console.<slug>.blynk.ae
Your lake
Trino, Polaris, Iceberg, and the plan-first console. Release 0.2.8 is the current alpha.
Alpha release 0.2.8 · invite only
blynk is a dedicated AI lakehouse platform. You describe the outcome. The agent drafts a validated plan. You approve. Governed Iceberg stays inside your stack.
The lake product · release 0.2.8
This is the appliance loop that runs inside your stack — on
console.<slug>.blynk.ae. Pick an intent and watch it refuse to guess.
Project bound to lake.sales · federated: pg_prod, mysql_billing. Every guarded write waits for your approval. Lake Temporal runs the plan after you say yes.
Suggested intents
Guarded writes — ingest, write SQL, provisioning, drops — always stop for Approve.
Reversible write_safe actions can run autonomously, journaled with a one-click undo.
Where you work
Same appliance either way — your cluster, or a dedicated VPC. Analysts still land on
console.<slug>.blynk.ae. No shared query engine.
console.<slug>.blynk.ae
Trino, Polaris, Iceberg, and the plan-first console. Release 0.2.8 is the current alpha.
Work email on this site. Invite only — not self-serve.
We confirm the invite and the region. Destroy always takes two people.
We install release 0.2.8 into your stack. Nothing starts without an approval.
You launch console.<slug>.blynk.ae.
Why we build blynk
Teams still hand-stitch catalogs, query engines, pipelines, identity and AI tooling — then hope governance holds. A shared multi-tenant engine is faster to sell and worse to isolate. Handing an LLM the keys is faster still, and far worse.
Not a schema in a shared Trino. Isolation is the VPC or cluster, the catalog, and the identity realm.
The agent proposes; operators decide. Approval is a product primitive, not a policy document.
The model, the metadata and the bytes stay inside your perimeter. SQL never leaves blynk.
Iceberg, Polaris, Trino, SeaTunnel — fixed versions on release 0.2.8. Never :latest.
Platform · release 0.2.8
What runs on console.<slug>.blynk.ae after your stack is Ready.
Durable, named chat workspaces bound to a catalog and namespace. Messages, plans, runs and decisions live together, so the "why" survives long after the query.
information_schemaSeaTunnel writes into Apache Iceberg tables cataloged by Polaris, with per-user principals.
Attach Postgres, MySQL and JDBC catalogs to Trino, then plan across them in one prompt.
llama.cpp drafts plans, writes SQL and diagnoses failures — on hardware you control.
Every guarded mutation resolves to a plan a person signed. Reversible actions are journaled with the exact operation that undoes them.
Trino Gateway routing plus Superset dashboards, wired to the same governed tables.
Architecture
The lake talks to ports. Engines are adapters — swappable, versioned, and never
hardcoded into business logic. Release 0.2.8 is what we install into
console.<slug>.blynk.ae.
0.2.8 · versioned engines:latest
Alpha
Alpha is not a storefront. We provision a dedicated stack after we accept your request — AWS-hosted or on-premise.
What you get
console.<slug>.blynk.aeGovernance
An agent that can reach production is only safe if the blast radius is defined by the platform, not by prompt discipline.
sqlguard.ValidateWrite allowlists verbs and rejects federated sinks before a plan can be approved.
write_safe action records the operation that undoes it, so autonomy stays recoverable.
Trust
The lake is an appliance. We install it into an isolated stack you control — a dedicated
AWS account and VPC, or your own Kubernetes. Secrets never live in git. SQL never leaves
blynk.
AWS-hosted
One account boundary per customer. We apply Terraform + Helm into that perimeter.
On-premise
Same 0.2.8 appliance on hardware you already run. Overlay changes the adapters, not the product.
We provision a dedicated stack and you launch
console.<slug>.blynk.ae. Invite only. No shared query engine.