Kathy D’Agostino shares practical guidance on using AI in small businesses, especially professional-services firms. D’Agostino’s posts cover setting up repeatable AI processes, improving tasks such as proposals and pricing, and managing privacy and accuracy risks.
Weak AI output starts with unclear thinking, not a weak prompt.
Adopt AI as a staged system, not a one-time purchase.
AI risk is a leadership mandate, not an IT problem.
Check first: Confirm partnership terms, recent post results, audience job titles, and claim substantiation.
Main ideas
Weak AI output starts with unclear thinking, not a weak prompt · Adopt AI as a staged system, not a one-time purchase · AI risk is a leadership mandate, not an IT problem
Kathy D’Agostino shares practical guidance on adopting AI in small businesses, with particular attention to professional-services firms.
Their posts show how to turn AI use into repeatable work: start with a specific bottleneck, set clear constraints and safeguards, and check whether the process improved.
Examples range from pricing an unfamiliar project to preparing proposals, while other posts address data privacy and AI adoption in manufacturing.
D’Agostino connects hands-on AI instruction with the operating decisions around it.
In one post, they lay out a 30-day adoption plan that moves from finding a time-consuming task to setting privacy rules and measuring results.
Elsewhere, they show how to structure a project estimate from task scope, labor hours, and hourly rates, and explain how an AI policy can distinguish acceptable general problem-solving from sensitive customer or proprietary data.
This makes the work useful to readers looking beyond tool recommendations: it offers concrete ways to test workflows while keeping business risks in view.
The posts also include lighter, exploratory work, such as comparing ChatGPT and Gemini for data visualization, so the material is not limited to governance or process advice.
Kathy D'Agostino writes about how small firms should adopt AI.
Read more
Their argument is that the hard part is not the tool: unclear thinking, missing guardrails, and poorly matched tools cause most failures.
Adoption works best as a staged operational rollout, not a subscription purchase.
The posts return to a few priorities: think before opening a tool, pick one bottleneck and measure it, set data rules, match each model to the job, and use AI to protect margins and senior hours.
The advice is aimed mainly at professional-services firms under ten people, with one post extending to small and mid-sized manufacturers.
Weak AI output starts with unclear thinking, not a weak prompt
Their recurring argument is that leaders open a tool without a clear idea, type a one-line request, then blame the model for generic results.
Read more
The fix is to do the intellectual work first.
Two frameworks recur.
One maps four boundaries before prompting: Goal (the exact outcome), Obstacle (the friction point), Constraint (the non-negotiables), and Context (the operating reality).
The other gives the model a role, such as “You are my sales strategist,” rather than simply asking it to help.
The reasoning is consistent: give a model real-world limits and it can produce targeted operational work instead of fluff.
One post contrasts “write me a proposal” with a prompt that identifies the client as a six-agent realty firm seeking marketing support and requests a clean, one-page proposal in a professional tone with a 30-day expiration.
It reports that drafting time fell from two to three hours to about 20 minutes.
Another post shows how to create consistent observation reports: the goal is a format any team member can replicate; the obstacle is that each document takes too long and varies by writer; the constraint is that non-technical staff must be able to use it with minimal training; and the context is a fast-growing consulting firm producing client-facing documents.
Adopt AI as a staged system, not a one-time purchase
They frame adoption as an operational rollout: audit where time is lost, choose one high-friction problem instead of ten, establish a repeatable habit, set guardrails, measure the change, then apply the same pattern to the next bottleneck.
Read more
They report auditing more than 40 small professional-services firms and finding that almost none were actually using AI; they were only paying for it.
The test is direct: cancel every subscription tomorrow and ask whether the work process would genuinely change.
Three maturity stages follow: Solo and Silent, where people quietly use free tools with no policy or shared approach; Equipped, with paid secure accounts, a basic data policy, and shared prompts for recurring work; and Embedded, where AI is trained on the firm's own methods and wired into process automation.
The 30-day plan for firms under ten people assigns days 1-3 to auditing where the team loses time, including email, meeting follow-ups, and proposal drafting; days 4-7 to choosing one problem; days 8-10 to matching tools to tasks; days 11-15 to building prompting confidence; days 16-20 to setting privacy and quality guardrails; days 21-25 to measuring results, with email time cut in half and research sprints dropping from 90 minutes to 10; and days 26-30 to applying the pattern to the next bottleneck. The closing note calls day 31 a starting point, not a finish line.
AI risk is a leadership mandate, not an IT problem
Two posts frame ungoverned AI use as a leadership failure.
Read more
The term "shadow AI" describes employees logging into free public tools on personal devices and pasting sensitive company data in, not out of malice but to move faster.
D'Agostino separates the exposures: privacy and data leakage when free models train on inputs, hallucination and accuracy risk when unreviewed output reaches client deliverables, and third-party partner risk when vendors feed unverified data back into the process.
The countermeasures are written policy, verified privacy settings, and an appointed "AI champion" who does not need to be a technologist.
