Head-to-head · updated 13 September 2026

DataShield vs Informatica CDGC: the catalog knows the data is sensitive, so what stopped the agent?

Cloud Data Governance and Catalog is the governance module of Informatica's IDMC platform, and it is very good at its job. Point it at a thirty-year-old estate and it will crawl the metadata, draw lineage back to the mainframe, auto-classify columns with CLAIRE, hang a business glossary off the result, and publish governed data products into a marketplace. Their own page says 5,000-plus enterprises trust Informatica. We believe it. Active metadata management at that scale is hard, and they have been doing it since 1993.

We are not a replacement for that. DataShield is the layer underneath the datasets your agents actually query. Datasets are tokenized at ingest; agents query tokenized data over MCP; detokenization is a privileged, audited operation. Every governed tool call is checked against the agent's current authority before dispatch, and the decision is sealed into a hash chain you can verify without trusting us. A catalog tells an agent a column is sensitive. It does not refuse the call. Below is the honest split, including the many rows CDGC wins.

Get free Demo Center access Get an instant quote

The short version

Pick DataShield when

Someone will one day ask you to prove an agent's access log was not edited. An examiner, an internal auditor, or Article 12 of the EU AI Act. Our chain answers with math rather than a policy PDF. Run the verifier.
You need to pull an agent's authority mid-session and have the very next tool call fail. Not the next token refresh. The next call.
The control plane has to run inside your own network. CDGC is delivered as part of IDMC and we found no self-hosted edition. We self-host, or run you a dedicated single-tenant server.
You want a classification verdict you can re-derive. Ours is deterministic and stamped with a config digest, so the same column scores the same way next quarter. A model output is not.

Pick Informatica CDGC when

Your estate is wide and old. Mainframes, twelve warehouses, SAP, a BI tool nobody maintains. Their connector coverage took decades and we are not going to match it.
You already run IDMC. CDGC draws on the same consumption pool, so it is a line on a renewal instead of a new procurement. That is a large and boring advantage.
The buyer is the CDO, and the deliverable is a glossary, a stewardship org chart, and lineage a business analyst can read without help. Their UX is built for that person. Ours is built for an engineer.
You are standardising on Agentforce 360. Being owned by Salesforce then reads as a roadmap promise rather than a risk.

Bottom line: CDGC is the metadata plane for your whole estate. DataShield is the data plane for the datasets agents touch, where the obligation is enforcement and evidence. Most buyers who talk to us keep their catalog.

Feature by feature

Competitor cells describe what Informatica's public site and press releases say as of the date above. If we have mischaracterised something, email support@myorg.ai and we will correct it, credited.

