Inventory & purchasing for research labs

Your stock record says you have enough. Next week's experiments disagree.

Aliquot counts what is on the shelf, reads what your booked protocols will consume, and tells you what to order before the freezer runs dry — not after.

Orders are prepared for a human to send. Nothing is transmitted automatically.

DMEM · high glucose · 500 mL D5796
CONSUMPTION HISTORY No action

43 days of cover

10 bottles on hand, averaging 0.23 per day over the last 90 days.


BOOKED PROTOCOL RUNS Order today

Short by 6 bottles

6 runs scheduled in the next 14 days need 16 bottles. Supplier lead time is 12 days.


Every inventory tool shows you the first number. The second one is why an experiment gets cancelled on a Tuesday morning.

Reorder review
25 low items, batched across 8 suppliers
Counted without asking
9 movements logged from one protocol run
Grant clawback risk
$107k projected unspent on an R01, 81% elapsed
Fire-code check
8.0 L flammables in a cabinet limited to 4.0 L

The forecast

A burn rate is a rear-view mirror. Your calendar is not.

Move the controls. The bar on the left is what history predicts; the marker is when a reorder would actually arrive. Booked work is what turns a comfortable number into a cancelled experiment.

Runway simulator live
What history predicts 43 days of cover
What the calendar demands 16 of 10 bottles

What it does

Six jobs, all fed by the same count.

Inventory that only a diligent person keeps accurate is inventory that drifts. Every feature here writes back to the same lot-level record, so the count stays honest without anyone remembering to log anything.

Forecasting that warns early

Burn rate from real movements, plus demand from booked protocol runs, measured against each vendor's actual lead time.

Lot-level · FEFO

The last mile of purchasing

Low items batched by supplier to clear free-shipping thresholds, then turned into a requisition, a cart hand-off, or a drafted email.

3 order modes

Protocols as bills of materials

Mark a run complete and the reagents come off the shelf at the right volumes. Paste an SOP and the wizard parses the quantities and converts the units.

Auto-decrement

Freezer box maps

Every vial has a position. Find the one you need without holding the door open, and see which containers were never filed.

Grid positions

Chemical compliance

Incompatible neighbours flagged by hazard class, control-area volumes rolled up the location tree, and a CSV your EHS office will accept.

Fire-code totals

Grant burn-down

Spend against each fund, projected to the end date. Underspending an R01 is as much of a problem as overspending it — and it shows up months earlier.

Per-fund

Where things physically are

A 9×9 box, and the vial you are actually looking for.

Select a position to read its contents. Filled positions carry a lot number and an expiry; the empty ones are where the next vial goes.

Stocked Last vial Expired Empty
−80 °C · Freezer 2 · Rack C · Box 4

Compliance

The inspection finding you would rather write yourself.

Put containers on a shelf and see what an EHS officer would see. Segregation conflicts come from hazard class; the volume total is checked against the control area's configured limit.

Flammables cabinet · Bench B limit 4.0 L
Findings

Flammable volume in this control area

0.0 L of 4.0 L


Deliberate constraints

The things it will not do, on purpose.

Several of these look like missing features. They are load-bearing. A tool that spends money on its own, or overwrites a number a person is accountable for, is a tool a lab stops trusting after the first mistake.

Purchasing

Automation prepares orders. It never transmits them.

Every path ends at a requisition, a cart, or a draft email waiting for a person. The system never places an order on your behalf.

Data integrity

A quantity a machine read is a draft until a person agrees.

Parsed CSVs, scanned confirmations and imported protocols land as proposals with the source shown, not as silent writes to your stock record.

Safety

Fire-code limits are configured, never computed.

Your control-area limits come from your institution's own permits. Guessing them from a hazard table would produce a number nobody can defend.

Approvals

Approval requires a second person.

The requester cannot approve their own requisition over the threshold. Separation of duties is enforced in the authorization layer, not the UI.

Matching

Material matching stays conservative.

A near-match that silently binds the wrong reagent is worse than no match. Ambiguity is surfaced for a human rather than resolved by guesswork.

Budgets

Pending requests are excluded from fund totals.

A grant balance reflects committed spend. Counting requests nobody has approved yet would make every burn-down chart optimistic fiction.

Verification

What has been driven end to end.

Aliquot is pre-launch and has no customers yet, so there is nothing to quote you. This is the honest substitute: the behaviour that was exercised in a real browser or against a real database, and how.

Capabilities and how each was verified
CapabilityHow it was verified
FEFO consumption and lot maths8 integration tests against a real database
One-click order, all three modesBrowser: requisition YL-2026-0002 created, held for approval over the $500 threshold
Supplier batching25 low items grouped across 8 suppliers with the free-shipping gap shown
Protocol run writes back to stockBrowser: DMEM 2→1, DPBS 13→12.98, 9 movements logged, $91.62 of consumption costed
Protocol import wizardParsed a 6-step SOP, converted units, flagged the one line it could not convert
Barcode scanningBrowser: looked up T8787, took one, put it back; USB scanner Enter key handled
Messy CSV import9 awkward headers matched, including Cat #, Amount, on hand and Price ea.
Order confirmation parsingThermo tabular and Sigma labelled-block email formats, both extracted in full
Chemical segregationFound HCl beside NaOH, ethanol beside nitric acid, and 8.0 L against a 4.0 L limit
Grant burn-downAn NIH R01 at 55% spent and 81% elapsed projected roughly $107k unspent
Accessibility controlsHigh contrast, colour-safe palette, reduced motion and 120% text all applied and persisted
Test suite
82 passing, across ~19 suites
Data model
23 Prisma models, Postgres-portable
Bugs caught pre-release
7 including a prototype-pollution hole in the permission check
Tenancy
Multi workspace scope and permission checked together, never apart

Early access

Bring one freezer and one grant. That is enough to see it work.

Aliquot is onboarding a small number of labs by hand so the first counts are set up correctly. Tell us what you stock and how you buy it.