Aquaculture Software Development

A fish or shrimp crop is alive, growing, and impossible to count directly. When stocking, feeding, mortality, and water readings live in separate spreadsheets and notebooks, every number the farm plans on is already drifting. We build the system that holds one record per pond, from stocking to harvest.

  • One batch record per pond, tank, or cage, from stocking through grading to harvest

  • Feeding plans and feed conversion worked out per pond, in kilos and in money

  • Water quality readings tied to the batch in the water, with alerts when a reading crosses the line you set

  • Treatment and withdrawal dates the harvest cannot ignore

3+ years RetentionWeek 1 First milestoneFixed price Pricing

The problem

Sound familiar?

  • Feeding in one spreadsheet, mortality in a notebook, and water readings on a separate sensor screen, so no two people can agree on what is actually in a pond?

  • Finding out at harvest that a pond lost money, because feed cost, deaths, and treatments were never held against the same batch?

Short answer

RaftLabs builds custom aquaculture software for pond farms, shrimp farms, hatcheries, and tank facilities: one batch record from stocking to harvest, feeding plans with feed conversion per pond, mortality and treatment records with withdrawal dates, water quality trends and alerts, and cost per pond, scoped and priced before work starts.

Who this is for

Aquaculture teams we build for

Pond fish farms

Tilapia, carp, catfish, and similar species in ponds, with a record per pond instead of a notebook per worker.

Shrimp farms

Pond-by-pond stocking, feed tables by weight band, and mortality tracked across a cluster of ponds.

Hatcheries

Broodstock, spawning, and nursery phases, with the splitting and grading work that fills a hatchery week.

Tank and RAS facilities

Tank-level records with continuous water readings, where dense stocking leaves no room for a stale number.

01 Diagnosis

Problems we solve in aquaculture software

  1. 01
    Problem

    The production record is split across four places

    Cost

    On most farms the record of a crop lives in a feeding spreadsheet, a mortality notebook, a sensor screen, and the memory of whoever has worked there longest. Each one looks right on its own. Together they disagree, and nobody notices until someone needs a number. A split moves fish between two ponds, the feeding sheet gets updated, the stocking sheet does not, and every feeding plan and cost figure built on that number stays wrong until harvest. The cost shows up late: a bank, insurer, or buyer asks for a pond report and it takes a day of pulling files together and hoping the totals match.

    Solution

    One batch record per growing unit, where stocking, splits, feeding, mortality, and sampling are single events that update every figure they touch, means the next feeding plan and the next report are built on numbers that agree. The pond report becomes a page you open, not a day of work.

  2. 02
    Problem

    Feed conversion is a guess until the cycle ends

    Cost

    Feed is the largest cost on the farm, and it is committed weeks before the revenue it produces arrives. Feed conversion, the kilos of feed behind each kilo of growth, can only be worked out if feed issued and weight gained are recorded against the same pond over the same period. On most farms the feed is in one place and the sampling is in another, so the ratio is estimated at the end of the cycle, if it is worked out at all. The ponds that waste feed look the same as the ponds that convert well, and the farm keeps feeding both the same way.

    Solution

    Feed recorded per pond from the lot it came from, with sample weights against the same pond, turns feed conversion into a running figure in kilos and in money. The farm can see which ponds convert and which ones waste feed while there is still time in the cycle to act on it.

  3. 03
    Problem

    Water readings are not tied to the fish they affect

    Cost

    Dissolved oxygen, pH, and ammonia decide whether a pond thrives or crashes, and oxygen can fall fast overnight or after feeding. When readings live on a sensor screen that does not know which batch is in the water, a bad night shows up as dead fish in the morning and a timeline someone rebuilds by hand days later, from two systems, from memory. By then the pattern that caused it, the feeding before it, the weather around it, is gone, and the same crash can happen again in the next pond.

    Solution

    Readings stored against the growing unit and the batch in it, with alerts when a reading crosses the line set for that species and stage, put the oxygen curve and the mortality that followed it on the same page. The farm sees the fall while it can still respond, and the cause stays on record for the next cycle.

  4. 04
    Problem

    Treatments and withdrawal dates live on a whiteboard

    Cost

    When fish or shrimp are treated, they cannot be harvested or sold until the withdrawal period clears. On most farms that date lives on a whiteboard, in a notebook, or in one person's head. Ponds get split, batches get mixed, and the date does not travel with the fish. The risk is not a fine on paper. It is treated stock reaching a buyer, a consignment rejected, and a customer relationship the farm spent seasons earning damaged in one delivery.

    Solution

    A treatment recorded once, with its withdrawal period enforced on every harvest and sale for that batch, and the clock following the fish when a pond is split, turns a memory into a rule the system checks. A traceability question from a buyer gets a straight answer from the record.