What matters	DataShield	Informatica CDGC	Edge
Catalog and connector breadth	We register a provider over a live connection, scan it, and profile objects and columns in place with no rows leaving the source. That is PostgreSQL today. Snowflake, BigQuery, Databricks, SQL Server, S3 and Salesforce are declared with no handler yet, and we say so in the config rather than in a footnote.	The core product, and the best part of it. Automated metadata scanning across databases, files, cloud warehouses and BI tools, with decades of connector work behind it.	◇
Lineage	Typed lineage over a fixed predicate vocabulary, traversed with per-hop access gating: an edge never reveals an asset you are not allowed to read. It is derived from the pipelines that own the relationships, not a stored column-level graph.	End to end and column-level lineage with shared business context linking technical assets to glossary terms. Deeper than ours, and better looking.	◇
Classification method	129 field classes covering PII, PHI, financial data and secrets, including all 18 HIPAA Safe Harbor identifiers with a regulation reference on each. Regex plus checksum validation (Luhn, NPI, Verhoeff, ABA, IBAN, GTIN), column-name lexicons and anti-pattern suppressors. No model, so a verdict is reproducible from a config digest. It shipped recently, and we will not call it battle-tested.	AI-powered classification via CLAIRE, which auto-maps business terms to technical assets at a scale we cannot touch. Ask them to re-run a verdict from six months ago and show you the same answer.	◈
Sample handling	Catalog scans copy no column values by default. Storing plaintext samples needs an explicit acknowledgement and writes an audit row, cited to GDPR Article 5(1)(c) and minimum necessary.	Not described in their public material. Worth asking what CDGC retains in the cloud once it profiles a column of medical record numbers.	◆
Audit evidence	SHA-256 hash chain with Ed25519-signed checkpoints that are themselves chained. The verdict is three-valued: clean, attested damage, or tampered. Verification names the failure as tampering, insertion, deletion or truncation. Try the verifier.	Governance reporting and policy audit inside IDMC. We found no published cryptographic tamper evidence.	◆
Agent authorization	Every governed tool call passes a scope ceiling, a consented-tool allowlist, an authority tier, and a live revocation re-check before dispatch. It fails closed.	Policy-driven governance and access workflows over catalogued assets. The agent story is preparing trusted data for someone else's agent, not authorizing a specific call.	◆
Break-glass	Scoped, time-boxed emergency access for agents. It auto-revokes, it is step-up gated, and it cannot be quietly deleted from the log.	Not described in their public material.	◆
GDPR erasure	Crypto-shred of per-subject key material, plus ISO 27560 consent receipts. Actor identities in the chain are HMAC-committed, so the evidence still verifies once the subject is gone.	Governance policy and retention workflow. The underlying erasure mechanism is not described for CDGC.	◆
Tokenization	Deterministic, join-preserving, vault-reversible tokens applied at ingest, plus quasi-identifier generalization (dates to year, decade or age band; ZIPs to 3 or 4 digits; partial phones, SSNs and emails) with a measured cardinality-reduction score per column. Masking and generalization are features you switch on, not defaults.	Classification, policy and access control. Tokenization is not CDGC vocabulary. Informatica sells privacy capability elsewhere in the platform.	◆
MCP and agents	More than 200 MCP tools across Ontology, Auth, Corpus and Lighthouse. Tool tokens carry scope ceilings, delegation is RFC 8693 token exchange with an enforced ceiling, and every call is metered and attributed to the agent.	Real and dated. On 20 May 2026 they announced Metadata Explorer, MDM and data quality MCP servers, in preview on the AWS Agent Registry and generally available in Amazon Quick, both US regions. It serves metadata and master data to agents.	◈
Deployment	Self-hosted in your own cloud or data center, or a dedicated single-tenant server we operate. Docker images for Auth, Ontology, Corpus and Lighthouse, with a signed deploy manifest Guardian verifies. Ed25519 audit-signing keys can live in your own KMS or HSM. HMAC tokenization keys sit in your environment today, not in a KMS, and we will not pretend otherwise.	CDGC ships only inside IDMC, Informatica's cloud. Regional instances give you residency choices. We found no self-hosted or air-gapped CDGC edition.	◆
MDM alongside the catalog	In the same product: Fellegi-Sunter probabilistic record linkage, Jaro-Winkler, Levenshtein, Soundex and Double Metaphone comparators, LSH blocking, EM-trained parameters, five survivorship strategies and Population Stability Index drift monitoring on every promoted match config.	A separate IDMC module with a much longer track record, a multidomain UI, and an implementation partner ecosystem we do not have.	◈
Maturity signals	Auth, Guardian and Lighthouse are live in production (Guardian and Lighthouse since April 2026). SOC 2 not yet certified, and we say so.	Founded 1993. 5,000-plus enterprises claimed. Acquired by Salesforce in a deal that closed 18 November 2025.	◇
Pricing	Published model, scoped instant quote, no sales wall.	Consumption pricing in Informatica Processing Units, pooled across IDMC modules. No published rate for CDGC. G2's pricing tab for it is empty.	◆

◆ DataShield leads◇ Informatica CDGC leads◈ comparable

Informatica claims are drawn from informatica.com, Informatica press releases and Salesforce press releases, last checked 13 September 2026. We link them below rather than work from memory.

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

Proof that survives an audit

A log that can be silently edited proves nothing. Ours is a hash chain with signed checkpoints, and the verifier tells you what broke, not just that something did. Re-tamper a chain that was already marked as damaged and it un-attests itself. That is the property EU AI Act Article 12 and HIPAA §164.312(b) reviewers care about. Try it in your browser, no signup.

Authority that can change mid-flight

An analyst leaves on a Friday. Their agent is 20 minutes into a 40-minute job over a governed dataset. Here the next tool call is re-checked against current authority and fails closed. A catalog will still be labelling that column as sensitive while the job runs to completion. How Auth does it.

A catalog that keeps your values out of it

Our scans profile inside your PostgreSQL source and no rows leave it. Sample values default to omit, so nothing lands in the catalogue. If you want plaintext samples, you acknowledge it and we write an audit row. Ask any SaaS catalog what it keeps after it profiles a column of MRNs. See the Ontology surface.

Where Informatica CDGC is genuinely stronger

Breadth, and it is not close. Informatica has been building connectors since 1993, and CDGC inherits all of it. Lineage runs end to end and down to the column. The glossary and stewardship workflow are built for a governance office with real headcount, not for an engineer with a terminal. CLAIRE auto-maps business terms across an estate at a scale our deterministic classifier will never reach, and they publish a data marketplace so teams can request governed data products instead of filing a ticket. Their agent work is real too: MCP servers for metadata, data quality and MDM, shipped in May 2026 into the AWS Agent Registry and Amazon Quick. If your problem is that nobody knows what data the company owns, buy them, not us.

