I had over forty chief executive officers review this post tonight before I shipped it. A chief technology officer at a four point one billion dollar financial services company. A chief financial officer at a three hundred eighty million dollar annual recurring revenue software as a service company. A security chief at a forty-five billion dollar regional bank. A vice president of engineering at a Series C startup. Four chief people officers, one at a publicly traded enterprise, one at a private equity backed manufacturer, one at a six point eight billion dollar insurance holding company with unionized claims processors, and one at an eighty-five person AI-native startup in Brooklyn. They read the drafts, gave me specific feedback, and told me what made them want to close the tab. I revised the post four times based on their objections.
None of them were real.
Every one of those reviewers was a synthetic user, an LLM agent calibrated to behave like a specific executive with a specific background, specific pressures, and a specific disposition toward what I was selling. I gave them browsers. They navigated my site. They read my copy. They told me where I was wrong. The technology chief told me my title was boring and I changed it. The finance chief told me my economics section was napkin math, not a business case, and I rewrote it. The security chief flagged five security concerns I had never considered. The skeptical people officer at the manufacturer gave me a three out of ten on trust because I had not addressed what happens to the humans whose jobs this technology partially replaces, so I added a section on workforce transition and his score went to a six. The entire WordPress theme and plugin powering this site was designed the same way, iterating on executive synthetic user feedback alongside a synthetic user experience designer who audited every template for consulting-grade visual standards.
Look. That is what a synthetic user is. An agent with a browser, a persona, and opinions. It changes the fundamental economics of learning what your customers think.
Most teams are doing this badly. The traditional version of what I just described costs eight thousand dollars to fifteen thousand dollars and takes two to four weeks. You recruit eight people, put them in a room, and by the time the findings deck lands on someone's desk, the product has already shipped or the window has closed. The feedback is real. The timing makes it useless.
I wrote about the broader shift toward agent-driven validation in One Hundred Proof of Concepts a Day, where agents build, test, and discard a hundred proof of concepts overnight. Synthetic users are the same principle applied to feedback. Let the agent do the cheap validation so the humans can focus on the hard decisions. The organizations that get the rigor right build a capability that compounds, because every real conversation makes their synthetic users smarter, and every synthetic run makes their real conversations more productive.
But here is the thing. Most teams are not getting the rigor right. Either the team writes a two-paragraph prompt, calls it a persona, and gets back feedback that sounds authoritative but is untethered from real reader behavior. Or the team buys a self-serve tool, runs a survey against synthetic respondents, and treats the output as real market research.
Both produce confident-sounding fiction. The problem is not immature tooling or pressure to show AI adoption. It is that most teams have not done the hard calibration work that separates useful feedback from expensive hallucination.
So, what is a synthetic user actually? A synthetic user is not a chatbot wearing a name tag. It is an LLM, or large language model, agent calibrated to a specific role, industry, decision-making style, and set of priorities, with its own objections baked in.
Imagine you have a chief technology officer at a Fortune Five Hundred financial services company. She has been in the role for three years. She inherited a legacy modernization initiative eighteen months behind schedule. She reports to a board that wants AI wins on the quarterly earnings call. She has been burned by two consulting firms who promised transformation and delivered slide decks. She is skeptical, time-poor, and her default answer to a new vendor is no.
She was not real. But when you put your website in front of that agent and ask whether she would send this to her peer, the answer is a structured evaluation grounded in the same priorities and skepticism your actual reader carries.
I run about thirty of these against my own site. A private equity backed technology chief under margin pressure evaluates differently than a government chief information officer navigating the Federal Risk and Authorization Management Program. A security chief evaluates differently than a chief marketing officer. Our target audience is not one person. It is a dozen people with conflicting priorities, and you need to hear from all of them before you ship.
Not all of them should like what they see. Some of the most valuable synthetic users are anti-personas, agents deliberately calibrated to be hostile to your offering. My synthetic security chief is almost never happy. He flags security concerns on every page, questions every data flow, and gives me a trust score that rarely breaks a six. If your synthetic security chief loves your site, your synthetic security chief is broken. The anti-persona tells you what your hardest reader will actually think, and if you can survive that review, the real version is not going to surprise you.
