Head-to-head · updated 13 September 2026

DataShield vs Palantir Foundry: an operating system, or a governed data plane?

Palantir itself calls Foundry "The Ontology-Powered Operating System for the Modern Enterprise". For once, the marketing is really literal, too. The Foundry Ontology models your objects. Plus your actions, and your business logic. Then AIP goes and puts agents right on top. Airbus runs Skywise on it. Cleveland Clinic runs bed assignment on it. Do you want the work itself to run inside the software? Fine, then. Nothing here will talk you out of Foundry. We won't even try to, really.

Only one layer is ours to do. We sell it differently, too. DataShield is a checked data plane. It's built for agents. We classify the fields. We tokenize the data. Calls get an authorization check. We keep proof of it. It runs self-hosted. On your own hardware. No forward-deployed team, ever. A price you can read, right on the website. One naming note, before we start. It trips everyone up, each time. Palantir's Ontology is the object and action model inside Foundry. DataShield Ontology is our data platform. Same word, but another product. We'll flag both, each time.

Get free Demo Center access Get an instant quote

The short version

Pick DataShield when

Someone will eventually ask you to prove an agent's access log was not edited. An examiner, an auditor, or Article 12 of the EU AI Act. Our chain simply answers with math, always. Run the verifier.
You need to pull an agent's mid-session access. The very next tool call should fail. Not the next token refresh. The next call.
The data plane has to run on hardware you own. Your own team installs it. No vendor engineers in the building, ever.
You would just like to see a price, before you book a call. Ours is published.

Pick Palantir Foundry when

You want the operation to run inside the software: pipelines, apps, scheduling, digital twins, process mining. We build none of that, at all. We're not even close.
Your win condition is, simply, a modelled company, plain and simple. The Foundry Ontology binds data, logic, and action into one object model. Two full decades of that craft. A young vendor just can't copy it.
You are federal or defense. Their own material lists both FedRAMP and IL5/6. We have no accreditation path at all, today.
You'd really rather buy outcomes than parts, always. You want a bootcamp that lands a working use case, in just five days.

Bottom line: Foundry wants to be where your whole company runs. We want to be the layer that decides what an agent may touch and keeps proof of it. If you already own Foundry, this page is about that second job, not the first.

Feature by feature

Competitor cells describe what Palantir's public site and docs say. We use the date above. If we got something wrong, email support@myorg.ai. We'll fix it and credit you.