In a post aimed at small and mid-sized manufacturers, they acknowledge that fears of security leaks are valid but argue that paralysis is not the answer.
Three rules follow: confirm which model is being used and that your data is not used to train it; specify what data can be entered (operational friction points, logistics, general problem-solving) and what cannot (proprietary code, customer data, client schematics); and appoint someone curious and organized to track privacy settings and model changes.
They also recommend starting with one data-heavy bottleneck, such as a recurring supply-chain delay, a slow quoting process, or an engineering review that never gets finished.
They treat model choice as an operational decision.
Read more
In the adoption roadmap, ChatGPT is assigned to strategy and writing, Claude to deep document analysis, Perplexity to fast sourced research, and Gemini to teams already working in Google Workspace.
The stated rule: choose the tool that fits how a team already works, not the one getting the most attention.
A separate post compares ChatGPT and Gemini for data visualization, favoring Gemini for complex narrative compositions where precise object placement and text integration matter, and ChatGPT for speed, accessibility, and editing.
A visualization post describes generating an image of a disrupted supply chain as an intricate digital grid realigned by a centralized AI core, with sharp native text labels rendered directly on the graphic rather than corrected afterward.
They also describe building process visuals, such as a step-by-step flow of data from an input terminal to a processed output with labeled modules.
The post concludes that the choice is between a static summary and an immersive, data-driven experience.
Use AI to protect margins and senior hours, not just draft text
A recurring subject is document and pricing work that consumes senior time: proposals, pitch decks, standard operating procedures, client onboarding, and bids on unfamiliar projects.
Read more
They argue that the real waste is mechanical setup—formatting slides, choosing fonts, or guessing at a price—and that AI should take on that work so expensive people can focus on message and strategy. Their claim is that structured, bottom-up outputs are more defensible than a single generated number or a blank page.
For pricing, they walk through a small agency bidding on a multi-platform social media project: define the scope explicitly (12 unique posts, caption writing, graphic layout, scheduling, and monthly analytics), have the model estimate labor hours for each task, apply hourly rate assumptions, and produce a recommended monthly retainer price. For presentation decks, they describe pasting raw internal notes, a ten-page draft proposal, or a text SOP into an AI canvas engine such as Gamma, then using one global command like “replace formal language with plain English” instead of editing individual text boxes.
This is the clearest statement of the method seen across the posts: pick one bottleneck, create a repeatable process, and measure the change before moving on.
Read more
Read this one first.
A day-by-day rollout for professional-services firms with fewer than ten people.
Days 1–3 audit where the team loses time; the post names the usual answers: email, meeting follow-ups, and proposal drafting.
Days 4–7 focus on one high-friction task instead of ten, arguing that overly broad scope kills AI adoption in small firms.
Days 8–10 match tools to jobs: ChatGPT for strategy and writing, Claude for deep document analysis, Perplexity for fast sourced research, and Gemini for teams already in Google Workspace.
Days 11–15 build a repeatable prompting habit; days 16–20 set privacy, confidentiality, and quality-review guardrails; days 21–25 measure results using concrete markers (email time cut in half, research sprints down from 90 minutes to 10, first drafts 70–80% complete).
Days 26–30 apply the same pattern to the next bottleneck.
The closing note calls day 31 a starting point, not a finish line.
The post offers a free quick-start guide with a four-part prompt framework and ten ready-made prompts in exchange for a direct message.
For brands planning a campaign
The 30-day structure offers reusable campaign scaffolding: a workshop syllabus, serialized newsletter, or self-assessment checklist can map each day or phase to a content beat. A partner whose product addresses one step — meeting follow-ups, proposal drafting, or document review — can sponsor the corresponding day rather than the whole plan.
It answers the question a new reader is most likely to ask — what does this person teach people to do differently? — with a short, usable framework rather than a general opinion.
Read more
The post diagnoses a common failure: leaders open an AI tool without a clear aim, type a one-line instruction such as "write a proposal for a consulting client," then blame the tool when the output is generic.
Its blunt claim: "You don't have a prompt problem.
You have a thinking problem."
The fix is to write down a four-part frame before prompting: Goal (the exact uncompromising outcome), Obstacle (the precise friction point), Constraint (the non-negotiables), and Context (the environmental reality).
A worked example shows how to produce consistent observation reports: the goal is reports any team member can replicate; the obstacle is that each document takes too long and varies by writer; the constraint is that a non-technical staff member must be able to use it with minimal training; and the context is a fast-growing consulting firm with multiple project verticals and client-facing documents.
The post offers a four-part structural worksheet in exchange for a direct message.
For brands planning a campaign
The four-part frame can anchor a downloadable worksheet, webinar exercise, or sponsored prompt library. It can be taught in one sitting, making it a better fit for short-form placements than a longer training program.
Here, the advice moves beyond drafting and writing to the question that determines whether a services business should take on the work: margin.
Read more
The method applies to a commercial decision, not a content task.