This is not limited to external readers. I built one that represents an engineering manager at a two hundred person company that just announced an AI-first transformation. She manages twelve engineers. Three are worried about being replaced. Two are excited and already using Copilot without telling anyone. She has been told to redefine team roles for AI-native delivery but has received no framework, no budget for reskilling, and no clarity on what AI-native means. When I put a new role description or performance rubric in front of that agent, the feedback comes back grounded in the specific pressures of someone navigating that exact organizational moment. Your people team can hear the objections before the all-hands, not after.
OK. Let's talk about how they can use your software. Most synthetic user implementations stop at the survey layer. They ask questions and generate responses. That misses the thing that actually kills your conversion, which is the experience of navigating your product.
A synthetic user is an LLM agent. An LLM agent can use tools. The most important tool you can give it is a real browser.
Playwright, a browser automation framework, is the most common orchestration layer for web-based products. But the capability goes further. Anthropic's computer use application programming interface lets Claude take full control of a desktop environment, moving the mouse, opening applications, typing into native software, reading the screen, and deciding what to do next the same way a human would. OpenAI's Operator and Google's Project Mariner do similar things. These are production-capable tool-use interfaces, not research previews.
The frontier is moving beyond screens. Projects like OpenClaw and other open-source robotics frameworks connect LLM agents to cameras and physical sensors. A synthetic user that can see your physical product through a camera, evaluate the packaging, and assess the unboxing experience, that is early but not science fiction. The architecture is the same. A calibrated persona, a set of tools, and a feedback loop. Today the tool is a browser. Tomorrow it might be a webcam pointed at your retail shelf. The persona and calibration do not change. Only the interface does.
Here is how the browser loop works. An orchestration layer launches a Playwright session against your target page. At each step, the system captures a screenshot or accessibility tree snapshot and passes it to a vision-capable model along with the persona's calibration context. The model returns a structured action, such as click, scroll, type, or evaluate. Playwright executes and the loop continues.
You can point this at any browser-accessible surface. A fleet management dashboard behind a login, an internal procurement workflow, or a patient intake form in staging. The agent authenticates with a test account, navigates in role, and evaluates against the expectations of the person it represents.
I showed the output from one of these runs to a vice president of engineering at a financial services company I have worked with. She read the synthetic technology chief's objections and said, that is exactly what I said in the vendor review last quarter. How did it know that? It did not know. It was calibrated against the same behavioral patterns that informed her real objections. That was the moment I stopped thinking of this as an experiment and started thinking of it as infrastructure.
I run six synthetic users in parallel against a page. Twenty minutes total. You can do this in Claude Code, GitHub Copilot, Gemini CLI, or any agent framework that supports tool use. Roughly ten to fifteen percent of browser runs hit an obstacle that requires a retry, usually a modal or dynamic loading state. The other eighty-five to ninety percent produce actionable feedback. The failure mode is the agent got stuck on a cookie consent banner, not the agent gave me dangerously wrong advice.
Now, your marketing team can talk to them. A synthetic user is not a one-shot evaluation. It is a persistent agent. You can have a conversation with it.
Your marketing team can sit down with a synthetic technology chief and ask follow-up questions. You said the messaging was unclear. What specifically would you need to see in the first paragraph to keep reading? The agent responds in character, grounded in its calibration, with specific feedback.
I stopped treating synthetic users as evaluation tools and started treating them as simulated stakeholders I could workshop ideas with. I draft a homepage headline, run it past a synthetic chief financial officer and vice president of engineering, read their reactions, revise, and run again. The iteration cycle that used to take weeks, write, ship, wait for analytics, interpret, and revise, now takes an afternoon.
Your product team can do the same with feature messaging. Your sales team can rehearse objections. And the people who used to spend three weeks recruiting real technology chief interviews? They are now refining calibration documents, validating synthetic output against real conversations, and designing questions that make the real interviews sharper. The job did not disappear. The job got better.
