when the campaign needs a delivered Microsoft Fabric implementation and a training catalogue behind the argument.
Read more
The Fabric readiness post supplies a dated client story: a governed platform delivered in six weeks for a security services provider running hundreds of sites, with UK and EU data protection and ISO 27001 controls designed in at discovery, and source-system data quality issues found and fixed during the build.
The training post lists Microsoft Fabric, Power BI, data engineering, AI, GDPR and cybersecurity programmes, including certification preparation.
For a sponsor that must show implementation evidence and a path from content to paid enablement, this is a close fit for the same buyer task as an argument-led series.
Clive Lemmon’s material reads as a service provider’s offer: the Fabric case study and training menu are concrete, but there is less independent editorial reframing of procurement friction or the politics of adopting a new AI tool.
The question of where enterprise AI gets stuck and why buyers hesitate is answered indirectly through delivery capability, rather than through the kind of first-person argument David McCandless writes.
A sponsor seeking thought leadership that challenges the buyer’s framing may get a more conventional vendor-adjacent tone.
Choose Lucas Hathaway when the campaign’s centre of gravity is AI risk governance and vendor assessment.
Read more
The NIST AI RMF post gives a practical starting sequence: benchmark against NIST AI RMF or CRI’s Financial Services AI Risk Management Framework, write policies for how teams actually use AI, run a repeatable AI risk assessment inside the existing risk process, add AI items to vendor security reviews, and update incident-response playbooks for models acting on their own or giving confidently wrong answers.
The vendor-assessment post goes further: send a discovery email to critical and high vendors asking what AI is in the product, demand notice before new AI features are enabled on the tenant, and keep a dated record of what was asked, answered and decided.
That is specific, checkable governance detail for buyers who need to show examiners they have asked the right questions.
Lucas Hathaway’s material is narrowly framed around risk, compliance and assessment, with a free NIST AI RMF assessment as the offer.
It does not cover the data-platform-underneath argument, the state workforce-training funding angle or the delivery-room teaching method David McCandless uses.
A sponsor whose brief is broader enterprise AI enablement would get deep governance credibility but less of the narrative that moves a general business audience from hesitation to action.
The content may read more like a consultant’s checklist than a co-authored editorial series.
Choose Jeff Willinger when the campaign needs a Microsoft 365 Copilot adoption and Center of Excellence case study with measured outcomes and a partner-ecosystem angle.
Read more
The AI Center of Excellence post describes a $1.2B health and life sciences organisation with 400+ Copilot licences, no formal governance model and inconsistent usage, then lays out a 90-day phased model: vision and strategy, enablement infrastructure, policies and security, user training, business/IT alignment, and usage analytics.
It reports proposal creation time down 28%, executive prep time down 35%, 14 internal workflows automated, and Copilot expansion approved within six months.
The job announcement adds two decades in the Microsoft ecosystem, Microsoft MVP years, and a focus on Modern Work, Copilot, Power Platform, Dynamics and Microsoft co-sell.
For a sponsor that needs a Microsoft-anchored transformation story with numbers, this directly addresses the same corporate AI enablement brief.
The case study focuses on a single organisation and is heavily Microsoft-specific.
It does not address the procurement-friction argument or the governed data layer underneath the assistant, so a data platform sponsor may have less to build a series around.
The measured outcomes are presented as one client’s results, not as a research base, and the offer is closer to an ecosystem partner’s transformation playbook than to independent thought leadership.
If the need is a vendor-neutral argument about where AI adoption actually sticks, this alternative narrows the frame to Microsoft 365 Copilot and its governance model.
Five LinkedIn posts by David McCandless cover an argument that what companies call AI "risk" is mostly procurement paperwork, a post on the governed data layer beneath whichever chat assistant a company picks, three installments and announcements from a running AI rollout field-notes series, and the launch of a planned family business lessons series.
The posts here have no publication dates, so the ideas are not placed on a timeline; the order within the field-notes series comes from its part numbers, and one post states it follows an earlier one.
Comparison with three other creators draws on six of their posts dated between March and October 2026.
The posts here include no books, videos, transcripts, or other longer works, so whether David McCandless has signature work beyond these posts remains unknown.
The statements on this page come only from the posts listed below.