The post tackles a familiar problem: bidding on a project type the firm has never scoped.
It describes an "AI Intensive" session in which a small agency used an LLM to build a bottom-up financial estimate for a multi-platform social media project.
The four steps: define the scope explicitly (twelve unique posts, caption writing, graphic layout, scheduling, and monthly analytics reporting); ask the model to estimate operational hours for each task; apply hourly rate assumptions to calculate baseline internal cost; and produce a recommended monthly retainer price justified by the labor involved.
The post argues this protects margins and prevents firms from underpricing to win a contract, then absorbing the labor cost.
It offers to share the prompt blueprint in exchange for a comment.
For brands planning a campaign
This is the strongest fit for campaigns targeting agency owners and independent consultants: pricing and quoting are live pain points, not abstract curiosities.
A partner selling accounting, invoicing, or proposal software can tie directly into the pricing-audit offer.
A general AI-tools brand has less to connect to here than in posts about work processes.
The post offers a diagnostic for making the rest of the advice actionable: before deciding what to do about AI adoption, identify the firm's current stage.
Read more
It also shows how the writer supports claims about scale and evidence.
The post reports that an audit of more than 40 small professional-services firms found almost none of them were actually using AI, even where they were paying for it.
The test offered is direct: cancel every AI subscription tomorrow and ask whether the firm's work processes would genuinely change.
Three stages follow.
Stage 1, "Solo and Silent," is people quietly using free tools for small tasks with no policy, no shared approach, and no visibility into what is being typed in, including client data; the post estimates this is about 80% of firms.
Stage 2, "Equipped," is paid secure accounts, a basic data security policy, and shared prompts for recurring work such as proposals, reports, and client communications.
Stage 3, "Embedded," is AI trained on the firm's own methods, wired into real work automation, and funded by leadership as infrastructure rather than as an experiment.
The closing line is that counting software seats tells you nothing and watching how work gets done tells you everything; readers are invited to post their stage number and receive a reply.
For brands planning a campaign
The maturity-stage framing could power a lead magnet or interactive assessment: a three-question quiz that returns a stage and next step. It also lets a sponsor focus on a specific transition, such as moving from free tools to paid secure accounts, rather than AI adoption in general.
This covers the risk side of the argument made in the posts about work processes.
Read more
It also shows how the writer frames AI adoption as a leadership responsibility rather than a technology purchase.
The post asks owners whether employees are already using AI behind their backs.
It describes "Shadow AI" as staff logging into free public tools on personal devices and pasting in sensitive company data — not out of malice, but to move faster.
It separates three risks.
First, privacy and data leaks: free public models often train on their inputs, so pasting a proprietary client schematic, legal contract, or employee review can permanently compromise it.
Second, hallucinations and inaccuracy: models predict plausible answers rather than factual truth, so unreviewed output can introduce fabricated facts into client work.
Third, third-party partners: vendors and partners feeding work back into operations may handle data without oversight.
The post frames this as a leadership mandate, not a technical problem, and argues that without written guardrails a firm has a data liability rather than a technology strategy.
It offers a three-question self-assessment checklist in exchange for a direct message.
For brands planning a campaign
Use this item for campaigns on governance, compliance, or data protection.
It targets buyers with budget authority — owners or operations leads — rather than individual practitioners.
Because the post makes claims about how free public models treat inputs, campaigns should secure strong substantiation.
The method extends beyond office-based professional services to small and mid-sized manufacturers.
Read more
The post restates the guardrails in operational rather than knowledge-work terms.
The post opens with the objection "We're too small for AI," acknowledges that fears about security leaks and data exposure are valid, and argues that they need not cause paralysis.
It lays out three rules.
First, check the model and privacy settings before entering proprietary information: confirm that data is not used for training, then re-check periodically because settings, plans, and defaults change.
Second, set a written AI policy specifying what can be entered — operational friction points, logistics, and general problem-solving — and what cannot, including proprietary code, customer data, and client schematics.
Third, appoint an AI Champion: not a technologist, but someone curious and organized enough to track privacy settings and model changes as the tools evolve.
The post also advises against automating the whole company at once, recommending one data-heavy bottleneck, such as a recurring supply chain delay, a slow quoting process, or an engineering review that never finishes on time.
It cites that 80% of manufacturers plan to increase AI investment in the next two years, points to the pairing's location context, and invites direct messages.
For brands planning a campaign
Use this item for campaigns targeting industrial or manufacturing audiences rather than services businesses.
The "AI Champion" role gives partners a concrete focus for a recognition program, certification, or community.
The warning that privacy settings and defaults change over time also supports recurring content.
This is the clearest example of how the writer approaches visual work: it names specific tool strengths rather than offering general praise.
Read more
The post also shows an exploratory side of the practice, alongside the more settled workflows described elsewhere.
The post asks whether the classic text-and-icon infographic is finished and describes extensive experimentation with ChatGPT and Gemini for data visualization.