What matters	DataShield	Palantir Foundry	Edge
Operational apps and workflows	Simply not our job. We ingest, catalog, profile, transform, snapshot, and serve datasets. There's no app builder here, at all, none. No orchestrator. No digital twin, either.	The whole point. Pipeline Builder, Foundry Rules, dynamic scheduling. Process mining, streaming, a marketplace. Plus apps real operators use daily.	◇
Semantic model	DataShield Ontology gives you a catalog. Asset and column metadata, all there. Add a business glossary too. It's built from entity types. Add typed lineage you can trace. It has access gating at each single hop.	The Foundry Ontology models objects, actions, and business logic. All of it. AIP binds models and optimizers to it, treating them as tools. That's deeper than ours, for sure, no question. It's the reason customers actually buy in.	◇
Audit evidence	SHA-256 hash chain with Ed25519-signed checkpoints that are themselves chained. Verification names the failure: tampering, insertion, deletion or truncation. Try the verifier.	AIP's page says "All action can be audited". We found no public cryptographic proof system. Just ask them what stops a platform admin. From editing the record.	◆
Agent authorization	Each checked tool call passes a scope ceiling. It also carries a declared access tier. Plus a revocation re-check, before dispatch. We meter it and tie it to the agent. The call just fails closed, each time.	OAuth grant, app limits, and the user's own Foundry permissions. All three scope Ontology MCP tokens. That's real authorization, sure, at grant time. We found no per-call re-check described.	◆
Break-glass	Scoped, time-boxed emergency access for agents. It auto-revokes. It's both admin and IP gated. It can't be quietly deleted from the log, ever.	Not described in their public docs, that we could find.	◆
GDPR erasure	Crypto-shred of per-subject key material. Plus ISO 27560 consent receipts. HMAC commitments cover actor identities in the chain, not names. The audit still checks out, even so. That's after the subject is gone.	Purpose-based access controls and tight oversight are well laid out. The erasure method just isn't. Neither is what it does to old audit records, either.	◆
Tokenization and classification	We tokenize data right at ingest. Agents then query tokenized data over MCP. Detokenization is a privileged, checked step, each time. We classify fields against 129 built-in classes. PII, PHI, financial, and secret, in all. That includes all 18 HIPAA Safe Harbor identifiers. Results stay fully repeatable, with no model in the loop, ever.	Security markings, purpose-based access controls, and tight AI guardrails. Tokenization just isn't vocabulary they use, at all, in public docs.	◆
MCP and agents	There are more than 200 MCP tools. They spread across Ontology, Auth, Corpus, and Lighthouse. Auth issues MCP tool tokens with scope ceilings. Delegation always uses RFC 8693 token exchange, with an enforced ceiling.	There are two public MCP surfaces. Ontology MCP lets external agents read objects. It lets them run set actions, and query data. Palantir MCP gives builders 70+ tools. It just can't write ontology data. That's clean design, and it's clearly explained.	◈
Entity resolution	Fellegi-Sunter probabilistic linkage. It uses Jaro-Winkler, Levenshtein, Soundex, and Double Metaphone comparators. All of them. Add LSH blocking too. Plus EM-trained parameters, and five survivorship strategies. Drift monitoring runs on every single promoted match setup.	Foundry ships entity resolution right as part of the platform. The matching method just isn't public, in the detail ours is. Go compare them on a real file, not on a page.	◈
Deployment	Self-hosted, right in your own cloud or data center. Or on a dedicated single-tenant server we run for you. Docker images cover Auth, Ontology, Corpus, and Lighthouse. A signed deploy manifest, and Guardian checks it. Ed25519 audit-signing keys stay right on your instance. They sit in a key directory you fully control. Kid-keyed rotation always fails closed, on an unknown key. KMS or HSM custody is a built seam in Auth. It's not a shipped setup yet. HMAC tokenization keys sit in your own setup today. Not in a KMS, not yet.	Cloud, on-prem, and accredited government setups, including classified ones too. Virtual Tables let the Ontology read BigQuery, Snowflake, S3. Or Azure Blob. Right in place. They claim this needs "no data replication or duplication."	◇
Independence	We simply govern data. We don't own it. Same with running it. There's no incentive to keep your workloads anywhere in particular.	The platform under review owns all of it. Every piece. Think the checks, the semantic model, the apps, and the agents. For some buyers, that's the point. For others, it's the objection.	◆
Maturity signals	Auth, Guardian and Lighthouse are live. Guardian and Lighthouse have run since April 2026. SOC 2 isn't certified yet, and we say so.	Founded, way back, in 2003. NYSE listed, publicly. Forrester Wave Leader for AI/ML Platforms. Ranked first by Dresner in agentic AI, in 2025. A customer list with Airbus, Swiss Re, Cleveland Clinic. Plus others.	◇
Pricing	Published model, scoped instant quote, no sales wall.	Quote only, no listed price. Their sitemap carries 620 pages. None of them is a price list. Every single path ends at a form.	◆

◆ DataShield leads◇ Palantir Foundry leads◈ comparable

Palantir claims come from palantir.com and the Foundry documentation. Last checked 13 September 2026. We link the sources below, not from memory.

Three things you get here that you won't get from an enterprise operating system

Proof that survives an audit

"Auditable" is a property of a log. Tamper-evident is a property of math, plain and simple. Ours is a hash chain, with signed checkpoints, always. The verifier names what broke. Not just that something did. That's what reviewers are actually asking for. Think EU AI Act Article 12. Think HIPAA §164.312(b). Try it in your browser, no signup.

Authority that can change mid-flight

An analyst resigns on a Friday. Their agent is 20 minutes into a 40-minute job, still running. With DataShield, we re-check the next checked tool call. Against current access. It just fails closed, right then. Scoped OAuth tokens are good, sure. But a token issued this morning still believes in this morning. How Auth does it.

A stack you can install yourself

Docker images, a signed deploy manifest. Your own Postgres, your own keys. No bootcamp, at all. No forward-deployed engineers, either. Nor a five-day camp to get a first use case out. If your team can run a container, it can run this. See the architecture.

Where Palantir Foundry is genuinely stronger

Almost everywhere you look. Except our one layer. Foundry integrates the source systems, all of them. It models them. As objects and actions. Then it lets people build the apps. The ones the business runs on. Their AIP AgentCamp promises this: "deployed systems executing real work in your environment". That happens in a week. The Bootcamp before it went "From 0 to use case in 5 days". That's a whole delivery machine. We simply don't have one. Their accreditation list is long. Their own Builders page lists it. GDPR, HIPAA, FedRAMP, IL5/6, and SOC 2. That's a door we can't open yet. And the whole AIP guardrail story is well thought out. The Ontology binds approved logic. It treats that logic as fixed tools. AI-authored proposals can always route to a human, too. Virtual Tables let them read your warehouse. They never copy it. They've shipped two MCP surfaces, too. They laid out that split, quite clearly. That's more than most vendors managed this year.

