There are three reasons I wanted to meet. The first is a personal connection through cognition and the experience of building something beyond a conventional career path. The second is practical: I would like to understand whether there is a digital initiative or recurring workflow where engineering support would help. The third is the possibility of being useful to people in your wider network over time. Those goals do not require us to decide on a large partnership today. I suggest we spend a few minutes on context, give most of the time to a real example from your work, and finish with a concrete next step if one emerges.

The first connection is an interest in people’s experience: what draws their attention, what they find difficult, and what helps them learn. My five years in business development at Emotiv gave me a commercial perspective on EEG applications. Your work brings academic research and a practice of learning through small experiments. Building CTC added a third perspective for me: turning an idea into something a person can actually use. Those are different contributions. I would like to use the next few examples to make the connection concrete, then ask which questions are most relevant to what you are doing now.

My five years as Business Development Manager for Emotiv in Vietnam gave me a commercial lens on how people actually attend and react. Your pilot at King’s College London — with 36 neurotypical and 36 neurodivergent students across ADHD, autism, and dyslexia — really stood out to me. Even where EEG and eye-tracking signals differed, learning outcomes held steady, pointing to compensatory strategies and the power of inclusive design. Did that shape how you think about designing digital learning environments today?

I see this timeline as context for a conversation about choices. Your public work spans Google Fit, the Lysa project, writing and building Ness Labs, and academic research at KCL. My own starting point was business development at Emotiv before building CTC. I would not claim that our experiences or turning points were the same. What interests me is the practical question beneath both journeys: when a project teaches us something unexpected, how do we decide whether to continue, change direction, or stop? I can share one short example from a project I actually worked on, and would enjoy hearing which experiment changed the way you build.

These three threads help us ask better product questions. The hypercuriosity paper is a theoretical proposal, so I would not present it as established evolutionary fact. The later survey of five hundred and twenty-one adults reports associations between hyperactivity–impulsivity traits and self-reported curiosity, without establishing causation. The 2026 study on neurodiversity and generative AI in higher education adds qualitative accounts of support, authenticity, and dependence: twenty-four neurodivergent and neurotypical students used Microsoft Copilot in tasks, and fourteen joined focus groups. For a prototype, we can ask whether users understand the output, whether they can check its basis, and whether they retain control of the next action. A small usability study would answer those practical questions within its scope.

This map shows the ecosystem around a potential collaboration. On the Ness Labs side are distinct activities: newsletters and essays, Hypercurious, community and Fellowship, Featured Tools, and Anne-Laure’s academic work. The published audience numbers are context, not a forecast of users for a new product. In the middle is the proposed place to work together: one selected experiment with a named owner. On the CTC side are AI Studio for commercial automation, AI Auto for industrial IoT and AI, and PowerIT for engineering and secure delivery. The arrows represent routes we could explore. They do not imply that the systems are connected today or that a university collaboration has been agreed. Which part of this map contains work that matters now?

Let me introduce CTC through the work we want to make possible in Vietnam. Since 2025, our positioning has been to accompany the era of ideas: help a person or business move from a promising idea to a product they can use and operate. My earlier Emotiv business-development experience supplies context for EEG applications. For the Vietnam campaign and retail work, I handled client acquisition, product and proposal development, measurement experiments, data collection and analysis. The software showcases later make another part of that capability visible. What connects them is understanding the task before choosing the technology.

Since 2025, CTC Smart Tech has focused on partnering with lean organizations in the idea era: turning promising AI drafts into calm, dependable software. We always start from the recurring workflow that drains your team’s hours, integrate cleanly with the stack you already trust, and build in human review gates and clear handover so you stay in full control.

And third is our co-founder experience delivering the interactive EEG campaign for Audi Q8 in Vietnam alongside five live retail settings. Those projects forged our habit of combining refined user-facing design with the operational reliability required when real customers step into an experience every day.

A guided experience offers a useful pattern for a community experiment. Before the activity, the participant names a question and says what they expect. During it, the task remains simple and observations are easy to capture. After it, a facilitator can use Plus–Minus–Next to explore what helped, what was difficult, and what the person wants to change. The tool can make that sequence easier to follow and help preserve the record. It should not convert the participant’s experience into a label that replaces their own interpretation. This gives us something concrete to discuss with a facilitator: where does the existing activity need support, if anywhere?

