subinternal Talk to us

ResearchDecision AI

Growth Protocol vs subinternal

Both are reasoning platforms that keep the model from setting the answer on its own. They differ in who builds on them, what they reason over, and whether your data reaches a model provider at all.

QuestionGrowth Protocolsubinternal
What it isIn its own wordsA reasoning platformLicensed, and offered through EYA reasoning platform, and its buildersBSI built for you on our engine
What decides the answerWhere the model stopsRules and constraintsSays business rules set the valid routeCode, from your recordsThe model explains, never sets a figure
Questions it is built forFrom its own examplesStrategic growthNew categories, regional growth targetsWhy a number movedSales, finance, customers, operations
Where it runsWhose systems hold itIts Google Cloud stackGemini, with Elastic for searchYour cloud or your buildingOr with no internet at all
What a model provider seesWhere the data goesKept by none, it saysZero retention, by deployment designNothing, in private AIFrontier models off until switched on
Audit trailWhat is keptEvidence and rationaleSays they travel with every decisionEvery question and sourceEach forecast scored against what happened
Security reportAs each states itSOC 2 Type 2Report on requestSOC 2 compliantReport on request
Customers shownOn its own pages18 logosWith results it reports itselfNone namedResearch published with sources
PricePublishedNot publicIn writing after a first call
Growth Protocol’s column is what it says on its own pages and its partners’, read in November 2026 and cited below. subinternal’s is from our own pages. We are subinternal, so read this as one side’s comparison, made with the other side’s own words.

How do Growth Protocol and subinternal differ?

Growth Protocol sells a reasoning platform. It calls itself “the Enterprise Reasoning Platform”, for “decisions you can trust, explain, and audit”, and EY licenses it exclusively for its EY Growth Platforms practice.1, 4 subinternal has a reasoning platform of its own, the engine under Business Super Intelligence (BSI), and is also the team that builds on it. BSI joins every system a company runs and answers why a number moved and what comes next, and we build it for each company on that engine. Around it we set up the rest of a company’s AI, and build agents and search on its own systems.8, 10

So the first question is not which is smarter. It is whether you want to license a platform for growth strategy, or have a reasoning platform built onto your own systems, by the people who run it, where your data already lives.

What decides the answer?

In both, something other than the model. Growth Protocol says its deterministic reasoning “combines model output, agent actions, learned signals, symbolic constraints, business rules” and that “ontology, policies, thresholds, and commercial logic determine the valid route.” It calls this neurosymbolic AI.1, 2

In BSI, numbers are computed by code from the records, and the model explains them and never invents one. Revenue, pipeline and churn are written as code in one place, so every answer uses the same definitions, and anything that changes a record or reaches a customer waits for a person’s yes.8 The label differs; the principle is the same one: a model gives opinions, and code decides.

What questions is each built for?

Growth Protocol’s examples are strategic. Elastic’s account of it cites questions such as moving from personal care into pet care, or raising regional sales by 25%, and Growth Protocol reports analyzing more than 300 companies in days for a consumer goods client. It names insurance, industrials and consumer goods as its industries.5, 1

BSI’s are the questions a company asks of its own records: why margin moved, which deals are at risk, which customers are likely to leave, why orders ran late. Every answer is traced to the records behind it.8 A strategic decision rests on that internal record too, and BSI builds that part first.

Where does your data go?

Growth Protocol runs on Google Cloud with Gemini and Elastic.5 Its security page says “no client data retained by model providers”, and also that zero data retention patterns are “available by deployment design”, so it is worth asking which design yours would get.3 Zero data retention is a promise about what the provider keeps. The request still goes to the provider, and Google’s own documentation lists features, such as grounding with Google Search, that keep data however you set them.7

BSI runs in your own cloud account or on hardware inside your network.8 With private AI, nothing goes to a model provider, and it can run air gapped with no internet connection at all. Frontier models stay off until you switch them on, and even then they see no names.9 We wrote up the difference between zero retention and data that never leaves on its own.

What can each one show?

Growth Protocol shows more. It states a SOC 2 Type 2 report, shows 18 customer logos including P&G, Unilever and Nestlé, and reports “$170M+ profitable growth delivered” and “20X ROI”, without saying how those were measured.3, 1 EY confirms the license in its own release.4 Its Databricks partnership was announced in a paid press release.6

subinternal states that it is SOC 2 compliant, with the report on request, and names no customers on its pages.9 What we publish instead is our research, with every source cited. Neither of us has an independent evaluation that we could find. Ask both for references and for the report.

Which one fits?

This is our judgment, and we are one of the two. If you are an EY-Parthenon client working a growth question across markets and categories, Growth Protocol through EY is built for exactly that. If you want your own systems joined, every number traced to its record, AI set up and built across the business, and the choice of running it where no model provider sees your data, that is what we do.

The two are not always rivals. A strategy is only as good as the record under it, and many companies need that record joined and defined once before any strategic engine can be trusted with it.

What can’t this comparison tell you?

How either would do on your data. Every Growth Protocol fact here is what it, EY or a partner says, read in November 2026; we have not tested its platform, and no one we found has published a test of either of us. Products and partnerships change. Ask each for a working session on one of your own questions, and judge the answer and its trail.

References

  1. Growth Protocol, “The Enterprise Reasoning Platform.” Read November 2026. growthprotocol.ai
  2. Growth Protocol, “Why neurosymbolic AI.” Read November 2026. growthprotocol.ai
  3. Growth Protocol, “Security.” Read November 2026. growthprotocol.ai
  4. EY, “EY-Parthenon practice unveils neurosymbolic AI capabilities.” September 2025. ey.com
  5. Elastic, “Growth Protocol” customer story. Read November 2026. elastic.co
  6. Growth Protocol, “Growth Protocol partners with Databricks,” a paid press release. March 2026. finance.yahoo.com
  7. Google Cloud, “Vertex AI and zero data retention.” Read November 2026. docs.cloud.google.com
  8. subinternal, “Business Super Intelligence.” November 2026. subinternal.com
  9. subinternal, “Private AI on your own hardware.” October 2026. subinternal.com
  10. subinternal, “Enterprise AI setup, implementation and coaching.” October 2026. subinternal.com