It names three shifts: from abstract icons to photorealistic business scenarios; from manual text correction to sharp, accurate labels generated directly in the image; and from static snapshots to assets that narrate a process, such as data moving from an input terminal to a processed output through labeled modules.
The post distinguishes the tools: Gemini handles high-definition, complex narrative compositions where object placement and text integration matter, while ChatGPT offers faster, more accessible editing.
Rather than declaring a winner, it focuses on matching each tool to the design problem, then asks whether the goal is static summaries or immersive, data-driven experiences.
The accompanying graphic was generated with Gemini; ask to see it before assessing the visual work.
For brands planning a campaign
Use this item when a campaign needs an example of visual-content capability; it suits design, presentation, or creative-tool partners. Obtain the actual image before reproducing or building on the described graphic: the post includes only the post text.
This is a small, closed-loop example of the argument running through the other posts: removing mechanical setup frees time for the work that wins clients.
Read more
It also shows how a specific tool fits into the method.
The post starts with the thought, "I just need to find 4 hours this week to build this presentation," and argues that this is why proposals, pitch decks, and onboarding guides sit untouched for weeks.
The work people dread, it says, is alignment, font sizes, slide layouts, and template colors—not the thinking.
The recommendation: stop starting from a blank slide and use an AI canvas engine such as Gamma to build a structured draft.
The post describes three steps: paste messy notes, a long draft proposal, or a text SOP into the engine and let it map the presentation structure; use a global command, such as "replace formal language with plain English" or "make every heading a strategic question," to revise the deck at once; then use the two-minute first draft to free up attention for message precision and strategy.
It closes with an offer of a one-page operational guide in exchange for a comment.
For brands planning a campaign
Presentation and document-building tools are an obvious category fit.
The process—paste raw material, apply a global edit, refine the message—is concrete enough to demonstrate.
It pairs well with the pricing post, since both address the proposal and pitch stage of winning client work.
This is the most concrete demonstration of the prompting advice, with two worked examples and a small table of outcomes newcomers can compare with their own week.
Read more
The post opens with a familiar pattern—retyping the same prompt seventeen times a month—and calls it a trap: using AI as a to-do list instead of a system.
The remedy is to give the model a role, contrasting "write me a proposal" with "You are my high-ratio closing proposal writer," and "help me prep for this meeting" with "You are my sales strategist."
Two full examples follow.
The proposal prompt names a six-agent realty firm seeking marketing support and specifies a clean one-page proposal, a professional tone, no jargon, and a 30-day expiration.
The sales-call prompt identifies the person being met—an agency owner who wants to scale client acquisition but is stuck doing everything manually—and requests a tailored agenda and three questions to ask.
The post estimates that proposal drafting falls from 2–3 hours to 20 minutes, and sales-call prep from 45 minutes to 10.
It closes with an offer of ready-to-run templates in exchange for a comment.
For brands planning a campaign
The before-and-after table is the most quotable asset for advertising or landing-page use, and the two templates could serve as a low-friction lead magnet. Productivity or CRM partners could tie in with the sales-call example, which ends in a meeting rather than a document.
Matching tools to tasks across a small firm's week3
Producing the documents: proposals, reports, and decks3
A three-stage yardstick for real AI adoption2
Shadow AI, data exposure, and written guardrails2
Using AI for pricing and scenario modeling, not just writing1
AI adoption as a system, not a prompt trick
D'Agostino's most repeated argument: weak AI output usually starts with unclear thinking, not the prompt.
Read more
One post replaces a one-line instruction with a Goal-Obstacle-Constraint-Context rule mapped out on paper before opening any tool: define the exact outcome, friction point, non-negotiables, and operating context.
Another replaces generic requests with an assigned role such as "You are my high-ratio closing proposal writer," alongside a before-and-after table showing proposal drafts dropping from two to three hours to about 20 minutes and sales-call prep from 45 to 10.
A 30-day roadmap for firms under 10 people applies the same idea to adoption: audit where time is lost, pick one problem rather than ten, train for confidence, set guardrails, measure, then replicate.
Numbered frameworks, worksheets, day-by-day roadmaps, and before-and-after time comparisons; each post ends with a downloadable guide or template offered by direct message or comment.
The same argument is also aimed at small and mid-sized manufacturers rather than professional-services firms, with supply-chain delays, quoting, and engineering reviews as example bottlenecks.
For brands planning a campaign
A brand selling AI tools, training, or operations consulting can sponsor a structured program — a worksheet, role templates, or a 30-day rollout — rather than a prompt library. The stated time savings give sponsors concrete material for ad copy or a workshop offer.
Using AI for pricing and scenario modeling, not just writing
In one post, D'Agostino walks through a small agency bidding on a type of project it had never priced.
Read more
They describe building a context-rich prompt that makes the model work bottom-up: define the scope explicitly (12 unique posts, custom captions, graphic layout, scheduling, monthly analytics), estimate labor hours for each task, apply hourly rate assumptions, then produce a recommended monthly retainer price. The goal is margin protection — avoiding guesswork, underpricing, and contracts where labor costs get eaten.
