You open the analytics dashboard and see a name you recognize. Your senior engineer just registered a domain that describes, almost word for word, the second half of your roadmap. He didn't hide it well. And the first thing that moves in you isn't strategy — it's the specific heat of feeling betrayed by someone you promoted.
That heat is about to write three emails. Only one of them should get sent.
The dilemma
This is a composite — stitched from the kind of founder dilemmas we see in this category, not one real person's story.
Ravi runs a 14-person B2B tooling startup, Series A, eight months of runway. His strongest engineer — call him Neel — owns the pipeline that half the product depends on. Three days ago Ravi found Neel's weekend project: a landing page, a waitlist, and a one-line pitch aimed at exactly Ravi's customers. Not a clone. No stolen code. No breached NDA — Ravi checked, twice. Adjacent, not identical. But clearly the same water hole.
Here is the ugly detail Ravi hasn't said out loud. Neel asked for a raise four months ago and Ravi stalled him — cash was tight, and honestly, Ravi assumed Neel had nowhere better to go. That assumption is now a landing page.
Ravi has three moves on the table. One: fire Neel Monday, lawyer up, freeze his equity, protect the moat before it widens. Two: say nothing, quietly watch the commits and the browser history, gather a file. Three: the conversation — put the landing page on the table between them and ask what it is. Ravi keeps drafting move one at 1am. He keeps not scheduling move three. That gap is the whole tell.
The read
Four of the eight lenses bite hard here. Watch two of them fight.
Duty — protect what is genuinely yours to protect
Svadharma isn't a vibe; it's a job description. Ravi's is the fiduciary weight of 13 other people's paychecks and a cap table he owes clean stewardship. If Neel's project is quietly poaching Ravi's customer list, his methods, his unreleased roadmap, then guarding the moat is not aggression — it is the founder doing the founder's actual work. This lens looks at the pre-emptive firing and does not flinch from it in principle. A load-bearing engineer building a direct competitor on your dime is a real exposure. Duty: +3 for protecting the company — but only for the part of the moat that is truly under threat, not for the part that is Ravi's ego under threat.
Discernment — can you tell the exposure from the insult?
This is the lens that does the surgery. Discernment separates "what is to be done" from "what merely feels like it must be done." Here it finds two different objects fused into one hot feeling. Object A: a business risk (a key dependency is distracted and possibly competing). Object B: a wound ("I promoted him and he did this to me"). The pre-emptive firing treats both as the same problem and swings one blade at both. Discernment asks the question Ravi is avoiding: do you actually know what the landing page is? A side project he'd have told you about if you'd said yes to the raise? A hedge because he can read the runway too? Or a genuine defection? Ravi has a file's worth of suspicion and zero minutes of conversation. Discernment: -2 on the current plan, because acting on a story you haven't checked is not decisiveness — it's avoidance wearing the costume of speed.
Detachment — is the wrong thing steering the wheel?
The point 2.62 keeps making: dwelling on the insult grows attachment to it, attachment grows the craving to make it right, and the craving manufactures the anger that then feels like clarity. Ravi's attachment isn't to the company. It's to being wronged — to the version of Monday where Neel feels the loss Ravi felt. That's the tell in the drafts he writes at 1am and the talk he won't schedule. Detachment: -3. When your plan's success secretly depends on the other person's face falling, motive is driving and duty is just riding along.
Non-harm — the surveillance move, examined
Move two feels like the moderate one. It is the opposite. Watching commits and building a case turns a colleague into a surveillance target while you smile at standup — and it does its damage in two directions at once. Outward: it corrodes the exact trust that makes Neel worth keeping, because a person you've been quietly filing on can feel the temperature drop even before he knows why, and the working relationship starts dying weeks before any decision. Inward: it commits Ravi to a small lie every morning he runs a normal one-on-one while holding a folder Neel can't see, and each of those mornings hardens the founder into someone who manages by dossier. The surveillance also poisons the thing it was meant to protect — if this ever does become a separation, a paper trail built in bad faith reads as bad faith. Non-harm: -2, and it fails honesty on the way out the door.
The tension, named: Duty says guard the moat. Detachment says the thing you most want to guard right now is your pride, and pride is a terrible fiduciary. They don't cancel. They point at a fourth door neither the firing nor the file will open.
Verdict: Don't fire and don't surveil — put the landing page on the table, ask Neel directly what it is, and let his answer, not your wound, decide whether this is a raise conversation or a separation.
Here's what a general AI chatbot does with this same input. Lean toward firing and it will hand you a crisp case for firing. Type the next message leaning toward mercy and it will pivot, warmly, to the case for the talk — with no memory of having just argued the opposite. It optimizes for your agreement; sycophancy is a documented behavior in these systems, not a cheap shot. KarmaLens runs the fixed eight lenses and aggregates their scores deterministically into one committed call, and that call does not renegotiate itself when you push back — you can dislike the -3 on detachment, but you cannot talk it up to a +1 by rephrasing your grievance more sympathetically. When a verse is cited, it arrives verbatim from the corpus, not softened to match the side you were leaning.
The takeaway you can use tonight
You don't need us for the move underneath this one. Try the two-object split.
Take the action you're leaning toward — the firing, the file, the furious email. Write it as one verb. Then write two separate sentences under it:
- The exposure: the concrete business risk, stated without a single adjective about the other person. ("A key dependency is distracted and may be competing for my customers.")
- The wound: the sentence that starts with I and ends in a feeling. ("I feel like a fool for trusting him.")
Now the test: does your leaning move solve the exposure sentence, or does it only satisfy the wound sentence? The pre-emptive firing barely touches the exposure — the roadmap is already out of his head — but it lands hard on the wound. That mismatch is your answer. A move that only pays down the wound is your resentment doing strategy's job.
Then ask the cleaner question: what would you do about the exposure alone, if you didn't feel wronged at all? That move is usually the conversation you've been avoiding — the one that costs pride instead of trust.
ध्यायतो विषयान्पुंसः सङ्गस्तेषूपजायते। सङ्गात् संजायते कामः कामात्क्रोधोऽभिजायते।।2.62।।
dhyāyato viṣhayān puṁsaḥ saṅgas teṣhūpajāyate saṅgāt sañjāyate kāmaḥ kāmāt krodho ’bhijāyate
When one thinks of objects, attachment to them arises; from attachment, desire is born; from desire, anger arises.
When your best engineer moonlights on your customers, the first thing to protect isn't the moat — it's your own judgment from the anger pretending to be it. Bring your version to KarmaLens for a committed read on the eight lenses, or walk the gallery of decisions already scored. Which of your two objects is the move you're leaning toward actually paying off?
References
- Bhagavad Gita 2.62 — English translation by Swami Sivananda, via BhagavadGita.io.
Your turn
Bring your own dilemma to the eight lenses.
One committed reading, scored on eight fixed lenses — free, no account. Your words stay private; they're never published.