Join us at Dreamforce
Salesforce intelligence

Understand what you
inherited & built.

Salesforce orgs evolve for years, across changing teams, vendors and decisions. MetaLens360 analyses your metadata, code, dependencies, data and analytics, then explains what you inherited and built, how it all fits together, and where the risk sits — with the evidence behind every insight.

READING ORG · PRODUCTION
0components indexed · one org
0dependency edges resolved
0CRM Analytics assets parsed
0fields traced through lineage

MEASURED — every figure above was read from a live connected Salesforce org. Nothing on this page is an illustrative number, and where a value is unknown we say so rather than round it to zero.

The gap

Seven questions your org
cannot answer about itself.

Deployment tools move metadata between environments. None of them tell you what the metadata means, what it touches, or what happens if you change it.

MetaLens360 answers these — and tells you which answer it measured, which it derived, and which it could not establish at all.

What makes it different

Every answer carries
its evidence.

An intelligence tool that cannot distinguish measurement from guesswork is a tool you have to double-check — which means it saved you nothing. MetaLens360 grades every claim it makes.

Observed

Measured directly

Read from the org, the repository, or the platform's own APIs. A row count, a last-run timestamp, a declared foreign key.

Derived

Computed from what was observed

A dependency found by parsing Apex. A blast radius walked across the graph. Reproducible, and shown with its inputs.

Inferred

Pattern, not proof

A naming convention that usually means a reference. Labelled as inference so you weigh it accordingly.

Not established

Reported, never dropped

A rule that could not run. An asset the integration user cannot read. Absence is stated, because silence reads as safety.

Unknown never becomes zero.

This is the principle the product is built on. When a check cannot run, MetaLens360 says so rather than returning a clean result — because "we found no problems" and "we could not look" lead to opposite decisions, and only one of them is safe to act on.

The model

Nine sources.
One connected picture.

Intelligence comes from the joins, not the inventory. A dependency matters because it crosses a boundary; a change matters because of what sits downstream of it.

Connected

MetadataApex & LWCDependencies ArchitectureIntegrationsData volumes CRM AnalyticsEnvironmentsChange history

Produced

UnderstandingImpact analysisRisk detection Architecture intelligenceTechnical debt RecommendationsAI insights
Platform

Fourteen capabilities,
one connected model.

Grouped by the question each one answers rather than by how the product is built. Every one reads the same graph, so an answer in one is consistent with an answer in another.

AI, grounded

It explains your org,
not Salesforce in general.

Generic Salesforce advice is free and worthless. MetaLens360 AI reads the assets, relationships and lineage discovered in your org — and states what it could not see.

Evidence supplied to the model recipe: Account_Revenue_Recipe
sources: Opportunity (sObject) · Account (sObject) · FX_Rates (csv)
joins: 3 · computed fields: 11 · output: Account_Revenue_DS
consumed_by: 4 dashboards · 1 downstream recipe
last_run: 2026-08-21 02:14 · status: success · rows: 1,284,902
unreadable: 0
AI · from your metadata 6 observed inputs

Account_Revenue_Recipe converts opportunity revenue into account-level totals, normalising currency through an uploaded FX table rather than the org's own conversion rates — so its figures will diverge from standard reports whenever the uploaded rates are stale.

It is load-bearing: four dashboards and one downstream recipe read its output. The FX source is the weak link — a manual upload with no refresh schedule.

What it writes

Functional and technical summaries, architecture and dependency explanations, impact and root-cause analysis, technical-debt findings, CRM Analytics summaries, and recommendations drawn only from assets that exist.

Your credentials, your spend

Bring your own provider keys — eleven supported, with automatic failover. Choose whether a workspace shares one enterprise credential or requires each person to bring their own. Usage is metered per person either way.

How the AI layer works →

It declines to guess

When the evidence is thin, the answer says so. A summary generated from a partial retrieve is labelled as such rather than presented with the same confidence as a complete one.

CRM Analytics

The part nobody else reads.

Wave metadata is opaque, deeply nested, and usually undocumented. MetaLens360 parses recipes, dataflows, datasets, dashboards and lenses into a lineage you can follow end to end.

Sources & targets — traced, not listed
SourceKindTransformTarget datasetConsumed byEvidence
OpportunitysObjectAccount_Revenue_RecipeAccount_Revenue_DS4 dashboardsobserved
FX_Rates.csvManual uploadAccount_Revenue_RecipeAccount_Revenue_DS4 dashboardsobserved
CasesObjectService_Health_DataflowService_Health_DS2 dashboardsobserved
Account_Revenue_DSDatasetExec_Rollup_RecipeExec_Summary_DS1 dashboardderived from SAQL
Legacy_Pipeline_DSDatasetnothingobserved
Territory_ExtExternalnot readableunknownunknownaccess denied

Lineage you can walk

Move upstream and downstream from any asset. Field-level where the transforms allow it.

KPIs as actually coded

Measures extracted from SAQL and compiled aggregates — what the org computes, not what a data dictionary claims.