These five Vietnam projects show how I adapted the work across different retail settings, from finding the client and developing the product to running measurements and analysing results. The interaction changes when someone encounters a fragrance, listens to sound, tries a drink, explores a shoe, or touches a material. The footage helps us look at the activity and the guidance around it. We do not need to infer research accuracy or business impact from the display on a screen. The parallel for Ness Labs is that different roles need different support. A member trying an experiment, a facilitator leading a reflection, and an operator reviewing an incoming proposal should not be forced through the same interaction simply because the underlying technology is shared.

The practical bridge comes from my own delivery sequence: find the client and understand the question, develop the product and proposal, run a measurement experiment, then collect and analyse the results. The diagram separates that completed experience from the new adaptation on the right. For Ness Labs, we could apply that discipline to a short learning activity: state a question, act, record an observation, and reflect. This is our design inference. Neither the campaign nor the small pilot proves the value of a future Ness Labs product. We would observe the intended users and include the effort of reviewing any AI output in the evaluation.

The three circles describe a working relationship we could test. Research contributes disciplined questions and attention to the limits of a finding. An experimental mindset contributes a willingness to try something small, observe it, and change course. Engineering contributes a functioning workflow, reliable handoffs, and a practical understanding of what maintaining the tool will involve. The centre only becomes meaningful when those perspectives meet around an actual task. It does not follow from shared vocabulary alone. I would like us to identify the part that your team already does well and the part where outside help would add something specific. Then we can judge the fit through a small piece of work.

The examples become useful when they help us identify a real task. Your Omnipotence essay gives two concrete starting points: a community app idea and a team dashboard. I would first ask what has changed since then and which work still matters. Then we can explore three possible forms of support: a clearer internal handoff, a companion to an existing community activity, or an AI-assisted task whose sources and output can be reviewed. CTC brings three service lines, and PowerIT can take an existing AI draft through assessment, targeted fixes, retesting and operational handover. We would agree the release criteria before calling it ready. We can choose the smallest contribution that fits the current need.

In your February 2026 essay on The Omnipotence Dilemma, you wrote about experimenting with a community app idea and an internal team dashboard — and realizing that when AI lets us build anything in days, the bottleneck becomes curating what deserves to exist and maintaining it without burning out a three-person team. Before I show how we approach those exact two ideas, let me briefly introduce CTC so you have the context.

We organize our work around three pillars, with security built across all three. AI Studio crafts business and knowledge workflows where humans approve key decisions. AI Auto connects operational and industrial data streams into live dashboards. And PowerIT handles engineering delivery, security audits, and production readiness. For a lean knowledge team like Ness Labs, AI Studio and PowerIT work hand in hand.

With AI Studio, instead of building generic chatbots, we ground every workflow or companion strictly in your approved articles and archives. A person reviews important outputs before anything goes out, and we measure success by comparing preparation time, review effort, and operating cost against how your team works today.

AI Auto is our operational and industrial integration line — pulling data from systems, logs, or factory sensors into traceable dashboards and alerts. While a Ness Labs pilot obviously doesn't need industrial sensors, the exact same discipline of unifying scattered signals into one calm view applies directly to a three-person team.

And PowerIT is how we carry an AI prototype safely across the production finish line. We review system architecture, lock down permissions, cap API spend, verify fixes, and hand over clean documentation so your team can run or pause the tool anytime.

Whenever a small team considers opening a tool like that to thousands of readers, a natural question arises: is it safe? Verizon’s 2026 Data Breach Investigations Report shows that 48% of breaches involved a third party — up from 30% — and known-exploited flaws took a median of 43 days to patch. For a lean organization, security lives in the everyday tools, plugins, and access keys you rely on.

In the AI era, we see three distinct risk layers: attackers using AI to accelerate exploit discovery — like the first AI-assisted zero-day Google GTIG documented in May 2026; AI-written code shipping faster than human verification; and internal AI tools with broad permissions becoming bridges between systems, as highlighted in Binance’s initial November 2026 report. That is why putting an unreviewed prototype into production is an unmeasured reputation risk.

