NEXA is a free decision framework for emerging technology. It scores a vendor, product or capability gap across six dimensions — capability gap, differentiation, estate overlap, strategic alignment, time to value and reversibility — and returns one of four verdicts: run a proof of concept, scope a pilot, explore later, or park it.

How does NEXA decide the verdict?

Each dimension is scored zero to four against published anchors. The scores are weighted into a composite out of one hundred, which is then compared against a bar set by the size of the commitment being asked for. A small, reversible ask clears at forty-five; a platform-level commitment has to clear seventy-five. The arithmetic is published and reproducible — there is no hidden model.

Does NEXA use AI to make the assessment?

No. Every judgement is yours; NEXA supplies the structure, the scoring and the consistency. That is deliberate. A framework you can audit, defend in a steering committee and reproduce six months later is worth more than an opinion from a model you cannot inspect.

Why does it ask what I already own?

Because the most common and most expensive mistake in technology selection is buying something you already have. Estate overlap is weighted heavily, and an assessment that finds a direct equivalent already in place is parked regardless of how good the new option looks.

Is my data sent anywhere?

No. Every assessment stays in your own browser. There is no account, no server and no telemetry on the content of what you assess. You can export your assessments as CSV or JSON, or share one as a link that encodes the scores in the URL itself.

Can I use this framework in my own organisation?

Yes, with attribution. The dimensions, anchors, weights and verdict bands are published in full so they can be adopted, adapted or argued with. The entire framework — including the composite formula and the ordered verdict rules — is also available as machine-readable JSON at /nexa/data.json.

The framework

How NEXA reaches a verdict

Published in full, including the weights, so it can be adopted, adapted or argued with. There is no model and no hidden arithmetic — two people scoring the same thing the same way will get the same answer, which is the entire point of having a framework.

The six dimensions

Each is scored 0–4 against fixed anchors, then multiplied by its weight. The weighted total is normalised against a maximum of 24.00 to give a composite out of 100.

Capability gap

weight ×1.25

How much of this need is unmet by what you already run?

The cheapest capability is the one you have already bought. Nothing else matters until this is answered honestly.

0: Fully met today — we already do this
1: Largely met; only the edges are missing
2: Partly covered, with real friction remaining
3: Thinly covered — people are working around it
4: Genuinely unmet; nothing we own addresses it

Differentiation

weight ×1.15

How hard would an established vendor find it to replicate this within twelve months?

Most "innovation" is a feature your incumbent will ship next year. Paying for it twice is the most common avoidable mistake.

0: Commodity — several vendors already ship it
1: Incumbents are shipping this within the year
2: A real head start, but not a moat
3: Hard to copy — proprietary data, network or know-how
4: Structurally hard to replicate at all

Estate overlap

weight ×1.1

How much would this duplicate something you already own?

Duplication is paid for twice — in licence and in the integration tax of running two things that do one job.

0: We already run a direct equivalent
1: Close to tools already in the estate
2: Some overlap with what we use
3: Adjacent, with little real duplication
4: No overlap — this is clear ground

Strategic alignment

weight ×1

How closely does this serve a priority you have actually stated?

A technology with no sponsor and no stated priority behind it will stall after the pilot, however good it is.

0: Outside our current focus areas
1: Tangential to anything we have committed to
2: Adjacent to current priorities
3: Supports a stated priority
4: Directly advances a top-three priority

Time to value

weight ×0.85

How quickly can this prove or disprove itself?

Speed to evidence is a risk control. Anything that takes a year to fail teaches you nothing you can afford.

0: Multi-year before any real proof
1: Twelve months or more
2: Six to twelve months
3: Provable within a quarter
4: Provable in weeks, inside one sprint

Reversibility

weight ×0.65

If this turns out to be wrong, how cleanly can you walk away?

Reversible decisions deserve a fast yes. Irreversible ones deserve a slow one. This dimension decides which conversation you are having.

0: Deep lock-in; our data would be hostage
1: Costly and disruptive to unwind
2: Unwindable with meaningful effort
3: Standard formats; genuinely portable
4: Trivially reversible — switch it off and stop

The bar scales with the commitment

A score means nothing on its own. The same case that justifies a fortnight of someone's time does not justify a platform decision. The size of the ask sets the threshold the composite has to clear.

Size of ask	What it means	Must clear
Low	A few weeks of someone’s time. No procurement.	45
Moderate	A funded pilot. One team, one budget line.	55
Significant	Multi-team, contract negotiation, a real line in the plan.	65
Major	Platform-level. Hard to reverse once started.	75

The four verdicts

Clearing the bar gives a proof of concept when the decision is fast and reversible, and a pilot when it is not. Falling short by fifteen or less gives "explore later" — the case is real but something is not yet true. Below that, or where a direct equivalent already exists in the estate, the answer is to park it.

Run a proof of concept: The case clears the bar for what is being asked, and the decision is cheap to reverse. Prove it quickly rather than debate it slowly.; Next step — Agree the single question the proof has to answer, and the date it answers it by.
Scope a pilot: The case is strong, but the commitment is large enough or slow enough that it needs structure — a named sponsor, a scope and an exit.; Next step — Write the exit criteria before the start date. Define what failure looks like, in advance.
Explore later: Real potential, but something is not yet true — the gap is thin, the differentiation is unproven, or the timing is wrong. Keep it in view.; Next step — Record the one condition that would change this, and set a date to look again.
Park it: Either you already own this capability, or it is not differentiated enough to justify the disruption. Saying no here buys attention for something better.; Next step — Log the reasoning so the same vendor does not get re-assessed from scratch in six months.