Because the agent has a browser, you can say, go look at our competitor's pricing page and tell me how it compares to ours from your perspective. The agent navigates both sites, compares them in character, and gives you feedback specific to how that reader type evaluates competitive alternatives.
Right. Where do they fit? Every phase of the product lifecycle. This is what makes synthetic users a core capability. They fit everywhere a human would give you feedback, and the persona you need changes at each phase. Once you build the calibration infrastructure, you have a feedback engine that serves product, engineering, marketing, sales, customer success, and the people team simultaneously.
Start with discovery and research. Before a single engineer touches the codebase, a synthetic panel of your ideal customer profile evaluates your product brief. If three of five say the premise is weak, you saved a quarter of engineering time. I killed two content ideas based on synthetic panel feedback that would have taken weeks to write and months to discover were wrong.
Then there is build and validation. While the product is being built, synthetic users navigate your staging environment. A synthetic dispatcher evaluates your fleet management dashboard and might say, I expected to filter by vehicle status before route, but this forces me to pick a route first. You can integrate this into your continuous integration and continuous deployment pipeline as a pre-deployment gate. If a synthetic user that previously rated a flow positively now rates it negatively, the pipeline flags the regression before promotion.
Consider launch and go to market strategy. Your synthetic reader panel evaluates positioning, competitive messaging, and the launch page before the campaign goes live. If your marketing team is spending two hundred thousand dollars on a LinkedIn campaign, run the landing page past a synthetic version of your ideal customer profile first.
Next is pricing and packaging. You do not guess what the finance chief will object to. You build a synthetic chief financial officer calibrated to your actual reader profile and let her navigate the pricing page. Then build a synthetic startup founder with a different budget reality and compare the objections. Iterate before the first sales call, not after the first lost deal.
In sales, run your pitch past a synthetic version of your prospect's role and industry. The agent surfaces objections your team will hear in the real meeting, giving them time to prepare actual answers instead of improvising.
For retention and expansion, a synthetic churning customer evaluates your quarterly business review deck and tells you, this felt like a product demo, not a conversation about my business outcomes. A synthetic power user tells you whether the upgrade path is clear or feels like a bait and switch.
Finally, internal and people operations. A synthetic engineering manager evaluates your reorg announcement. A synthetic senior individual contributor tells you the new performance rubric does not account for prompt engineering. A synthetic new hire navigates the real onboarding flow through Slack, Notion, and the human resources information system, and tells you where she would give up and message her manager instead. The cost of getting a product page wrong is a bad quarter. The cost of getting a performance framework wrong is a year of attrition in your highest-leverage roles.
So, how do you build them? Most people expect me to say just write a good prompt here. That is like saying just write good code. Technically true. Practically useless.
Building a synthetic user that gives you useful feedback requires four things.
First, a calibration document. This is the foundation. It defines the persona's role, industry, organizational context, decision-making authority, risk tolerance, known objections, competitive alternatives, and default disposition toward your category. What makes them skeptical. What makes them lean forward. What makes them close the tab.
The calibration document is not a paragraph. It is a page, sometimes two. The ones that work best are grounded in the most specific behavioral detail. My synthetic technology chief for financial services runs two pages. The calibration for a synthetic startup founder is shorter because the persona is less constrained. Specificity is not about length. It is about knowing which details change the output.
Second, behavioral anchoring to real humans. This is where most efforts fail. If you invent the persona from your marketing team's assumptions about the reader, you will get feedback that confirms those assumptions. You are testing your assumptions against your assumptions.
Ground it in actual customer interviews, sales call recordings, objection patterns from your customer relationship management system, and churn reasons from your customer success team. If the model is built on assumptions instead of evidence, the feedback is fiction. A research director at a mid-market software as a service company once told me this whole approach was expensive hallucination with a persona costume on. Without behavioral anchoring, she was right.
The best synthetic users are calibrated from the entire surface context of your application. Support tickets, social mentions, application logs, net promoter score comments, community forums, sales call transcripts, and churn surveys. Every surface where a customer expresses an opinion or encounters friction is a calibration input. I wrote about how to instrument this in The Customer Product Operating Model.