Here's the honest push-back. It's narrow, on purpose. Foundry's control is excellent, sure. But it's Foundry's own. The permissions themselves live inside that same platform. So do the markings. So do the ontology, the agents, and the audit trail. You'd be asking that same party to prove nothing changed. But that party is the same one that runs the database. That's fine, until the day someone wants proof. A regulator, a counterparty, or your own board. Proof that doesn't depend on trusting one vendor. Our chain checks out offline. Right on your laptop. Against keys you alone hold. Gartner expects most unauthorized agent transactions through 2028 to be internal policy violations. Not attacks. Internal violations are the exact case. That's where an in-platform log is weakest. So go run the work on Foundry. Keep the evidence somewhere the work simply can't reach.

Questions worth asking both of us

These are the questions we would want answered. That's if we were the ones buying. Ask them on every call. Ours too.

Can you cryptographically prove an audit log entry wasn't deleted?

DataShield: yes, it can. Each record commits to the one right before it. Checkpoints are Ed25519-signed, and chained, always. Verification tells deletion apart from truncation, and apart from tampering too. Run it against a sample chain at /verify. Palantir: their AIP page claims all action gets checked. Foundry's access controls are, in fact, tight. We found no public tamper-evidence system at all. Just ask them what a platform admin can change. And how you'd even know.

What happens to a revoked agent mid-session?

DataShield re-checks access on each checked tool call. Revocation lands on the next call, and the context downgrades. Palantir's Ontology MCP docs describe OAuth scopes. Plus app limits, and per-user permissions. That's sound enough grant-time authorization. We just couldn't find a mid-session re-check. Ask how long an agent keeps working. That's after you pull the human's access.

How does GDPR erasure interact with the audit trail?

DataShield crypto-shreds per-subject key material. Then it issues an ISO 27560 consent receipt. HMAC commitments cover actor identities in the chain, not names. The evidence still checks out. Even once the subject is gone. Palantir documents purpose-based access controls. Plus retention too, at the platform level. What erasure does to old audit records, we could not find. Just ask for the method, not the policy.

You both say "Ontology". Are these the same thing?

No, and we should both have probably picked another word. Palantir's own Ontology is the object and action model. It sits at the heart of Foundry. It represents your whole company. That way apps and AIP agents can act on it. DataShield Ontology is our own data platform. Think ingest, catalog, field classification, tokenization, master data, RAG, and analysis. All of it, reachable over MCP. If your problem is modelling how the business works, that's theirs. If your problem is governing what an agent may read: that's ours. So is proving what it did.

Is DataShield a Palantir Foundry alternative for mid-market buyers?

For part of the job, yes. Palantir has come down-market, quite deliberately. Palantir for Builders offers startups and SMBs "full access to AIP and Foundry". The page says big data solutions shouldn't just be for big companies. That's a really serious motion. But it's still a form. A conversation. A delivery job. Maybe what you need is simply the checked data plane. Just under your agents. Installed by your own team. Running on your own hardware. At a price you can read today. Just buy that from us. Just skip the sales cycle. Do you need the whole operation modelled and rebuilt? You want them instead.

Does DataShield have SOC 2?

Not yet, and we won't imply otherwise. Palantir's Builders page lists SOC 2, FedRAMP, and IL5/6. They're really not overselling it. Auth is live. It ships with a public threat model. Plus a verifier anyone can run. Guardian and Lighthouse have been live since April 2026. Design-partner terms include source escrow. That way a small vendor isn't a single point of failure. Details on the security page.

Foundry's hero: "The Ontology-Powered Operating System for the Modern Enterprise", with the primary CTA "Get Started with the Ontology". - palantir.com, 13 Sep 2026
AIP describes the Ontology as "a decision-centric system at the heart of AIP", with granular AI guardrails and the line "All action can be audited." - palantir.com, 13 Sep 2026
Ontology MCP lets external AI agents read objects, execute predefined actions, and query ontology data. Palantir MCP gives builders 70+ tools, but "cannot write actual ontology data." - Palantir Foundry docs, 13 Sep 2026
Ontology MCP authorization is OAuth 2.0. It uses authorization-code or client-credentials grants, with tokens "scoped to the operations and resources you have granted to the application." - Palantir Foundry docs, 13 Sep 2026
Palantir for Builders takes Foundry and AIP to startups and SMBs. It lists GDPR, HIPAA, FedRAMP, IL5/6, and SOC 2. - palantir.com, 13 Sep 2026
AIP AgentCamp promises "No Presentations. No Theoretical Pilots." and "deployed systems executing real work in your environment"; the AIP Bootcamp before it ran "From 0 to use case in 5 days." - palantir.com, 13 Sep 2026

Other head-to-heads

Same market

Get free Demo Center access

You've seen the proof

Ready for a number? Scope your deployment and we'll price it against your own economics.

Get your quote →