Because we have never looked inside your stack, this is never a diagnosis. You do not need bank-grade security — just calm answers to four practical questions: How do readers verify official messages? Who holds backup keys to core accounts? How is member data isolated? And what human gate triggers before an AI tool acts? Your platforms may already cover most of these well.

For any gap that remains, each question maps to one small step where you keep all passwords and final control. The first two steps can be done in a 45-minute guided screen-share session where you stay at the keyboard. And while packaging this as a standalone client audit service is a new expansion for CTC, it is built directly from the engineering discipline we apply to our own products.

When an AI prototype is close to serving real members, our PowerIT flow moves through four clear steps: Assess, Fix, Retest, and Handover. A fast demo proves a workflow can run when everything goes right; production readiness proves you stay in control when something goes wrong — backed by a 10-item handover dossier and tested recovery steps.

This diagram summarizes the six go-live readiness gates and 14-item checklist we verify before recommending any AI tool for live community use: strict permission boundaries, isolated secrets, spend limits, human approval gates, audit logs, and a tested kill-switch and recovery playbook.

Now let’s bring this directly to the ideas from your February essay. Hypothesis 1 is a calm 'Team Pulse' dashboard. Instead of a complex reporting suite, we start with one weekly handoff: connecting read-only to your status signals to generate a Monday morning brief with source links, open decisions, and recurring reader questions. Your operations lead can immediately judge whether it saves real time every week.

Hypothesis 2 is a lightweight Community Companion alongside an existing cohort or Fellowship activity. A participant frames one small action, records what happened, and reflects using Plus–Minus–Next — with AI surfacing relevant Ness Labs essays only when invited. Facilitators see only what members explicitly choose to share, keeping human warmth at the center without adding moderation hours.

For evidence-grounded AI, a useful first task might be comparing a small set of approved documents. The output would be a draft table whose claims link to the relevant passages, with missing information and uncertainty kept visible. A human reviewer remains responsible for the interpretation and final use. We would evaluate whether the tool makes that review easier and more accurate, including the time spent finding and correcting mistakes. Your writing about the illusion of clarity and your 2026 study on neurodiversity and generative AI make that question especially relevant. We should not assume that adding citations makes an answer correct, or that a small prototype can perform an entire academic review.

This is the design principle I would like to explore across the examples. Before showing an AI-generated explanation, we might invite the person to make a brief prediction or explain the idea in their own words. Then the interface can make sources and uncertainty easy to inspect, followed by a deliberate choice about the next action. That sequence draws a practical question from your essay about the illusion of clarity. It is not a validated intervention simply because the theory is appealing. We would observe whether people understand the task and can judge the result, and whether the extra step earns its place or merely adds friction.

The first two examples show work in Vietnam: an application CTC states is in operation, and a reporting pilot. The security checks just discussed describe a service to scope for each product. TinyExp is a separate, independent prototype for exploring a Ness Labs learning journey. It gives us an artifact to discuss: our independent attempt to translate a reading of Tiny Experiments into an interaction. The next question is how someone would move through that experience and what we would learn by observing them. We can make the intended journey explicit, connect it to the wider learning loop, and define a small delivery and evaluation process. That is what turns a prototype from an interesting demonstration into a useful experiment. The review should reveal both what worked and what needs to change, and leave the owner able to decide whether the work deserves further investment.

Let me briefly show three real examples from Vietnam that ground our work. First is ECT AI Consumer Research — a bilingual platform architected and commercially operated by CTC. It guides participants through sensory tasks and gives researchers a clean, role-based analytics workspace. We can walk through any permitted flow together if useful.

Second is PA Pilot — a controlled five-person reporting pilot comparing Insight and MN8 device streams across two videos. Every summary card stays linked to its exact timestamp while participant identities remain completely anonymized. That same principle of keeping every synthesis anchored to its source context powers how we design community tools.

To make that tangible, we built TinyExp — an independent prototype inspired by Tiny Experiments. It guides a reader through framing a repeatable PACT in 60 seconds, logging curiosity-driven check-ins, and reviewing with Plus–Minus–Next to decide whether to persist, pivot, or pause. We built it as a conversation piece so we can critique together what genuinely helps a learner reflect.