A note on data privacy. Use anonymized and aggregated patterns, not raw transcripts. If you are sending calibration data to a third-party model provider, ensure your data processing agreement covers this use case. Regulated jurisdictions such as those under the General Data Protection Regulation, the California Consumer Privacy Act, and the Health Insurance Portability and Accountability Act require legal review before feeding interview data into any external model. This is not optional.
If you do not have four data sources, start with win and loss analyses from your sales team. It is the highest-signal starting point. You can build a useful synthetic user from win and loss data alone and refine it as you get more inputs.
Third, behavioral validation. Once built, test it against known stimuli. Give it a competitor's website or messaging that performed poorly and compare its response to what real humans said. Do the objections overlap?
I measure this as an objection overlap rate, which is the percentage of objections raised by the synthetic user that also appear in real customer feedback. After three rounds of calibration, my synthetic technology chief hit roughly seventy to seventy-five percent objection overlap with feedback from real technology chiefs I have spoken with. I am not really good at mathematics, but I think that means the agents are catching about three out of four real objections, right? The twenty-five to thirty percent they miss tends to be context-dependent. Relationship history, internal politics, or budget timing. Things a simulation cannot know because they live outside the page.
Fourth, ongoing refinement. Every time real feedback surprises you, ask whether your synthetic user would have caught it. If not, update the calibration. A calibration update takes ten minutes. The validation run takes twenty.
I do this once a week. I compare real traffic and bounce rates to what my synthetic chief officers predicted. Did the synthetic technology chief say this post would get forwarded? Did it? Did the synthetic finance chief say the economics section was weak? Did readers drop off there? The gap between prediction and reality is the calibration signal. Every week, that gap gets smaller.
Over time, you create new personas as patterns emerge. Traffic from a segment you had not considered, such as a mid-market head of data, a non-technical chief executive officer, or a procurement officer, and you build a synthetic user for it. The persona library grows with your understanding of your market.
Right. What can go wrong? I would trust this methodology less if I only told you what it does well.
Synthetic users can be confidently wrong. I had a synthetic chief financial officer tell me my pricing page needed a detailed return on investment calculator before the call to action. Real finance chiefs wanted fewer fields, not more. The synthetic user was reasoning from a generic finance chief archetype instead of the specific type of finance chief who visits a consulting site. Recalibrating to a time-poor finance chief evaluating a services engagement, not a software purchase, fixed it. But if I had acted on the original feedback without validation, I would have made the page worse.
They do not model trust accumulation. A real reader builds trust over multiple touchpoints. A blog post, a LinkedIn comment, a referral, then a site visit. A synthetic user evaluates a single page in isolation. It cannot simulate relationship history.
They are not accessibility audits. A synthetic user can flag confusing navigation or unclear content hierarchy. It cannot replicate the experience of a screen reader user or a keyboard-only navigator. Accessibility requires specific tooling and real users with real accessibility needs.
Internal synthetic users carry higher stakes. Bad feedback about a pricing page costs you a suboptimal landing page. Bad feedback about a reorg plan or career ladder costs real people real career decisions. A synthetic panel can tell you the reorg announcement will confuse engineering managers. It cannot tell you the reorg will break a trust relationship between two specific leaders that took three years to build.
The speed can create false certainty. Twenty-minute feedback cycles are seductive. The risk is that teams treat synthetic feedback as ground truth and skip the real research. Synthetic users are a pre-filter, not a verdict. The moment you stop validating against real humans, you are optimizing for your model of the customer instead of the actual customer.
Look at the economics. A typical eight-person usability study costs eight thousand dollars to fifteen thousand dollars and takes two to four weeks. A synthetic panel run takes twenty minutes and the tooling is effectively free if you already use Claude Code, Copilot, or Gemini CLI.