For brands planning a campaign
This opens a financial angle that most AI content misses: pricing software, accounting services, or advisory firms can sponsor a campaign around scenario modeling and quoting rather than content generation.
D'Agostino reports auditing more than 40 small professional-services firms and finding that almost none were actually using AI — they were paying for it.
Read more
Their test: if every AI subscription were cancelled tomorrow, would the workflow change?
They lay out three stages.
Solo and Silent: people quietly use free tools for small tasks, with no policy, shared approach, or visibility into what data is being entered.
Equipped: paid secure accounts, a basic data security policy, and shared prompts for recurring work such as proposals and reports.
Embedded: AI trained on the firm's own methods and wired into workflow automation, funded like infrastructure rather than an experiment.
Counting software seats tells you nothing; watching how work gets done tells you everything.
For brands planning a campaign
The maturity ladder offers a ready-made campaign hook: a self-assessment quiz, tiered content, or a diagnostic offer. It also shows what the audience D'Agostino addresses is asking for: most firms they describe are stuck at stage one, making entry-level education and policy templates likely next purchases.
Two posts treat risk as a leadership problem rather than a technical one.
Read more
The Shadow AI post describes employees logging into free public tools on personal devices and pasting in sensitive company data, not out of malice but to move faster.
It names three exposures: privacy and data leakage when free models train on inputs; hallucination and accuracy risk when outputs reach client deliverables without human verification; and third-party partner risk when vendors feed unverified or biased data back into the workflow.
A companion post aimed at small and mid-sized manufacturers offers countermeasures: confirm which model and privacy settings are in use, write an explicit AI policy covering what data may and may not be entered, appoint an organized non-technologist as an AI champion, and start with one data-heavy bottleneck.
For brands planning a campaign
This is the strongest governance content in these posts, suited to sponsors selling secure or enterprise AI seats, data-protection services, and audit or policy templates. D'Agostino addresses the owner or operations lead who signs off on the purchase, not the individual user.
Matching tools to tasks across a small firm's week
D'Agostino treats tool choice as an operational decision.
Read more
In the adoption roadmap, they assign ChatGPT to strategy and writing, Claude to deep document analysis, Perplexity to fast sourced research, and Gemini to teams already living in Google Workspace.
Their rule: the best tool is the one that fits how a team already works, not the one with the most hype.
A separate post compares ChatGPT and Gemini specifically for data visualization, arguing that AI can now render sharp native text labels directly onto graphics and build narrative process visuals such as a supply-chain disruption or step-by-step data flow, with Gemini favored for complex compositions and ChatGPT for speed and editing.
The visualization post is an outlier: it speaks to design and future-of-work readers rather than the small professional-services and manufacturing operations audience D'Agostino addresses in other posts.
For brands planning a campaign
D'Agostino is credible on head-to-head comparisons and multi-tool workflows, making them a fit for platforms that integrate several models or vendors willing to appear in a balanced comparison. The "fits how your team already works" framing keeps the content practical rather than promotional.
Producing the documents: proposals, reports, and decks
A recurring subject is document work that consumes senior time — proposals, sales-call preparation, standard operating procedures, client onboarding, observation reports, and presentation decks.
Read more
In the deck post, D'Agostino describes importing raw internal notes, a ten-page draft proposal, or a text SOP into an AI-powered canvas tool such as Gamma, letting it map the material into presentation cards, then using single global commands like "replace formal language with plain English" instead of editing text boxes one by one. The framing is direct: executives waste premium hours acting as junior graphic designers, while a complete first draft in under two minutes frees mental bandwidth for the message and strategy.
For brands planning a campaign
Sponsors in document automation, presentation software, and productivity suites have an obvious entry point. The argument targets the person whose hours are expensive — the firm owner — rather than a junior staffer.
Across the nine posts, the only third-party businesses Kathy D'Agostino names are AI software vendors, recommended as tools rather than identified as sponsors, clients, or partners.
Read more
No post discloses employment, business ownership, an agency relationship, an affiliate arrangement, or a paid partnership.
ChatGPT and Claude appear repeatedly in recommended workflows; Gemini, Perplexity, and Gamma each appear in specific task examples.
Kathy sells their own services and offers free guides through direct-message prompts, but does not name the business or firm they work through.
Ask for their employer or business affiliation.
OpenAI (ChatGPT)
Mention
Recommended for strategy and writing, building a bottom-up pricing and scenario model, and role-assignment prompting.
Read more
Presented as a product recommendation, with no disclosed commercial relationship.
Named as the engine that generated the graphic in the data-visualization post, praised for complex narrative compositions and native text labeling, and recommended for teams already working in Google Workspace.
Read more
Presented as a product recommendation, with no disclosed commercial relationship.
Kathy D'Agostino's posts offer a working path through AI adoption for firms with fewer than ten people.
Read more
The fastest way to explore them is by problem rather than date.