Disposition, with a brake

Keep, fix, consolidate, retire or build — and nothing is marked for retirement while something still reads it.

Who it is for

One model, read nine ways.

Enterprise

Built to be handed
to a security review.

Read-only by design

MetaLens360 reads. It does not deploy, modify, or write back to your orgs.

Credentials encrypted at rest

AES-256-GCM with a key held in a managed secret store, never in the database it protects.

Workspace isolation

Every record is scoped to a workspace, and reads are filtered by it — including for administrators.

Roles and permissions

Per-workspace role definitions over a permission model, not a fixed set of tiers.

AI access controls

Decide whether AI runs on shared enterprise credentials or requires each person to bring their own.

Full call auditability

Every AI request, its provider path, and its cost are recorded — and nothing is silently pruned.

Start

Point it at one org and see what comes back.

Connect read-only, let the first sync run, and read the org overview it produces. If it tells you something you did not already know, keep going.

Platform

Everything reads
one connected model.

Fourteen capabilities — listed in full below — all read the same graph of your metadata, code, data and change history. That shared foundation is the point: a dependency shown in impact analysis is the same dependency the architecture view drew, so two teams looking at the same org never end up arguing from different numbers.

How it works

Four stages, each inspectable.

Stage 1

Connect

OAuth to each org, read-only. Optionally connect the repositories that hold your source of truth.

Stage 2

Retrieve

Scheduled or on demand. Every run records what was requested, what arrived, and what failed — so a partial retrieve is never mistaken for a complete one.

Stage 3

Resolve

Parse code and configuration into a dependency graph, each edge labelled with how it was established.

Stage 4

Explain

Rules, comparisons and AI read the resolved model — never the raw dump — so every output traces back to a source.

Also included

The operational surface.

Sync & schedule center

Durable schedules, concurrency limits, per-org history, and the reason any run fell short.

Diagnostics console

Live logs, operation timelines, and AI-assisted log analysis for when a sync misbehaves.

Data model browser

The platform's own schema, read live — every table, column and relationship, labelled enforced or by convention.

CRM Analytics intelligence

Wave metadata,
made legible.

CRM Analytics is where Salesforce estates go undocumented. Recipes nest transformations several layers deep, dashboards bind to datasets by name, and the lineage exists only in the head of whoever built it — if they are still there.

What it reads

Recipes, dataflows, datasets, dashboards, lenses, templates and XMD — parsed from the real payloads rather than the documented schema, because the two differ.

What it produces

A dependency graph across the whole estate, field-level lineage where transforms permit, a KPI dictionary read from SAQL, health findings, and a disposition per asset.

Thirteen views

From inventory to decision.

Architecture

The estate as a real dependency graph, not a component count.

Sources & targets

Every input traced to every output it reaches.

Recipes

Parsed transformations, joins, and computed fields — with the source visible.

Fields

Field-level lineage across three confidence tiers.

KPIs

Measures as actually coded in SAQL and compiled aggregates.

Lineage

Walk upstream and downstream from any asset, with history.

Health

Rules with earned severity — and the checks that could not run, listed.

Disposition

Keep, fix, consolidate, retire, build — gated on what is load-bearing.

Assessment

A generated implementation review grounded in the discovered estate.

Worked example

Metadata in. Explanation out.

The same recipe, shown as the platform sees it and as it reads back to a human.

Parsed from the recipe body nodes: 14 · joins: 3 · filters: 6 · computed: 11
sources: Opportunity, Account, FX_Rates.csv
output: Account_Revenue_DS (1,284,902 rows)
downstream: 4 dashboards, Exec_Rollup_Recipe
schedule: daily 02:00 · last 30 runs: 30 success
AI summary grounded in 5 observed facts FX refresh cadence unknown

Functional. Produces account-level revenue for executive reporting, converting opportunity amounts to a single reporting currency before rolling them up.

Technical. Fourteen nodes, three joins, eleven computed fields. Currency conversion uses an uploaded rate table rather than the org's conversion rates, which is the single largest source of divergence from standard reports.

Risk. The FX table is a manual upload with no discoverable refresh schedule. Four dashboards depend on this output, so a stale upload propagates silently.

The AI layer

Grounded, graded, and paid for
by whoever asked.

MetaLens360 AI is not a chat window bolted to a product. It reads the resolved model — the same graph every other surface reads — and writes explanations that cite what they were built from.

It reads your org, not the internet

Every prompt is assembled from assets discovered in your environment: component bodies, dependency edges, run history, row counts, drift results. No generic Salesforce advice.

It states its own limits

An explanation built from a partial retrieve says so. An asset the integration user could not read is named as unreadable rather than omitted from the picture.

Provider architecture

Bring your own keys.

Eleven providers

Configure any subset. The chain tries them in your order and fails over automatically.

Adaptive routing

Ranking adjusts on live evidence — rate-limit windows, recent failure rates, latency — and every routing decision is explained.