The interesting question is what happens to the people who do usability research today. Your best user experience researchers become persona calibrators, the people who ground synthetic users in real behavioral data, validate against real outcomes, and catch the confident fiction that uncalibrated agents produce. The role changes. The skill set evolves. If your AI adoption strategy starts with headcount reduction, you will lose the institutional knowledge that makes the AI useful in the first place.
The real investment is time. Four to eight hours per persona for the initial calibration, a day or two of engineering for the orchestration layer if your team already uses Playwright, and two to four hours per month recalibrating against real-world feedback. Unlike a usability study, the capability does not reset to zero after each engagement. The calibration documents get better. The persona library grows. The validation data accumulates. This is an asset that compounds, not an expense that evaporates.
That does not mean you skip the usability study. You run the synthetic panel first, fix the obvious problems, then run the study on a version of your product that is already better. The study stops catching obvious problems and starts catching the hard ones.
A redesign takes eleven weeks to build. Six weeks to get feedback. Then a rework because the messaging missed. Two engineers for eleven weeks is roughly eighty-five thousand dollars in loaded salary, plus the opportunity cost of the six-week delay. A synthetic panel before launch would not have caught everything, but it would have caught that this messaging lands with directors but not technology chiefs. That single insight, twenty minutes before launch instead of six weeks after, saves the entire rework cycle.
We need better research, not less research. Synthetic users do not replace user research. If you use them as a replacement, you will build a product optimized for your model of the customer instead of the actual customer.
But that is not a weakness. It is a description of how to use them correctly. Before you run a usability study with eight real participants, run the same flow past your synthetic panel. The panel catches the obvious problems. You fix them. The real participants see a better version. Their feedback is more useful because they are not wasting time on problems you could have caught without them.
It is the same logic as linting before code review and code review before quality assurance. Catch the cheap problems early so the expensive validation steps can focus on the hard ones.
Start with one. Do not build thirty synthetic users. Build one.
Pick your most important reader persona. The one whose objections keep you up at night. Build a calibration document grounded in real data. Win and loss analyses are the fastest starting point. Not a generic enterprise technology chief, but someone specific. The industry, the company size, what she inherited, what her board is asking for, and what burned her last time.
Point that synthetic user at one page. Your homepage. Your pricing page. Whatever matters most.
Read the feedback. Compare it to what your real customers have told you. If the synthetic user surfaces objections that match what you have heard from real people, you have a working tool. If it misses, recalibrate and try again.
One synthetic user, one page, one hour. That is the cost of finding out whether this works for you.
And once it works, think about what happens when you add synthetic users to your agent swarm. You are not running one panel manually and reading the results over coffee. You have agents building the product, agents testing the product, and now agents evaluating the product as calibrated readers, all running in parallel. The iteration speed changes completely. I revised this post four times tonight based on synthetic executive feedback. Four full rewrites, with scores tracked across every iteration, in a single session. How long does your current feedback cycle take to produce four validated iterations of anything?
How long is your current feedback loop? From shipping a change to learning whether it worked for the right reader? Six weeks? Eight? A full quarter?
During that gap, how many decisions are you making based on what you think the customer wants instead of what a calibrated, validated synthetic version of that customer would tell you?
You do not have to answer that to me. But you should answer it to yourself. Every week that gap stays open, you are shipping into the dark. The organizations closing it are still talking to real customers. They are just arriving at those conversations with better questions because they already eliminated the obvious mistakes.
How many obvious mistakes did you ship last quarter that a twenty-minute panel run would have caught?
Hope is not a feedback strategy. Your product is not a fiction project. What happens if you do not get it right?
Now, let's look at what the synthetic chief officer panel actually said about this post. Every reviewer below is a synthetic user. They were calibrated to specific executive profiles, given access to this article, and asked to score it, react honestly, and answer whether they would send this to a peer.
Jordan Whitfield, a chief technology officer at Ashford National Group, a four point one billion dollar financial services company with twelve hundred engineers, gave it a seven out of ten. He liked the calibration methodology, particularly the objection overlap rate and weekly recalibration. He suggested handing it to a platform team for a prototype. However, he docked points because the economics section treats compliance as a footnote, which is a critical path at his scale. He would send this to his head of platform engineering and his vice president of client experience.
