Kadeck vs Lenses
ComparisonsKadeck vs Lenses, compared
Key takeaway
Kadeck and Lenses are both commercially licensed tools on top of a Kafka cluster somebody else runs. Lenses reaches topic data through SQL, Kadeck through a filterable record browser. Kadeck Enterprise costs 32 US dollars per user per month with a ten-user minimum; Lenses DevX Team starts at 4,000 a year for fifteen users. Kadeck stays per seat at any size, while Lenses moves to negotiated pricing past fifteen, and needs Postgres for HQ plus an Agent database per cluster, where Kadeck Teams is one Docker image. Kpow by Factor House is licensed per cluster at a published price.
What is Lenses?
Lenses is a commercial Kafka governance and data exploration platform that sits on top of existing clusters. It is a Kafka client rather than a proxy, so it does not sit in the data path, and KRaft clusters need no modification because the Agent connects as an ordinary client. The architecture is a central Lenses HQ node with lightweight Agents deployed per cluster, and one Agent reaches one Kafka cluster. Celonis acquired Lenses in early 2022, and 6.2.6 shipped on 19 August 2026.
- SQL Studio: queries a topic in SQL rather than in consumer code, which puts topic data in reach of an analyst.
- SQL Processors: Kubernetes-native stream processing defined in SQL on top of Kafka Streams.
- Topology and lineage: one picture across producers, topics, connectors and consumers.
- Two ladders: DevX for the UI, governance and SQL surface, and K2K for replication.
What is Kadeck?
Kadeck is a commercially licensed Kafka management and data exploration tool from Xeotek. It is not open source, and it supports Apache Kafka, Redpanda and Amazon Kinesis. The fact that shapes everything downstream is not a feature: it ships as two separate product lines rather than as two rungs of one ladder. Kadeck Desktop is a native application for Windows, macOS on both Intel and Apple Silicon, and Linux, so an individual engineer needs no container runtime. Kadeck Teams is the web product, distributed exclusively as a Docker image.
- Record view: Avro decoded through a Schema Registry and presented in a columnar layout rather than as raw bytes.
- QuickProcessor: a JavaScript expression derives calculated fields, with no streaming application behind it.
- Dead-letter recovery: isolate, transform inline, preview, re-ingest, then purge what has been processed.
- Metadata layer: schema fields documented without modifying the Schema Registry or bumping a version.
What is the official 2026 pricing of Kadeck and Lenses?
Kadeck’s Enterprise floor and Lenses’ DevX Team floor land within 160 US dollars a year of each other, and on neither side is that the headline tier. Kadeck’s RBAC with LDAP and OpenID Connect, its data masking and its audit logs are Enterprise-only, at 32 US dollars per user per month with a ten-user minimum, so the entry cost is 3,840 US dollars a year at any team size below ten. Lenses puts SSO, SAML and RBAC at DevX Team, from 4,000 US dollars a year for up to fifteen users. Two products whose headline tiers look an order of magnitude apart converge the moment access control becomes a requirement.
After that they diverge. For a team of fifty, Kadeck stays arithmetic at 19,200 US dollars a year, while fifty users is past the Lenses Team cap of fifteen and the deal becomes a Multi-Kafka Enterprise contract priced by negotiation. Two smaller things move the total. Kadeck Professional at 19 US dollars per user per month is the Desktop line, and a desktop licence is not upgradeable into Enterprise. Lenses carries a second meter, because exactly-once replication, offset replication or schema migration puts a team on the K2K ladder on top of DevX, from 1,000 US dollars a month with five clusters included and 200 a month per cluster after.
Where does each one run out?
Lenses’ control plane is wider than it first looks: HQ plus its Postgres, plus one Agent and one Agent database per Kafka cluster, and the Community edition carries the same dependency. Its own Helm quickstart creates lenses_hq and lenses_agent before either component starts.
- Masking: data policies match on field name across every registered dataset and are global, with no way to escape a policy even as an admin.
- Portability: SQL Processors are proprietary and compiled, so anything built on them is re-implementation if the team leaves.
- Permissions: 6.2.5 split
UpdateTopicDetailsinto two actions, so custom roles granting the old one needed re-granting by hand. - Weight: a server-side JVM backend runs alongside the web interface, which is heavy for an engineer who only wants to look at a local topic.
Kadeck’s Teams edition ships as a Docker image and nothing else, and the documented production Kubernetes path wants a persistent external database.
- Licence checks: every container start triggers an online validation, so air-gapped sites need offline activation and CI deployments need challenge-response activation.
- Free tier: one cluster connection, so covering Dev, Test, Stage and Prod is a paid plan on its own.
- Write path: read-oriented, with no streaming of derived data back into a topic and no Kafka Streams or ksqlDB integration.
- Export: CSV respects the column filters applied in the record table and JSON does not.
Which should you pick?
Pick Kadeck if:
- the tool is for one engineer with one laptop and one cluster today
- a native installer with no container runtime and no database is what is wanted
- the work is recovering from failures rather than reporting on them
Pick Lenses if:
- people who will not write consumer code need to answer their own questions
- SQL over a topic is the capability being bought
- stream processing defined in SQL is part of the plan
That last pair is the one axis on which these two are genuinely not substitutes: a record browser with filters is a debugging surface for the person holding the incident, and SQL is a self-service surface for everybody else. If access control has become somebody’s requirement, count the whole cost rather than the tier. Both floors sit within 160 US dollars a year of each other, and what differs is what happens next: Kadeck bills every seat forever behind a ten-seat minimum, while Lenses covers fifteen users flat and then moves to a negotiated contract.
Kpow: no seat count to negotiate, no control plane to run
Kadeck stays per seat at any size behind a ten-seat Enterprise floor, and Lenses covers users flat only up to fifteen before the deal becomes a Multi-Kafka Enterprise contract priced by negotiation, so both roads eventually turn into a headcount conversation. Both also cost infrastructure before they cost people: Kadeck’s Teams edition wants a persistent external database, and Lenses needs a central HQ node on PostgreSQL plus a separate Agent and Agent database for every cluster it reaches. Kpow by Factor House avoids both: licensed per cluster at a published price, adding an engineer never moves what you pay, and there is no external database, no sidecar and no persistent volume behind it — a single stateless JVM container, configured through environment variables, reaching up to 12 clusters from one instance.
No seat count to negotiate and no control plane to keep alive is the whole saving. Check the number before a headcount conversation becomes a contract negotiation — it is published in full on Kpow’s product page.
Sources
- Apache Kafka documentation on authorization
- Apache Kafka documentation on basic operations
- Apache Avro documentation