This is the journey I would like to discuss with a participant or facilitator. First, name a small question and what you expect to happen. Then try the action and record a brief observation. Plus–Minus–Next creates a moment to describe what helped, what was difficult, and what you want to try next. The final choice may be to continue, change the approach, or pause. A tool can preserve those steps and make them easy to revisit. It can also support sharing, but only the part the participant chooses to share. This is a proposed interaction to test, so we should ask where it feels useful and where it adds unnecessary work.

The loop brings the earlier pieces together. We begin with a useful signal from a conversation, an essay, or a recurring task. We shape an idea, using AI where it helps, and select a bounded test. Engineering makes that test usable. People try it in the intended context, and a Plus–Minus–Next review helps us decide what to retain, change, or stop. Only then do we consider sharing the result more widely. This is a proposed collaboration process, not a claim that such a system already operates across Ness Labs and CTC. The loop is healthy when it creates a useful decision, including a decision to stop adding software.

PowerIT’s proposed contribution is the work between an interesting draft and a workflow someone can inspect and use. We can begin with notes, a short walkthrough, or code that you have permission to share. First we check what exists and whether it is useful to retain. Then we agree on one user journey, the dependencies, and the acceptance checks before implementing it. Verification should cover the actual task and the relevant failure cases, rather than relying on a polished demo alone. Finally, we document what was built, what remains limited, and who owns operation and support. The depth of each step depends on whether we are delivering a prototype or preparing a production release.

So how do we work without draining a founder’s bandwidth? True to your book, we use a 14-day scoped Tiny Experiment. It takes just 30 minutes of your time in Week 1 to align on one bottleneck; we build in an isolated sandbox using synthetic data, run our security and quality checks in Week 2, and review together on Day 14 so you can consciously choose to persist, pivot, or pause.

We compare the prototype against your current manual process across three clear dimensions: time saved in preparation and review, source-traceable accuracy, and whether the team owner feels the benefit clearly outweighs ongoing upkeep.

If one of the ideas is useful, the next step is to make it clear enough for both sides to evaluate. We should name the task, the owner, the scope, and what a good result would look like. Commercial terms and the responsibilities after delivery belong in that discussion as well. There may also be relevant opportunities in the wider network over time, but the first contribution should stand on its own. I suggest we capture the brief in four fields, agree what information is needed next, and choose a review time that works for the people involved. That leaves us with a practical decision rather than a vague intention.

The third goal is to be useful beyond a single project, if the fit develops. Joe’s partnerships role and the Featured Tools interviews suggest a context where founders discuss how their products work. Examples such as Flowing, Littlebird, and Linkflare give us topics to understand, not an assumed referral channel. Fellows who are building something might also have a specific technical question. Separately, researchers may need software for a defined task, but any academic work would follow its institutional, ethics, data, and procurement process. There is no need to arrange introductions today. A relevant connection should arise from a clear need and the willingness of the people involved.

I also want the commercial path to be clear. A scoped pilot should specify the outputs, timing, price, ownership, acceptance, and handover before either side commits resources. If the result is useful and there is continuing work, we can then discuss the most suitable arrangement: a fixed-scope project, an engineering pod with clear responsibilities, or a retainer for ongoing support. Co-building a product would need a separate business case and agreement. Academic procurement would follow the relevant institution’s process. We would choose any ongoing model together after reviewing the pilot and the actual work ahead. The question for today is which kind of engagement would fit how your team normally works with external partners.

That brings us to the most important part of today — listening to you. Looking at these four open questions: across readers, members, and your three-person team, where does recurring friction still live, what already works well, and what would genuinely protect your bandwidth right now?

Whatever we uncover today, there is zero timeline pressure. We can simply send you a one-page recap note with the four-question security self-check (Option A), schedule a 45-minute guided screen-share hygiene session where you keep all passwords (Option B), or bookmark a 14-day sandbox experiment for whenever your bandwidth opens up (Option C). Thank you again for a wonderful conversation, Anne-Laure.