Rachel Goldstein, the chief financial officer at Vertex Software as a Service, with three hundred eighty million dollars in annual recurring revenue, gave it a six out of ten. The unit economics framing got her attention, specifically the eighty-five thousand dollar salary cost for a rework cycle. But she needed a full cost model including calibration time and break-even points for different persona counts. She would not send this to a peer but would send it to her chief people officer to read the economics section before asking for more headcount.
Kevin O'Brien, the security chief at Ridgeline Federal Savings, a forty-five billion dollar financial services firm, gave it a seven out of ten. He called the anti-persona concept the first honest thing he had read about security roles in two years. He would want more detail on the data flow architecture and data processing agreement coverage. He would send this to his vice president of application security as a threat model exercise.
Marcus Thompson, a vice president of engineering at RouteCast, a Series C company with forty-two million dollars in annual recurring revenue, gave it an eight out of ten. The browser-based validation loop stopped him from skimming. He noted that his team recently spent six weeks on a dashboard redesign that failed because nobody simulated a dispatcher under pressure. He would send this to other vice presidents of engineering in his Y Combinator batch.
Elena Wright, the chief people officer at Clareo Health, a nine hundred million dollar digital health company, gave it an eight out of ten. She screenshotted the internal synthetic user section for her leadership team. She found the idea of testing reorg announcements against calibrated agents to be a critical checklist for organizational change. She would send this to the human resources chief at a one point two billion dollar digital therapeutics company.
Derek Rawlings, the chief people officer at Ridgewell Industrial, a two point four billion dollar company with twelve thousand employees, gave it a five out of ten. He felt the article ignored the impact on non-technical employees, such as his seventy-two hundred quality inspectors. He called it another case of consultants selling transformation to the people who need it least. He would not send it to a peer but would send it back to the author with a note to write about the workforce impact.
Zara Okonkwo, the head of people at Cadence Labs, an eighty-five person startup, gave it an eight out of ten. She admitted her team has been doing a lazy version with simple prompts and made-up personas. The calibration rigor section hit her hard. She plans to take the framework to her next people operations sync and would send the article to other heads of people at startups.
Margaret Chen-Liu, the chief people officer at Harborline Insurance, a six point eight billion dollar company with twenty-two thousand employees, gave it a six out of ten. She saw applications for testing modernization communications but felt the article did not understand institutional trust debt from failed change programs. She would send it to her head of internal communications as background reading.
Sarah Chen, the chief information officer at Pacific Northwest Health System, a three point two billion dollar healthcare organization, gave it a five out of ten. She found the data privacy treatment too shallow for a regulated health system. Before piloting, she would need to know exactly where calibration documents are stored and what the business associate agreement coverage looks like. She would not send it to another healthcare executive.
Tom Brennan, the chief information officer at Midwestern Mutual Insurance, an eight billion dollar company, gave it a four out of ten. With thirty years in the industry, he was skeptical that an LLM could truly produce feedback equivalent to an actual technology chief. He noted that a seventy percent overlap rate means it is wrong thirty percent of the time, which often includes the most important details. He would only reconsider if he saw a live demo with his compliance team present.
Michael Zhang, the chief technology officer at Stratal Software, a private equity backed company with two hundred million dollars in revenue, gave it a seven out of ten. He liked the idea of pre-empting objections from private equity sponsors on a pricing page. He wanted more information on how to sell this internally when investors are focused on headcount reduction. He would send this to other portfolio company technology chiefs in his fund.
Diane Foster, the non-technical chief executive officer at Lakeview Health Partners, a five hundred million dollar healthcare company, gave it a six out of ten. She understood the opening story but felt she lost the thread during the technical sections on Playwright and continuous integration. She needed a clearer explanation of how this affects patient satisfaction and margins for her board. She would describe the concept to her chief operating officer but would not send the article.