The override

If estate overlap scores 0 — you already run a direct equivalent — and the capability gap scores 1 or less, the verdict is "park it" regardless of how well everything else scores. No amount of strategic alignment justifies buying a second one.

Starting from a gap instead of a vendor

When there is no vendor yet, the same scores produce a routing decision rather than a purchase decision.

Adopt what you already own: Something in your estate already covers this. The work is adoption and enablement, not procurement.
Extend or build on what you have: You are part of the way there. Extending an existing platform will almost always beat introducing a new one.
Worth scouting the market: Nothing you run genuinely delivers this. This is the case where looking outside is the right answer.

Starting scores from the radar

Where what you are assessing matches a technology on theEmerging Technology & Innovation Radar, NEXA offers starting values for two dimensions. Only two, because only two are properties of the technology itself. The other 4 — capability gap, estate overlap, strategic alignment and reversibility — describe your organisation, not the product, and guessing them would be dishonest. Every suggested value is overridable, and nothing is applied unless you ask for it.

Ring → differentiation

Radar ring	Suggested score	Why
Adopt	1	Proven and widely available, so an incumbent already ships it.
Trial	2	Established enough to pilot, but not yet commodity.
Assess	3	Early enough that few vendors deliver it credibly.
Hold	1	Either commoditised or problematic. Rarely a moat.

Adjusted by momentum: new this edition raises it by one, cooling lowers it by one, both clamped to the 0–4 scale.

Time horizon → time to value

4 — radar horizon “Now”
3 — radar horizon “6-12 mo”
2 — radar horizon “12-18 mo”
1 — radar horizon “2-3 yr”

Anything with no recorded horizon is left mid-scale at 2.

The due-diligence set

Eight domains, ordered by where your own scores say the risk sits. These are the questions that get skipped in an enthusiastic pilot and discovered in an unhappy renewal.

Integration surface: Which of your core systems must it connect to, and are those connectors real today or on a roadmap?
Identity & access: Does it support your identity provider, single sign-on and joiner-mover-leaver process without custom work?
Security posture: What independent assurance exists, and what is the blast radius if this vendor is breached?
Data residency & privacy: Where does your data physically sit, who can access it, and is it used to train anything?
Scale & resilience: Has it run at your volume, for someone comparable, with an SLA that means something?
Data governance & exit: On the day you leave, what exactly do you get back, in what format, and how long does it take?
Supportability & viability: How long is the runway, who answers at 2am, and what happens to you if they are acquired?
Commercial shape: How does the price move as you succeed with it? Model the cost at ten times today’s usage.

Questions to put to the vendor

“What makes your approach genuinely hard for an incumbent to copy within twelve months?”: Separates a durable advantage from a feature that your existing vendor will ship for free next year.
“What would a ninety-day proof have to demonstrate for us to commit real budget?”: A vendor who cannot answer this has not thought about your decision, only about their pipeline.
“Who is your most similar customer, at our scale, and may we speak to them unaccompanied?”: The word "unaccompanied" is the whole question. Reference calls with a seller present are theatre.
“What does your product deliberately not do, and who should not buy it?”: A vendor with no honest answer here will not tell you the truth about anything harder.
“On the day we leave, what do we get back, in what format, and how long does it take?”: Exit terms are cheapest to negotiate before you sign and impossible to negotiate after.
“How does our cost change if usage grows tenfold?”: Turns a per-seat price into the number that will actually appear in a future budget.

Common questions

What is NEXA?: NEXA is a free decision framework for emerging technology. It scores a vendor, product or capability gap across six dimensions — capability gap, differentiation, estate overlap, strategic alignment, time to value and reversibility — and returns one of four verdicts: run a proof of concept, scope a pilot, explore later, or park it.
How does NEXA decide the verdict?: Each dimension is scored zero to four against published anchors. The scores are weighted into a composite out of one hundred, which is then compared against a bar set by the size of the commitment being asked for. A small, reversible ask clears at forty-five; a platform-level commitment has to clear seventy-five. The arithmetic is published and reproducible — there is no hidden model.
Does NEXA use AI to make the assessment?: No. Every judgement is yours; NEXA supplies the structure, the scoring and the consistency. That is deliberate. A framework you can audit, defend in a steering committee and reproduce six months later is worth more than an opinion from a model you cannot inspect.
Why does it ask what I already own?: Because the most common and most expensive mistake in technology selection is buying something you already have. Estate overlap is weighted heavily, and an assessment that finds a direct equivalent already in place is parked regardless of how good the new option looks.
Is my data sent anywhere?: No. Every assessment stays in your own browser. There is no account, no server and no telemetry on the content of what you assess. You can export your assessments as CSV or JSON, or share one as a link that encodes the scores in the URL itself.
Can I use this framework in my own organisation?: Yes, with attribution. The dimensions, anchors, weights and verdict bands are published in full so they can be adopted, adapted or argued with. The entire framework — including the composite formula and the ordered verdict rules — is also available as machine-readable JSON at /nexa/data.json.

Run an assessment Download the framework (JSON)or view it in the browser See the technology radar

The NEXA framework is free to adopt, adapt and cite with attribution to Shailesh Kumar (vantessence.com). It is independent, carries no sponsored placements, and is directional guidance to focus deeper diligence — never a replacement for it.