LHO Features Walkthrough
A features overview — what the Lakehouse Optimizer does across cost, optimization, governance, and monitoring, and the value it delivers.
- 1 01 · What the Lakehouse Optimizer is
- 2 02 · The leadership altitude — Executive Insights
- 3 03 · Cost you can trust
- 4 04 · Catch problems before the bill
- 5 05 · Chargeback and accountability
- 6 06 · Where the money goes — and proving the architecture
- 7 07 · Know what to change
- 8 08 · Stop paying twice — consolidation and governance
- 9 09 · Always-on assurance
- 10 10 · Proof it worked
- 11 11 · The business case
The Lakehouse Optimizer (LHO) is Blueprint's Augmented FinOps platform for Databricks. It blends human insight with AI across the whole FinOps lifecycle — Observe → Analyze → Forecast → Optimize → Govern — and watches your Databricks estate and the cloud resources underneath it, because the true cost of a data platform is both.
This article walks LHO's capabilities feature by feature, with the business value beside each one. To keep it concrete, it follows one anonymized example: a fast-growing, PE-backed services company that had just moved its platform onto Databricks and gone all-in on serverless — growing about 22% a year and acquiring 10–15 businesses annually. What a company like this needs isn't another dashboard; it's confident consumption as it scales, and a cost story its leadership and sponsors can trust.
Get more from every Databricks dollar — LHO makes the money you spend on Databricks and the cloud visible, trustworthy, and smaller, and gives your teams the exact, ranked list of what to do next. FinOps outcomes without a FinOps team.
01 · What the Lakehouse Optimizer is
LHO page: Landing → Forecasting
LHO sits better-together with Databricks — the deep FinOps layer on top of the platform, not a replacement for it. Everything rolls up to four value pillars. A good first session lays out that map, then follows whichever pillar the customer cares about most — in our example, cost control.
Pillar | What it does |
|---|---|
Optimization | Ranked, evidence-backed recommendations across cost, performance, and orchestration — including vendor consolidation. |
Cost Management | Trustworthy cloud + Databricks cost, forecasting, budgets, KPIs, and threshold-based control. |
Unity Catalog Migration | Assess the leftover Hive/legacy footprint and finish governing your estate. |
24/7 Monitoring | The always-on foundation — cost, performance, and orchestration, weekends included. |
LHO delivers FinOps outcomes without a FinOps team — and it pays for itself quickly.
👇 Click to expand and see product screenshots
02 · The leadership altitude — Executive Insights
LHO page: Executive Insights
For a growing, acquisitive company, the hardest question isn't what happened — it's where do we focus? Executives who own Databricks live in a sea of consumption: many workspaces, spend split across Databricks and the cloud, often other platforms too. The tools they have show trends in arrears and can't even tell whether a problem sits on the Databricks side or the cloud side. At sophisticated organizations, a team can spend weeks exporting data, building spreadsheets, and assembling slides just to put one prioritized view in front of a decision-maker.
Executive Insights — the Executive Insights Engine, Blueprint's headline LHO launch and its first premium, licensable app — collapses that into an instant, prioritized summary: the leadership view of the entire Databricks estate on one page. It answers three questions an executive actually asks.
What do I have? · How am I doing? · What should I do about it?
The top of the page is the estate's posture at a glance: a single Health Score (how "% healthy" the estate is), total cost with a conservative Potential Savings range, the Databricks Cost Plan status with an early warning if committed capacity is trending to run out before the plan ends, plus estate posture — workspaces, % on Unity Catalog, regions, metastores — and the daily-cost, cost-per-TB, and throughput KPIs with their trends.
Below that, Top Optimization Opportunities organizes everything into three lenses — Orchestration (the vendor-consolidation view, the default), Performance, and Cost Savings — each with ranked, dollar-and-workload-sized opportunity cards carrying a Low/Medium/High complexity badge. Every opportunity is one click into LHO's Recommendations, pre-filtered, so the executive picks the theme and the team lands on the exact, actionable list.
Insights, not dashboards — surface what matters, hide the rest. Executives skim "are we healthy, where do I focus?"; their managers and platform owners drill into a tile and act.
Ask Genie
Even simplified, an executive will have a follow-up — "what does suboptimal cost actually mean?" — or need to narrow 100+ workspaces to the one they care about. Ask Genie takes the question in plain English and answers it right in the page, drawn only from data they're already allowed to see. Powered by Databricks; personalized; built to help a customer run their estate with confidence.