David Park, the chief operating officer at Draymark Logistics, a one point eight billion dollar freight company, gave it a seven out of ten. The dispatcher example resonated with him, as a recent two point three million dollar redesign was rejected by operations staff. He wondered how to scale calibration for fourteen thousand workers across two hundred facilities. He would send this to his vice president of process excellence and to a peer chief operating officer.
Raj Patel, the data chief at Ashford Capital Holdings, a Fortune One Hundred company with thirty-eight billion dollars in assets under management, gave it an eight out of ten. He felt the calibration rigor separated this from synthetic data theater. He would send this to his counterparts at other large banks and to his head of model risk management.
Patricia Williams, the human resources chief at Brevian Analytics, a one point two billion dollar company with thirty-two hundred employees, gave it a seven out of ten. She felt the article focused too much on product and marketing. She wanted to see more examples related to organizational design, such as career ladder rollouts and manager enablement. She would send this to her vice president of talent strategy.
Elena Vasquez, the chief marketing officer at Clearpoint Software as a Service, a one hundred fifty million dollar company, gave it a six out of ten. She was excited by the idea of talking to a synthetic technology chief but felt the article pivoted too quickly back to engineering workflows. She noted that marketing teams do not always have developers available to build these loops. She would only send it if there were a version focused on marketing implementation.
Gregory Haines, a general counsel at an American Lawyer fifty firm, gave it a five out of ten. He noted that the article ignored the legal and ethical implications for law firms, such as attorney-client privilege and the work product doctrine. He felt a managing partner would see no relevance to their business.
Dana Kowalski, the technology chief and co-founder at Threadwork Labs, a Series A startup with thirty employees, gave it a four out of ten. Her team has already been using synthetic users for eight months. She felt the article was for beginners and wanted an advanced playbook on persona drift and version-controlling calibration documents. She would send it to other founders who have not started yet.
Sandra Kim, the chief technology officer at Veridon-Praxis Technologies, a two point one billion dollar merged entity, gave it a seven out of ten. She is managing the merger of two incompatible engineering organizations and felt a shared synthetic reader panel would force alignment on customer definition. She would send this to her integration lead and to the technology chief of the acquired company.
Ingrid Magnusson, the security chief and technology chief at NordFinans Group, a six hundred million dollar fintech company in Stockholm, gave it a seven out of ten. She emphasized that in her world, where data is processed and whether it crosses an EU border is the entire project. She appreciated the acknowledgement of regulated jurisdictions. She would send this to her counterparts at other Nordic firms with a warning about data sovereignty.
James Okonkwo, the chief information officer at the U.S. General Services Administration, gave it a five out of ten. He noted that federal procurement moves much slower than the article assumes. He would need an authority to operate and a Federal Risk and Authorization Management Program authorized model provider. He felt the article understood the acronym but not the actual constraint.
Yuki Tanaka, the chief technology officer at Mitsuhara Industries, a twelve billion dollar manufacturer in Osaka, gave it a six out of ten. He worried about how direct criticism from an agent would be received in a culture that values consensus. He would suggest his digital transformation team explore how to adapt the output for their decision-making process.
Carlos Mendez, the technology chief at VoltaPay, a three hundred million dollar fintech in Brazil, gave it a seven out of ten. He liked the compression of feedback loops for moving targets like regulations. He noted that vision-capable models require significant bandwidth and that he would need personas calibrated for the Latin American market.
Amir Hassan, the technology chief at Tawazun Digital, a sovereign wealth tech initiative in Abu Dhabi, gave it a six out of ten. He felt the article undersold the capability by focusing on business to business sales. At a national scale, he would need hundreds of synthetic users across multiple languages and literacy levels. He would send this to other sovereign digital leads but noted the need for localization.
Lisa Nakamura, the technology chief at Crestline Retail, an eight billion dollar company, gave it a seven out of ten. She saw value in testing checkout flows during busy seasons like Black Friday. She noted that reader personas shift throughout the year and that the article did not address seasonal scaling. She would send this to her vice president of digital experience.