Start with the two posts on framing: one replaces a one-line instruction with a Goal-Obstacle-Constraint-Context rule mapped out on paper before any tool is opened; the other assigns the model a role such as "You are my sales strategist," with before-and-after times for proposal drafts and call prep.
Next, read the 30-day rollout for small firms and the three-stage maturity test, which show how they approach adoption: audit where time leaks, pick one bottleneck, measure it, then repeat.
The governance posts on shadow AI and manufacturing data rules address risk.
From there, the subjects narrow to using a model to price a project, producing proposals and decks, and AI-generated data visualization, a topic aimed more at designers than operations leads.
Several posts end with a guide or worksheet sent by direct message, so ask for those resources if you need more than the posts themselves.
Like D'Agostino, Futia writes tool-specific, do-this-next posts rather than general commentary about AI.
Read more
Both take one model and show the concrete jobs it can handle: Futia turns a single product photo into dozens of campaign images by uploading reference shots for product accuracy, swapping models for persona tests, and rendering text inside the image so a hook can be tested in any language; D'Agostino walks through naming a client and scoping a one-page proposal, or mapping a task's outcome, friction, constraints, and context before opening a tool. Both post on LinkedIn, end with a practical playbook, and measure the change in hours or weeks rather than enthusiasm.
Futia focuses on creative production for ecommerce and agency teams — photoshoot costs, ad fatigue, aspect ratios, localization — and builds his material around one image-generation model. D'Agostino writes for professional-services firms and small manufacturers, emphasizing adoption processes, data policy, and what a firm should stop paying for, with creative output as a side topic rather than the center.
Patel teaches the same underlying move as D'Agostino: stop re-explaining the task and encode the way you work once so the model can reuse it.
Read more
His Claude Skills post describes writing a creative process as a simple markdown file that triggers automatically for research, briefs, and copy, and composing overlapping skills in one conversation; D'Agostino's equivalent is assigning the model a role and reusing a structured four-part brief instead of retyping instructions.
Both write for owners and small teams already paying for tools and seeking repeatable output.
They also extend the argument to how AI changes how businesses are found or chosen, with Patel focusing on SEO, AI Overviews, and AI recommendations.
Patel's audience is ecommerce brands and creative agencies, and his material is closer to a tool-setup manual — skill architecture, templates, composability. D'Agostino focuses on operations and leadership: maturity stages, written data rules, pricing decisions, and who inside the firm is responsible, without the same depth of tool configuration.
Pascua and D'Agostino diagnose the same operational waste.
Read more
Pascua counts the two-to-five-minute tax of hunting for an update, an SOP, a file, or a task across Asana, Sheets, Docs, and Drive, then builds a single source of truth in Notion with dashboards for sales, marketing, ops, finance, legal, and exec, plus templates and automations.
D'Agostino describes documents that take too long and look different depending on who writes them, then recommends a repeatable brief, a shared prompt, and solving one bottleneck at a time.
Both write as practitioners describing their own builds, target small businesses and solopreneurs, and offer paid or gated follow-up resources at the end of a post.
Pascua's work is product-shaped: Notion templates sold through a shop and custom workspace builds, so the recommendation is usually a specific tool and setup to buy. D'Agostino's work focuses on advisory and training — frameworks, a 30-day rollout, policy language, and tool selection across ChatGPT, Claude, Perplexity, Gemini, and Gamma — with no single platform at the center.
Kathy D'Agostino's posts target people who run small businesses and fix internal process problems: owners and senior staff at professional-services firms — law, accounting, agencies, consulting — and small to mid-sized manufacturers.
Read more
They name that readership directly, use hashtags such as #SmallProfessionalFirms, #SmallBusiness and #SolopreneurLife, and usually close with an action for whoever owns the process: DM a keyword for a guide, checklist or worksheet.
One post carries #HudsonValleyManufacturing, a regional thread alongside the national small-business framing.
The only published audience figure is their LinkedIn follower count: 2,755 as of October 8, 2026.
The content maps to campaigns aimed at small-firm operations decision-makers interested in AI adoption strategy and risk mitigation, covering written AI policies, data-handling mistakes, price and scope modelling, proposal and slide production, and repeatable process design.
The posts don't establish whether the audience includes the operations managers you want to reach.
For a placement decision, ask D'Agostino for recent post-level results — impressions, reactions, comments, and any click or DM conversions — for two or three posts closest to your campaign, plus any follower-demographic data or audience survey they can share.
If reaching operations leaders rather than small-business owners generally is essential, ask specifically for job titles and seniority.
Kathy D'Agostino's posts do not offer a sponsorship, paid-placement or co-marketing format.
Read more
Instead, they show a consistent inquiry mechanic she runs herself: several posts ask readers to send a LinkedIn direct message with a keyword—“Day 31,” “AI blueprint” or “AI audit”—in exchange for a quick-start guide, worksheet or self-assessment checklist.