Here is the push-back, and it is two-part. First, CDGC lives in Informatica's cloud. There is no self-hosted edition we could find, so your metadata, your lineage and your classification results sit with a vendor. For some regulated buyers that ends the conversation before price comes up. Second, on the day the deal closed, Amit Walia called the combined company the "Switzerland" of AI-powered data management. Switzerland is a striking word for a business unit whose president now carries a Salesforce title and whose stated job is powering Agentforce 360. Maybe the roadmap stays neutral. Ask them how a non-Salesforce request gets prioritised against an Agentforce one, and write down the answer.

Questions worth asking both of us

These are the questions we would want answered if we were the ones buying. Ask them on every call, ours included.

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

DataShield: yes. Each record commits to the one before it, checkpoints are signed and chained, and verification tells deletion apart from truncation and from tampering. Run it against a sample chain at /verify. Informatica: CDGC reports on policy and governance activity inside IDMC. We found no published tamper-evidence mechanism. Ask them to show one, and ask who can edit the store it sits in.

What happens to a revoked agent mid-session?

We re-check authority on every governed tool call, so revocation lands on the next call and the context downgrades. CDGC governs access to catalogued assets through policy and workflow. We could not find a mid-session cut-off in their public docs. Ask how long a compromised agent keeps working after you pull its access, and get the number in minutes.

How does GDPR erasure interact with the audit trail?

We crypto-shred the subject's key material and issue an ISO 27560 consent receipt. Actor identities in the chain are HMAC-committed, so the evidence still verifies after the subject is gone. Informatica supports governance policy and retention at the platform level. The erasure mechanism for CDGC is not public. Ask for the mechanism, not the workflow.

We already run IDMC. Do we drop CDGC for DataShield?

No. If you are paying for IDMC, CDGC is close to free at the margin and we cannot cover your estate. Keep it as the metadata plane. Put us under the specific datasets your agents query, where the job is authorizing the call and holding evidence of it. The question to settle internally is not which catalog, it is which layer the obligation lands in when a regulator asks what an agent did on 3 March.

Informatica ships MCP servers now. What's different about yours?

Theirs serve metadata, data quality and master data so an agent can reason about your estate. Announced 20 May 2026, preview on the AWS Agent Registry, generally available in Amazon Quick, US regions. Good design for a catalog. Ours is where the governed data itself gets queried, so the tool token carries a scope ceiling, the call is authorized before dispatch, metering is attributed to the agent, and the decision is sealed into the chain. Different jobs. Ask them for a production reference running agent workloads, not a registry listing.

Does it matter that Salesforce owns Informatica now?

It might. The $8 billion deal closed on 18 November 2025 and the site now brands as Informatica from Salesforce. If you run a mixed stack, ask how CDGC roadmap items get ranked against Agentforce 360 ones. We are small and free of all that, which cuts both ways: no parent steering us, and no parent balance sheet either. Our design-partner terms include source escrow for exactly that reason.

Does DataShield have SOC 2?

Not yet, and we will not imply otherwise. Auth is live with a public threat model and a verifier anyone can run. Guardian and Lighthouse have been in production since April 2026. Details on the security page.

CDGC's own tagline: "Simplify, automate and scale data and AI governance to deliver trusted business outcomes." The page cites 5,000-plus enterprises and names AI-powered classification and automated lineage as headline capabilities. — informatica.com, 13 Sep 2026
Salesforce completed its acquisition of Informatica on 18 November 2025; Amit Walia called the combined company the "Switzerland" of AI-powered data management, and Steve Fisher framed it as a governed data platform for Agentforce 360. — Informatica news release, 18 Nov 2025
Salesforce agreed to acquire Informatica for approximately $8 billion in equity value at $25 per share, to build "a unified architecture for agentic AI." — Salesforce press release, 27 May 2025
Headless data management for AWS: Metadata Explorer, MDM and data quality MCP servers, in preview on the AWS Agent Registry and generally available in Amazon Quick, US regions, plus CLAIRE Agent skills in global preview. — Informatica news release, 20 May 2026
CLAIRE is positioned as "Trusted Enterprise AI, Powered by Metadata" and the page carries vendor-published impact figures including $63.6M total financial impact over five years. — informatica.com, 13 Sep 2026
No published price for CDGC: G2's pricing tab for the product returns no data. — G2, 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 →