Alex Rivera, the technology chief at Storyarc Media, a two billion dollar entertainment company, gave it a six out of ten. Since AI is their product, he needed synthetic users to evaluate narrative tension and playlist discovery, not just pricing pages. He would not send this to his peers in media.
Sven Eriksson, the technology chief at a fifteen billion dollar Scandinavian telecom, gave it a six out of ten. With fourteen million subscribers, he already has plenty of production telemetry. He saw the most value in using synthetic users for internal network engineering tools, a use case he felt was buried in the text. He would send it to his operations lead.
Henrik Johansson, the technology chief at GrainWise, a four hundred million dollar precision agriculture company, gave it a five out of ten. His users are often in fields with no cell coverage using tablet interfaces in direct sunlight. He needed an approach that accounted for offline-first experiences and non-screen interfaces. He would only forward a version that addressed these agricultural constraints.
Ananya Mehta, the technology chief at Proteon Life Sciences, a six billion dollar pharmaceutical company, gave it a seven out of ten. She liked the structured approach for her regulatory affairs team. However, she noted that calibration data in her field is often protected health information or proprietary clinical data, making compliance a major project. She would send this to her head of research and development informatics.
Richard Townsend, the technology chief at Steelvine Holdings, a three billion dollar publicly traded company, gave it a seven out of ten. Under activist investor pressure, he valued the speed of the feedback loop. He wanted to know how to translate loop compression into the operating leverage metrics that investors track. He would send this to his vice president of product to pilot against their client portal.
Margaret Thornton, a board member for companies in tech, healthcare, and financial services, gave it a six out of ten. She saw significant governance implications. She wanted to know the board's liability exposure when management relies on AI-generated personas instead of real market research. She would mention the concept in board strategy sessions but would not send the article.
Fiona McAllister, the technology chief at Ironstone Resources, a five billion dollar Australian mining company, gave it a five out of ten. She runs technology across remote sites with a fly-in fly-out workforce. She needed examples applied to industrial operations like control rooms or safety reporting. She felt her peers would close the tab because it was too focused on software as a service.
Wei Lin, the technology chief at a four billion dollar super-app platform in Singapore, gave it a six out of ten. He operates across eight markets where personas shift fundamentally between cities like Jakarta and Ho Chi Minh City. He needed an approach for multilingual calibration and cultural context. He would send this to his head of product to adapt the methodology for the Asia-Pacific region.
Nadia Osei, the technology chief at KwikSettle, a fintech company in Lagos, gave it a five out of ten. Her users often use sixty dollar Android devices on two G networks or transact through agents. Her synthetic users would need to evaluate non-browser interfaces. She felt the argument for synthetic users is even stronger in markets without traditional research infrastructure.
Amara Washington, the technology chief at Beacon University Systems, a six hundred million dollar educational technology company, gave it a seven out of ten. She saw immediate use for testing student portals but needed an approach that handled the complexity of faculty governance and academic integrity. She would send this to her academic lead and product vice president.
James Chen, an engineering director at a Fortune Five Hundred retailer, gave it a six out of ten. He found the browser-based validation loop useful for simulating warehouse managers on tablets. He needed advice on how to pitch this to his vice president so it would not sound like just another AI side project. He would send this to other engineering directors.
Sarah Martinez, an engineering manager in enterprise software, gave it a five out of ten. She worried about her team's anxiety regarding AI. She felt the article needed a clear reskilling plan for engineers. However, she found the section on internal synthetic users for organizational change to be the most useful part.
Robert Kim, a vice president of engineering at a large enterprise with over three hundred engineers, gave it an eight out of ten. He found synthetic user panels in the development pipeline to be a concrete, reportable capability for his board mandate. He wanted more evidence of how this scales to hundreds of pages across multiple product lines. He would send this to other vice presidents under similar pressure.
Finally, a staff engineer named Dev gave it a seven out of ten. He liked the technical walkthrough of Playwright and vision models. He appreciated the honest failure rates. He wanted to see the actual calibration document and prompts rather than keeping the implementation behind a consulting engagement. He would send this to other staff engineers evaluating agent frameworks.