Mentoring early-career PMs, and why the tools stay free
How I mentor students and early-career PMs and BAs, and why TaskCue, Finance, and Medical Claims stay free.
Outside delivery work I mentor students and people early in their careers as product managers and business analysts. I also build small GenAI products and leave them free. The tools are not a business, and the mentoring is not a pipeline into one. This note is what I actually share, what I will not pretend to offer, and why the three live apps stay unpaid.
I have been at this profession for more than twenty-five years, across telecom, healthcare, retail, analytics, and construction. The titles changed — senior technical project manager, delivery lead, product owner, business analyst — and the work stayed product-shaped: discovery, a roadmap, a cross-functional release, and an outcome someone could check. That is the material I mentor from. I am not interested in teaching a personal brand.
Who I am talking to
Students, and PMs and BAs in their first roles, who are trying to learn how software actually leaves a team. Sometimes they are also trying to understand GenAI tooling without drowning in demos. I am happy to talk. The useful conversations are specific. Bring a backlog you cannot order, a stakeholder you cannot align, a requirement you cannot tell from a solution, or a side project that is growing features and not growing users. I can react to that. I cannot usefully react to "how do I break into tech?" as an abstract question, and I will not role-play a hiring process with you.
I am a delivery lead by day. Since January 2026 that has been Systems Ltd, for a major telecom client: release outcomes, alignment across engineering, QA, BA and product, and the business, and GenAI embedded in the delivery workflow. I do not bring that client's internal detail into mentoring conversations. I bring the judgment, and I point at public work when a concrete example helps.
What I share
Four themes, because they are the ones I have scars in.
Roadmaps and backlogs. A roadmap is a sequence of outcomes, not a catalogue of ideas. Prioritisation is the act of leaving something out and saying so. When someone is stuck between "the stakeholder asked" and "the team can deliver", I use Veroxos: I was dual project manager and product owner, and manual reporting fell by 60 percent because real-time KPIs replaced packs people were still building by hand. A request for another manual report is often a product failure you are about to institutionalise.
Requirements with BA discipline. I certified CBAP years ago, and the part that lasted was not the credential. It was refusing to solution a sentence I had not understood. Early-career BAs are often rewarded for producing artefacts quickly. GenAI makes that reward dangerous, because a plausible specification is now cheap. I ask them to show me the user, the workaround, and the non-goals before they show me a generated draft. If those three are fuzzy, the draft is decoration.
GenAI as a shipping practice. I use Claude Code, and Google AI Studio when the product wants it. I do not teach prompt collections. I teach a loop: write the boundary, implement one slice, review it yourself, ship or delete. The public examples are the free tools below. On client-side product builds I will also mention two published results, without stretching them: Dental EHR, where tighter scope and GenAI-assisted R&D cut QA effort by 35 percent, and BarkCrush, where AI-assisted web and mobile delivery made releases 22 percent faster. I tell people those numbers have owners and contexts. They are not a promise about their own team.
Healthcare and compliance, at the level of delivery judgment. Panacea reached 100 percent Meaningful Use Stage-2, and more than $450,000 in penalties was avoided. I will talk about what that demanded of a plan: the external consequence named up front, scope cut to what operators can sustain, QA on the critical path, no checklist theatre. I will not invent measure-by-measure instructions. Anyone mentoring in a regulated domain owes people that restraint. Confident fiction is worse than a short answer.
Why the tools are free
TaskCue, Circles Finance, and Medical Claims are live, and they are free. TaskCue, at tasks.samikhanapps.com, is shared tasks and reminders without a heavy project tool. I was product owner and builder. The stack is Claude Code, React, Node.js, and MySQL. Circles Finance, at finance.samikhanapps.com, puts billing windows, leave, and a budget ledger in one place so spreadsheets stop drifting. Claude Code, React, and MySQL. Medical Claims, at medical.samikhanapps.com, is a PWA for family medical expenses and a monthly claim PDF, built with Google AI Studio. It is not an EHR.
I built them for my own use and for anyone with the same problem. They are also the cleanest teaching aids I have, because a learner can open them. A case study they cannot click is a story. A URL they can click is a product, with all the small decisions still visible in what was left out. TaskCue does not have a programme-management layer. Medical Claims does not pretend to be a clinical system. Circles Finance covers a billing cycle instead of "finance" in the abstract. Those omissions are the lesson.
I am not pitching, and I do not want these treated as a template for shipping a startup with an assistant. The narrower lesson is: solve a real workaround, keep the scope smaller than your enthusiasm, review every slice, and be accurate in public about what the thing is.
If a login screen on a workplace tool shows Systems Ltd branding, that is because some of my workplace tooling runs in that context. It is not a claim that the three free apps are client products, and it is not an invitation to treat them as one.
How a conversation usually goes
I ask what they are accountable for this month — the release, the requirement, the side project — and what they are about to add that the user did not need. Most early trouble I see is addition. Subtraction is the skill.
If they are using Claude Code or a similar assistant, I ask for the boundary they wrote before the session: user, workaround, non-goals, one slice, and how they will know the slice failed. If they are a BA between a stakeholder and a team, I ask which sentence both sides would sign. An assistant can put two wordings next to each other. It cannot take the meeting where someone has to choose.
The invitation
The portfolio is samikhanapps.com. Start with the three live tools if you want the build side. The community section of the site is the short version of how I like to contribute: practical roadmap and risk conversations, GenAI workflows that end in production, and healthcare lessons that stay inside what I can actually defend.
If you are a student or you are early in PM or BA work and you want to talk, find me on LinkedIn. Bring the specific problem. I will answer in the same register I use at work: concrete, limited to what I know, and uninterested in hype.