A big client wants a feature that guts your product

A note from the archive: this post, from August 2026, follows the earlier KarmaLens method, which scored every reading on eight lenses from −5 to +5. Today a reading asks the decisive facts first and makes one call; the lenses below are a way of looking at the decision, not a score. See how a reading works.

The offer lands on a Tuesday. Eight months of runway, one signature, and the payroll email you've been dreading un-writes itself. Then you read the scope: they want a build that quietly breaks the one thing your product was actually for. And you hear yourself say the sentence founders say. We'll refactor it back later.

The Dilemma

This is a composite — stitched from the kind of contract-vs-purpose fork we see often in the founder category, not one real company.

Maya bootstrapped a scheduling tool for solo clinicians. The whole point is that it does one thing without configuration — a solo practitioner opens it and it just works. Seven weeks of runway left. Then a regional chain, forty locations, wants in. The number is real: eight months of runway, paid quarterly.

The catch is in clause four. They want role hierarchies, admin overrides, a custom approval flow, and a bulk-import their ops team designed. None of it fits the solo user. Worse, the admin layer forces a login gate on the front door — the exact friction the product was built to remove. And clause nine locks the roadmap to their change requests for twelve months.

Maya's engineers can build it. That's not the question. The question is what the product becomes while they do. Every sprint for a year answers to one client's ops team. The solo clinicians — the people the thing was for — get a login wall and a backlog that never clears. Maya tells her co-founder they'll take the money, harden the base, and rip the enterprise layer out in year two. She has said this before, about a smaller contract, and year two never arrived. The ugly detail she hasn't said out loud: part of her wants the contract because building for one demanding client is less frightening than finding out whether solo clinicians will actually pay.

The Read

Four of the eight lenses bite hard here. Watch two of them go to war.

Duty (svadharma): +1

Duty asks what work is genuinely yours. Maya's dharma is the solo clinician who can't afford an ops team — that's the person the product exists to carry. The contract is real work, and shipping payroll is a duty too; a founder who lets the company die on principle has served no one. But this duty is split down the middle. The build serves survival and betrays the mandate at the same time. And the enterprise layer isn't the mandate — it's another company's roadmap wearing Maya's logo. Score stays low and positive: the money is a duty, the build is not hers.

Wider welfare: -1

This lens counts everyone the decision touches. The enterprise client benefits — genuinely, they get a tool shaped to them. The team keeps their jobs for eight months. But the solo clinicians, the entire base the product promised to serve, get a login gate and a frozen roadmap. The people with the least leverage lose the thing built for them. Welfare doesn't just count the loud client; it counts the quiet many who never signed a clause.

Detachment: -3

Detachment asks whether you can act without clutching the outcome. Here it flags the real driver. Maya isn't reaching for the contract because it's right — she's reaching because the alternative is finding out whether her actual bet works, and that's terrifying. The offer is a way to not look. "We'll refactor it back later" is the story detachment sees through: it's the sentence that lets you say yes to the fork without admitting it's a fork. This is the lowest score, and it's not close.

Discernment: -2

Discernment separates action from inaction dressed as action, and real risk from imagined risk. It sees that taking the money feels like the responsible, non-action move — but it's the bigger bet, not the safer one. Twelve months of locked roadmap converts a product company into a contract shop that happens to own IP. The imagined fear is "we run out of runway." The real, under-weighted risk is "we spend a year building the wrong thing and emerge with no product and no base." Saying yes to the wrong revenue can cost more than the runway it buys.

The tension: Duty (+1) wants Maya to keep the lights on — you can't serve solo clinicians from a dead company. Detachment (-3) says the lights are the excuse, not the reason. That's the fork. One lens is protecting the mission; the other is exposing the flinch wearing the mission's clothes.

Verdict: Take a scoped version that serves the base or walk — but the year-long roadmap lock as written trades your product for its own funeral costs, and "refactor it back later" is the fork pretending to be a bridge.

Here's what a general AI chatbot does with this: lean toward the contract and it will help you defend taking it; say "I think this betrays my users" and the same model will help you defend walking away — sycophancy is a documented behavior, and an agreeable assistant is built to match your posture, not hold one against it. KarmaLens runs the opposite way. Before making the call, it asks the decisive unknowns first — the questions that actually determine the answer — and the reading that comes back doesn't renegotiate itself when you push back. Where the eight lenses of the Gita bear on the decision, the reading uses them as instruments to show the weighing honestly; where a verse genuinely fits, it's included, not paraphrased to flatter your lean. If the facts don't support a call, you get an honest abstention rather than a verdict dressed up to match what you wanted to hear. You don't get a mirror. You get a call.

On the split-duty read, the point 3.35 keeps making is the one Maya keeps talking past — the mandate:

The Takeaway

You don't need us to run the test that matters most here. Call it the refactor-clock test.

Write the reassuring sentence you're telling yourself out loud: "We'll take it and undo the damage later." Now do three things.

1. Put a date on "later." Not "year two" — an actual quarter. If you can't name one, the refactor isn't a plan, it's a permission slip.
2. Check whether you've said this exact sentence about a past yes. If you have, and "later" never came, that's your prior — weight it.
3. Ask the one question that cuts through the money: does this revenue extend the thing I'm building, or replace it? Extend means the base gets more of what it came for. Replace means a different company grows inside your logo.

Then weigh both paths — take-it and decline-and-stay-lean — through duty and detachment, and see which path only looks safe because it lets you avoid a scarier question. The runway buys time. It's worth checking what you're spending the time on.

Want the full eight-lens read on your own version of clause nine? Bring the contract to KarmaLens for a committed call. If you'd rather watch a few land first, browse the gallery.

The money extends the thing you're building, or it replaces it. Which sentence is true when you say it slowly?