What sets LHO apart is the data behind it: nobody else collects this depth of telemetry, combines it with cost data from both Databricks and the cloud, and layers on years of proven recipes for spotting and sizing optimization. That's what turns the conversation from "look what LHO found" into "let's optimize together — here's where to start, it's already surfaced."
A very sophisticated customer with massive in-arrears dashboards still couldn't tell whether an issue was on the cloud side or the Databricks side. With LHO they solved a storage problem they'd fought for months — and previously spent weeks exporting LHO data into spreadsheets and slides to brief leadership, work Executive Insights collapses into an instant summary. Multi-region customers (two Databricks deployments in different regions) use it to see their entire footprint and the top one-to-five areas to focus.
👇 Click to expand and see product screenshots
03 · Cost you can trust
LHO page: Forecasting
There are two halves to a Databricks bill — the Databricks charges and the cloud underneath — and both can carry discounts. LHO shows the true, discount-accurate cost of both: negotiated DBCU plans, cloud savings plans and reservations, enterprise agreements. Whatever you see in the cloud portal, LHO shows exactly that — no separate, un-reconcilable number. And it works the other way too: understand your spend and forecast it reliably, and you're in a position to negotiate a better commitment — or optimize first and negotiate from a leaner, truer baseline.
Cost Plans — "cost you can trust." Two halves of the bill, both discounted, tracked so nothing pre-paid is wasted.
👇 Click to expand and see product screenshots
04 · Catch problems before the bill
LHO page: Cost Over Time → Daily · Incidents
The daily view is where surprises surface. In our example customer's data, most days sat in the low hundreds of dollars — then one day spiked to about $2,200. One click on the spike drilled into that day, and the breakdown showed it was almost entirely Artificial Intelligence (~$1,700) — foundation-model serving that had been spun up for a short experiment. Had that endpoint run for days or weeks, the bill would have been far larger. In the same session, an AI cost-incident policy was configured and wired to the team's email group — so the next time daily AI cost jumps, the right people hear about it within hours, not at month-end. Nothing gets stopped; you simply stop being surprised.
Incidents — "catch problems early." Automatic alerts as problems happen, each explained, with the fix attached — before the bill lands.
👇 Click to expand and see product screenshots
05 · Chargeback and accountability
LHO page: Cost Over Time → Breakdown by Tag
Workspaces, jobs, and clusters are how engineers think — but a CFO or a sponsor thinks in divisions, teams, projects, initiatives, acquisitions. Tag your Databricks assets and LHO reports cost by those business entities, and just as importantly surfaces the untagged / orphan spend so you can drive coverage toward 100%. For a roll-up — assets spread across dev, staging, and prod, and across acquired businesses — that's the difference between a bill nobody owns and a cost story leadership trusts.
Every dollar gets an owner. The untagged remainder shows exactly where tagging discipline still needs to land.
👇 Click to expand and see product screenshots
06 · Where the money goes — and proving the architecture
LHO page: Cost Over Time → Feature · Serverless
Break spend down by feature and the architecture shows up plainly: in our example, serverless compute dominated — about 69% of the year, split almost evenly between serverless jobs and pipelines. Serverless is powerful, but it's a black box — you pay the bill and trust the platform, with no cluster internals to tune. So how do you know it's the right call?
LHO's answer is evidence: run a workload on classic infrastructure alongside serverless, and LHO exposes cluster utilization, spillage, parallel execution, and under/over-provisioning — then compares cost per TB and per hour across both.
An architecture decision stops resting on faith. This is where LHO shifts from a dashboard into a decision instrument.
You can't get more from every Databricks dollar until you can see where the dollars go and prove the cheaper way is actually cheaper.
👇 Click to expand and see product screenshots
07 · Know what to change
LHO page: Recommendations · SQL Warehouses
LHO doesn't just show the bill — it hands you a ranked, evidence-backed list of what to change, each item with a root cause and a dollar impact.
In our example most of that lived in SQL warehouses: one warehouse was up 1h38m and idle 1h36 of it, active barely two and a half minutes for 114 fast queries. The fix is the auto-shutdown timeout, and LHO drops the exact API snippet into the recommendation.