Others ask readers to comment “CODEX,” “slide deck,” “pricing audit” or a 1/2/3 answer to receive a one-page resource.
These are lead-capture routes for their own advisory and training work, not published partnership inventory.
At least one post says plainly that working through AI adoption with a firm is what they do.
Before planning a campaign, ask about rates, deliverables, lead time, disclosure, and their policies for client data and AI tools—a precondition their posts set for the AI work they take on.
Strong fit. The content topic, framing, and stated readership all line up with this brief. Before money moves, agree on partnership terms and substantiate claims carried over from D'Agostino's posts.
Read more
Brief
A vendor of secure or enterprise-grade AI accounts, plus data-protection and policy tooling, wants to reach owners and operations leads at small professional-services firms — law, accounting, agencies, consulting — and small manufacturers. The campaign theme is governing AI use and keeping confidential client or customer data out of free public tools.
Reasons
The Shadow AI post offers a campaign theme for buyers in this category. It lays out three exposures they sell against: privacy and data leakage when free public models train on their inputs, accuracy failures when unreviewed output reaches client deliverables, and third-party partner risk when vendors feed unverified data back into operations. It also speaks to buyers at their level: the post positions safe adoption as a leadership mandate, not a technical problem, and addresses the owner who signs off on the purchase.
A second post offers countermeasures for a campaign: verify which model and privacy settings are in use, then re-check them because defaults change; write an explicit policy specifying what data may be entered (operational friction points, logistics, general problem-solving) and what may not (proprietary code, customer data, client schematics); appoint an AI Champion who is explicitly not a technologist; and start with one data-heavy bottleneck rather than the whole company. These actions and a named role provide a campaign spine for a worksheet, workshop, or serialised series.
The three-stage maturity ladder gives the campaign a diagnostic, not just an opinion. Stage 1 is quiet use of free tools with no policy or visibility; Stage 2 is paid secure accounts, a basic data-security policy, and shared prompts for recurring work; Stage 3 is AI trained on the firm's own methods and funded as infrastructure. The test D'Agostino offers — cancel every AI subscription tomorrow and ask whether the work actually changes — works as a quiz or self-assessment lead magnet. The sponsor's product fits naturally into the Stage 1-to-Stage 2 move from free tools to paid, governed accounts.
The posts address owners and senior staff at professional-services firms and small to mid-sized manufacturers, as indicated by hashtags such as #SmallProfessionalFirms and closing actions aimed at whoever owns the work. That is also the campaign's target role.
The format is reusable. Numbered frameworks, day-by-day roadmaps, checklists, and before-and-after time estimates recur across the posts. Several already offer a guide, worksheet, or self-assessment in exchange for a direct message or comment. A gated resource campaign can use this existing mechanic.
D'Agostino already teaches this material to the audience the brief wants, in plain operational language — 'if it's not written down, it's not a policy' — the register a risk-and-governance campaign needs, but most AI-adoption content lacks.
Tradeoffs
The posts document no sponsorship format, rate card, deliverable list, lead time, or disclosure statement. Their inquiry mechanic is a direct message in exchange for a free resource, aimed at D'Agostino's own advisory work. Agree on terms, disclosure, and fulfilment before proceeding.
The posts teach method rather than pitch products. A sponsor expecting product-led demonstrations must supply the demo, comparison, and proof.
Claims in the posts — an audit of more than 40 small professional-services firms, an estimate that about 80% sit at Stage 1, and the statement that 80% of manufacturers plan to increase AI investment — are reported as D'Agostino's own observations, not independently verified. The sponsor should substantiate them before republishing.
D'Agostino reports a LinkedIn follower count of 2,755 as of October 8, 2026. The posts give no impressions, engagement, job titles, or seniority data. Ask for those metrics before relying on the follower count as evidence of reach to budget-holders.
When to choose a specialist
Poor fit as briefed. It could work if the target buyer shifted to the owner or operations lead who approves the purchase, and the content moved beyond the implementation layer.
Read more
Brief
A developer-facing LLM platform — offering evaluation, fine-tuning, and deployment tooling — wants hands-on implementation content for data engineers and platform or IT leads at mid-market firms, with trial signups from technical practitioners as the objective.
Reasons
The reviewed work addresses non-technical owners and operations leads. The manufacturer post explicitly says the AI Champion need not be a technologist; it asks for someone curious and organised to track privacy settings and model changes. That is the opposite of the reader a fine-tuning platform needs to reach.
None of the nine posts contains implementation material: no code, architecture, or model evaluation or deployment methodology. Tool guidance stays at the level of which tool fits which job — ChatGPT for strategy and writing, Claude for document analysis, Perplexity for sourced research, Gemini for teams already in Google Workspace. The stated rule is that the best tool is the one that fits how a team already works. Even the data-visualisation post, the most tool-focused item in the set, compares output quality and ways of working, not technical capability.
