7
Optimization Opportunities Identified
35
Upgrade Risk Inventory
$10,650
Combined services pipeline savings identified
AcuLens reads how a system is actually used - screens touched, roles active, customizations exercised - and turns it into decision-ready reports for developers, account teams, and management, without a human manually assembling the data. Aistrium built it first for Acumatica ERP, and the same engineering approach can be built for any system with activity logs, role-based access, and a customization history, including banking core platforms.
Every core business system builds up the same blind spot as it ages. Which screens carry real work, which customizations are dead weight, and which habits would break if the system changed? That knowledge lives in people's heads, and it's different at every installation.
The blind spot costs money long before an upgrade. Teams pay for capability they never adopted, build manual workarounds for things the system already does, and depend on one person for processes nobody else knows. When an upgrade does arrive, test scope gets set in meetings, from memory.
The usual fixes don't solve it. A consultant interviewing staff for days collects what people remember. A generic analytics tool logs raw clicks without knowing what they mean for the business.
The usual way to answer these questions is a consultant sitting with a client for days, asking what people remember doing, or a generic analytics tool that logs raw activity without understanding what any of it means for the business. Neither approach scales, and both mean guessing at exactly the problem an AI Agent is built to take on directly.
Learns the installation, rather than requiring it to be configured - modules, custom screens, user roles, and thresholds are learned entirely from that instance's own data
Separates judgment from arithmetic - facts, counts, and risk scores are computed with fixed rules every time; AI reasoning is applied only to the narrative explanation, never to the numbers
Protects individual privacy by design - users appear as aliases, small groups are suppressed entirely from per-person reporting, and the identity map never leaves the customer's own instance
Never touches the business itself - no names, balances, customer or vendor records, or document contents are ever recorded or transmitted; only metadata about how the system is used.
The same AI Agent architecture applies to any system with activity logs, role-based access, and a customization history with the stakes rising accordingly: a banking core carries regulatory audit exposure on top of the operational risk any upgrade carries.
Aistrium (former Sprinterra) designed AcuLens to run without anyone on the customer's team configuring it or filling anything in. Deployment is one standard customization package and one dedicated API user.
A lightweight component installed directly on the customer's own deployment that reads navigation and activity logs, which screens were opened, which commands were run, which fields changed, with no performance impact.
Facts, counts, and risk scores are computed with fixed rules every time, so the numbers never depend on an AI's interpretation.
Language-model reasoning applied only to the explanations built on top of those facts - never to the numbers themselves.
Users appear as aliases, groups of three or fewer are suppressed entirely from per-person reporting, and the alias-to-identity map never leaves the customer's own deployment.
Every documented change to the system itself, cross-walked against what's actually used, so the impact of a change is scoped from real usage, not memory.
Upgrade Readiness & Test Scope for developers and QA, Usage Insights & Optimization for the account team, and Workforce & Process Analytics for client management - all generated from the same underlying record.
AcuLens never records or transmits the customer's business data itself: no names, balances, customer or vendor records, document contents, or credentials. It reads metadata about usage with users appearing as aliases, groups of three or fewer suppressed entirely from per-person reporting, and the alias-to-identity map never leaving the customer's own deployment.
it's how Aistrium approaches any system where understanding real usage, not assumptions, is critical to scoping change safely.
AcuLens is built as a multi-way benefit, not a single customer tool. The roles below apply whether the system runs through a vendor-and-partner channel (like ERP) or was built and is maintained directly by one team (like a custom platform built for a single business).
AcuLens records how a system is actually used and turns that record into findings a team can act on. Manual work gets matched to capability the business already owns. Processes that depend on one person get flagged early enough to train a backup. The most frequent errors and slow screens get ranked, each with its fix.
On a mid-size distribution installation, AcuLens found that 11 of 38 custom screens hadn't been opened in twelve months. It also cut the regression test scope for the 2025 R2 upgrade by roughly half.
Behind it is one design rule: fixed rules compute every figure, and AI only explains what the figures mean. Aistrium (former Sprinterra) applies the same approach to other core systems. If nobody on your team fully knows how your system is used, we can build this for yours.
Upgrade Risk Inventory
Effort
A global manufacturer ran engineering change through email chains, with no reliable record of who reviewed what or which active projects a change affected. We built the review, approval, and impact workflow around the ERP instead of inside it, so supplier and component changes now trace automatically to every project they touch.
See the full case study
A lightweight collector records which screens, workflows, customizations, and integrations carry real work, and turns that into a usage-ranked test plan, a risk register, and a performance baseline. Counts and risk scores are computed with fixed rules every time, so AI reasoning shapes the explanation and never the numbers.
See the full case study
Lauren B Jewelry runs client records, inventory, work orders, appointments, invoicing, and financial reporting through LB.NET, an ERP we built from scratch and still maintain. Its hardest rule: one-of-a-kind stock priced correctly at every moment across the sales floor and the Magento storefront, with a sale on either channel locking the item everywhere else instantly
See the full case studyThirty minutes, no pressure, just a conversation about what’s next.
Fill in the details and we'll be in touch within one business day.
CASE STUDY · AI-ACCELERATED USAGE INTELLIGENCE