Full call traces

Prompt, response, provider path, and cost for every call. Nothing pruned.

Shared credentials

Everyone in the workspace generates on the enterprise key. Nobody configures anything. Usage is still recorded per person, so spend stays attributable.

Personal credentials

Generating requires a key the individual owns. Nothing is inherited from an administrator or a colleague, so every token traces to whoever spent it.

Reading is never gated. Under either setting, everything already generated stays visible to the whole workspace. The setting decides only whose credential is spent producing something new.
What it writes

Nine kinds of generated intelligence.

Functional summaries

What the org does, in business language, from what it contains.

Technical summaries

How it is built, where the complexity sits, and what that costs.

Architecture explanations

Layers, clusters, and where coupling concentrates.

Dependency explanations

Why this component touches that one, and through what.

Impact analysis

What a proposed change reaches, and how confidently.

Root-cause analysis

Applied to debug logs, failed runs, and drift.

Technical-debt insights

Dead code, duplicated logic, and automation nobody owns.

CRMA implementation summaries

What the analytics estate actually computes.

Recommendations

Drawn only from assets that exist — never a generic best-practice list.

Solutions

One model,
nine different jobs.

Each role arrives with a different question. They resolve against one graph, which is why the answers do not contradict each other across teams.

Situations

Where it earns its place.

Inheriting an org nobody documented

The original team has moved on. Start from a generated functional and technical summary, then walk the dependency graph outward from whatever you were asked to change.

Planning a migration or consolidation

Establish what actually exists across environments, what is genuinely used, and what is safe to leave behind — with the evidence attached.

Reducing release risk

See what drifted between environments and what sits downstream of each change before it ships.

Cleaning up technical debt

Find dead code, orphaned automation and unread datasets — with a brake that refuses to mark anything load-bearing for removal.

Standing up a Center of Excellence

Compare orgs against each other, track drift over time, and give every team the same picture.

Scoping a consulting engagement

Produce an evidence-based assessment of an unfamiliar org in hours rather than weeks.

Trust

What MetaLens360 does
with your data.

Written to be read by a security reviewer, not a buyer.

Access model
Read-only OAuth per org. MetaLens360 retrieves metadata and counts; it does not deploy, modify, or write back.
Credential storage
Refresh tokens and provider keys are encrypted at rest with AES-256-GCM. The encryption key is held in a managed secret store, never in the database it protects.
Tenancy
Every record carries a workspace identifier and reads are filtered by it. Cross-workspace visibility is an explicit, separately controlled capability.
RBAC
Permission-based, with per-workspace role definitions. A workspace can redefine what its own roles may do without a code change.
SSO
Supported. Password credentials are optional and are stored only as a hash.
Audit
Actions are recorded with actor and timestamp. Users are deactivated rather than deleted so audit records keep resolving to a real person.
AI data handling
Prompts are assembled from your metadata and sent to the provider you configured, under your own credential. MetaLens360 does not train models on your data.
AI access control
Per-workspace: shared enterprise credentials, or each person supplies their own. Access is never inherited from an administrator.
AI auditability
Every call is recorded with its prompt, response, provider path, token counts and cost. Records are not pruned.
Multi-org
Any number of connected orgs and repositories, compared and tracked independently or together.
Ask us for our current attestations. We list a compliance framework once it is certified, never while it is in progress — the same standard we hold the product to, where nothing is reported as established until it has been. Write to security@metalens360.io for the current status of any specific framework, or to request our security documentation.
Pricing

Scoped to your estate,
not to seats alone.

What drives cost is the size of what you connect — orgs, repositories, and how much metadata they hold — rather than how many people read the results.

Evaluation

Connect one org

Run a full sync, read the generated overview, and walk the dependency graph. Enough to judge whether it tells you something you did not know.

Start
Team

Multiple orgs, scheduled

Continuous sync across environments, drift tracking, repository intelligence, and the full AI surface on your own provider keys.

Talk to us
Enterprise

Multi-workspace

Workspace isolation, SSO, per-workspace roles, per-user AI credential enforcement, and cross-org administration.

Talk to us
We price against your estate, not a list. The number that matters depends on how many orgs and repositories you connect and how much metadata they hold, so we quote once we know — rather than publish a figure we would immediately have to qualify. AI usage is billed by your own provider directly to you: MetaLens360 does not resell inference or mark it up.
Get started

Four steps to
the first answer.

Nothing is deployed and nothing is modified. The connection is read-only throughout.

Step 1

Connect an org

Browser OAuth, or an SFDX auth URL. A refresh token is required — pasted access tokens expire within hours.

Step 2

Run the first sync

Retrieval and indexing. The run reports what arrived and what did not.

Step 3

Add your AI key

Any of eleven providers. Optional — the platform is fully usable without AI.

Step 4

Read the overview

Start with the generated org summary, then follow the dependency graph outward.

Talk to us

Tell us how many orgs you run and what you are trying to find out, and we will scope from there. Evaluations start with a single read-only connection.

hello@metalens360.io