There's a bigger, LHO-only win too: BI tools hold ODBC/JDBC connections open, so a warehouse thinks it's busy and never auto-stops — burning money while nobody runs a query.
LHO's SQL Warehouse Auto-Stop watches for actual queries and stops the warehouse itself, with full auditing.
Recommendations — "know what to change." Not just the bill: root cause, the fix, and the dollar impact, ranked.
👇 Click to expand and see product screenshots
08 · Stop paying twice — consolidation and governance
LHO page: Vendor Consolidation · Unity Catalog Migration
The single biggest lever on total cost of ownership isn't shaving a warehouse — it's not paying for the same thing twice. Duplicated storage, cross-vendor data movement, parallel pipelines, and multiple platforms each needing their own team all inflate TCO.
A Databricks-centric architecture delivers the lowest TCO and highest ROI available — and LHO is what makes consolidation evidence-based: it shows the third-party data footprint of each Databricks workload and ranks which are worth migrating, so a customer consolidates the right things, not blindly.
Governance is the same discipline: LHO's Unity Catalog assessment finds the leftover Hive/legacy footprint and gives prescriptive steps to finish — because plenty of teams believe they're fully migrated and aren't.
On a consolidated estate Blueprint has seen total cost of ownership drop by more than half — operating expense from ~$4M to $1.8M, with 4× price-performance — retiring legacy platforms onto Databricks.
Vendor Consolidation — "stop paying twice." See the third-party footprint and a ranked shortlist of what to move onto Databricks.
👇 Click to expand and see product screenshots
09 · Always-on assurance
LHO page: Incidents · Executive Insights
Underneath every pillar is continuous monitoring. As long as the instance is up, LHO watches cost, performance, and orchestration around the clock — weekends included — and fires the notifications you've configured whenever something leaves normal conditions.
In our example it showed its worth in an unglamorous way: the customer's instance had been shut down over a weekend, and LHO's own monitoring caught it. This is the Govern stage of Augmented FinOps — thresholds, alerts, and policy guardrails that prevent overruns rather than explaining them afterward.
👇 Click to expand and see product screenshots
10 · Proof it worked
LHO page: Optimization Review - Jobs
Optimization only counts if you can prove it. Optimization Review compares any two periods — a single job, all jobs, or prod only — and shows how cost, data usage, and the KPIs are trending, with every configuration change on a timeline.
After a change you can ask plainly: did this help, or make things worse? Even for a serverless-heavy estate it still tells you whether cost is quietly creeping up on your jobs, and where.
This is the loop that closes Augmented FinOps: Observe → Analyze → Forecast → Optimize → Govern, then back to Observe with evidence in hand.
👇 Click to expand and see product screenshots
11 · The business case
LHO page: Executive Insights
Most people look at a Databricks invoice and think that's the cost. It isn't — it's one line in a much bigger total cost of ownership: compute, storage, cloud infrastructure, data engineering, operations, and people. LHO makes the Databricks and cloud slices visible, trustworthy, and smaller, trims the operations slice, and — through consolidation — attacks the biggest lever of all. And as AI and LLM workloads scale, more cost lands on the platform, so cost discipline becomes inseparable from AI value. Blueprint cites the Gartner view that "FinOps is critical to maximizing the ROI of AI agents." LHO is the Augmented FinOps layer that keeps an AI-scaling platform efficient.
For our example customer, a first look delivered the proof that matters at first contact: the team investigated a real cost spike, understood it, and left with a live guardrail — value inside the first hour, and a monthly check-in cadence to build on. That's the pattern: first look, first win, then a relationship.
IGS Energy — 83% Databricks cost optimization ·
Vālenz Health — $140,000 annual savings, live in ~2 weeks · a content-management provider — 35% lower annual cloud spend, $45k saved shutting one idle cluster, a memory leak caught 8 days before it would have crashed prod · consolidation — >50% lower TCO ($4M → $1.8M, 4× price-performance).
Save 30% on total cost of ownership and improve performance by 50%.
Executive Insights is the leadership view of your Databricks estate — one page with a health score and the top few places to focus, across all your workspaces, powered by the telemetry and cost data LHO already collects, with Genie so you can just ask in plain English. Executives see where to focus; their teams click straight into the fix.
👇 Click to expand and see product screenshots
See it on your own numbers. Explore the Lakehouse Optimizer 👇