The posts do not identify engineers, data teams, or IT decision-makers as part of D'Agostino's audience. They describe the intended readership as owners and senior staff in professional services and small manufacturers, and give a follower count, but no titles, seniority, or industry composition. Ask for audience data before committing to a technical trial-signup goal.
The commercial mechanics also do not support a performance-style placement. The posts document no sponsored format, rate card, or measurement practice. Before planning a paid placement with defined deliverables, disclosure, and tracked conversions, ask for those details.
This is a mismatch between the reader and the brief, not a judgement that D'Agostino's material is weak. A platform whose actual buyer is a small-firm owner weighing a move from free to paid tools would be a different brief, much closer to the good fit above.
A worked campaign example
Concept
The Shadow AI Audit: a governance campaign built on D'Agostino's three-part risk framing and three-rule countermeasure, for a vendor selling secure AI accounts and policy tooling to small professional-services firms and small manufacturers.
Content spine
Post or article one — the diagnostic. Explain the three-stage ladder in plain terms and include the cancel-your-subscriptions test. Call to action: invite readers to reply with their stage number, as the original post does. The campaign opens with a question readers can answer in one word.
Post or article two — the risk. Cover the three exposures: free public models that train on inputs, unchecked output reaching client deliverables, and partners feeding unverified data back into operations. Keep the framing at the owner's level, not the technical level.
Post or article three — the fix. Offer three actions: confirm the model and privacy settings and re-check them on a schedule; write down what data may and may not be entered; appoint an AI Champion who is not a technologist. Close with the single-bottleneck starting point.
Gated resource. Create a one-page AI policy worksheet using the two-column list from D'Agostino's manufacturer post: operational friction points, logistics, and general problem-solving on one side; proprietary code, customer data, and client schematics on the other. Add the short self-assessment. The sponsor can present its paid secure-account product as the route out of Stage 1 when delivering the worksheet.
Recurring beat. D'Agostino's caution that settings, plans, and defaults change over time can become a quarterly privacy-settings update, giving a retainer a reason to continue after the launch series.
Fulfilment and call to action
Reuse the mechanic already visible in the posts: invite people to comment or send a direct message with a keyword to receive the resource. If leads belong to the sponsor, the sponsor should handle fulfilment and follow-up. State plainly that the content involves a commercial relationship.
Proof and guardrails
Attribute D'Agostino's internal figures to them as personal observations, or replace them with figures the sponsor can substantiate. Do not present them as independently verified.
Do not describe a product control or feature as working unless it is built. The campaign is about policy and account choice, so describe the sponsor's actual capabilities at that level.
Confirm the audience assumption early. Before committing a full budget, ask for post-level results on the two or three posts closest to this theme—including impressions, reactions, comments and direct messages—and any shareable follower breakdown by title or seniority. These results could materially change the plan.
Agree terms, disclosure, deliverable list, and lead time in writing, since no sponsorship format is documented and nothing can be assumed from the posts.
What success looks like
Readers identify their stage and request the policy worksheet. The sponsor can track how many requests come from owner or operations-lead roles—the group that decides whether a firm moves off free tools.
Why this angle
The campaign draws on D'Agostino's strongest content, addresses the budget holder and turns the topic into a gated resource they already offer in a different form. It educates first and positions the product at the end, matching the structure of the posts.
Consider Mike if the campaign shifts from AI governance for small firms to AI-generated ad creative for direct-to-consumer brands.
Read more
His posts describe turning product photos into campaign variations and using AI for persona testing, ad formats, and localization.
That is a different buyer and campaign task. The posts focus on creative production, not policies for confidential data, risk controls, or AI adoption across professional-services firms and manufacturers.
Consider Mitesh if the campaign shifts to ecommerce discovery or AI-supported creative workflows.
Read more
His posts address how brands appear in traditional search, AI Overviews, and AI recommendations, and describe using Claude Skills to standardize creative work.
Neither post addresses the governance and risk-mitigation task in a campaign aimed at small-firm owners deciding how to govern confidential business data in AI tools. The posts target ecommerce brands and creative teams, rather than those small-firm owners.
Consider Aaron if the campaign’s main need is organizing small-business operations in Notion.
Read more
His posts discuss building a centralized workspace for tasks, procedures, sales, and other business information, including the problems that arise as a business grows.
The posts make a case for operational systems, not AI adoption strategy or AI risk mitigation. Ask Aaron for guidance on data protection and AI policies before considering them for a campaign focused on those topics.
Nine LinkedIn posts by Kathy D'Agostino on her profile on October 8, 2026 cover AI adoption roadmaps, prompt framing, data and governance rules, pricing and scenario modeling, document and deck production, and AI-generated data visualization.
The posts do not show post dates, so the order of her thinking and any change over time cannot be established.
The topics span small professional-services firms, with one post aimed at small and mid-sized manufacturers.
The posts here do not include the guides, worksheets, and checklists several posts offer in exchange for a direct message or comment; video or transcripts; or any books, shows, or earlier career work.
Those remain open gaps for anyone researching her name beyond her current AI-adoption writing.