02 What the system covers

Software for each part of the production cycle

  1. Batch and pond records

    One record per pond, tank, or cage, from stocking to harvest: placements, splits, transfers, merges, and sampling, with the history of where each batch came from carried forward. The standing stock is shown as what it is, an estimate built from recorded events with its last sample date, not a hard count the data cannot support.

  2. Feeding plans and feed conversion

    A daily feeding plan worked out from your feed table, by species, weight band, and stage of the cycle. What was actually fed is confirmed at the pond, skipped days are marked, and feed conversion is reported per pond in kilos and in feed cost per kilo gained, priced from the feed lot used.

  3. Mortality, treatments, and withdrawal

    Deaths recorded per unit as they are found, with the likely cause noted where your team records one. Treatments carry their withdrawal period with the batch, and the system holds harvest and sale until the period clears, including after a split or transfer.

  4. Water quality trends and alerts

    Dissolved oxygen, pH, ammonia, and temperature readings held against the growing unit they came from, whether they arrive from fixed sensors or a probe at the pond side. Trends sit next to the production record for the same batch, and alerts go to the people on duty when a reading crosses the threshold you set.

  5. Harvest, grading, and settlement

    Harvest weights and size grades captured at the pond side instead of estimated later, tied back to the batch they came from. Buyer settlement is worked out from measured weights, so the yield of a cycle can be compared pond by pond and season by season.

  6. Cost per pond and reporting

    Feed, seed, treatment, labour, and energy costs held against the pond that used them, so margin per pond and cost per kilo harvested are reports, not reconstructions. Period reports for owners, banks, and insurers export cleanly, and a closed month stays closed once it is signed off.

Integrations

Connected to the pond and the office

Sensors, probes, and accounting connect into one record, because a reading that is not tied to a batch helps nobody.

Water data
Dissolved oxygen sensors · pH and ammonia probes · Temperature and level sensors
Farm systems
Accounting integration · ERP integration · Excel and PDF reporting exports
A discovery phase scopes the sensor and system connections first; they drive most of the cost variance.

03 Buy, rent, or own

Stay on the platform while your species, stages, and sites fit what it assumes.

The aquaculture software question is really about whether your production model fits the product, or the product has started deciding how you farm.

Buy or rent

  • An aquaculture platform, while your operation fits its model

    Faster to start and sound when your species, stages, feed tables, and site structure match what the product assumes.

  • Spreadsheets, while one person can still hold the farm

    Keep the sheets while the farm is small, the maintainer is not a single point of failure, and reports are rare.

Own custom software

  • When your species or stages do not fit the product model

    Your team records the real operation in side notes because the system has no place for it.

  • When sensors and accounting need to meet in one record

    Readings, feed lots, and costs only become useful when they land on the same batch.

  • When the production record is an asset you want to own

    The history of every pond and cycle stays in systems you control, with no per-seat lock-in.

Bottom line

Stay on the platform while your operation fits its model and the fees stay reasonable. Own the system when the workarounds, the disconnected sensors, or the reporting burden cost more than running your own record.

04 How we work

How an aquaculture software project runs

  1. Step 01
    01

    Discovery and connection scoping

    Two to three weeks mapping how your farm actually runs: the species and stages you work through, the units you stock, your feed tables, and every sensor, probe, and office system the software has to talk to. This step surfaces the connection requirements that cause rework when they are found halfway through a project.

  2. Step 02
    02

    Data model around batches and units

    We design the record around growing units and batches, the objects your team actually works in, and the connection layer for sensor feeds and accounting. The riskiest connection, usually a sensor feed or a feed-lot costing rule, is tested with real farm data early, because it is the most common cause of a timeline slip.

  3. Step 03
    03

    Working software in two-week cycles

    You review working software every two weeks, not wireframes. Each cycle delivers a usable slice: the batch record, then feeding, then water quality and alerts, then cost reporting. Scope is fixed and priced before the work starts, so the number does not move unless you change what you are asking for.

  4. Step 04
    04

    Testing on the farm and handover

    We test pond-side entry and sensor connections under your real site conditions, including ponds with weak or no signal, because a record with gaps fails exactly when it matters. You get a documented, tested system, your data in your control, and a handover so your team or ours can run it.

05 Track record

What we have shipped in operations software

Countries reached on a multi-carrier shipping platform we shipped
70+
Shipments processed in year one on that platform
2,000+
Customers migrated to it with zero downtime
200+
Fixed-cost rebuild, scoped up front
14 wk

08 Why us

Why choose us?

  • 01

    The record is designed around your farm

    Species, stages, unit types, and feed tables are set up for your operation, not picked from a fixed list written for someone else's farm.
  • 02

    Connections are scoped before the price

    Sensor feeds and accounting connections are mapped in discovery, so the fixed price covers the part that usually breaks an aquaculture project.
  • 03

    Fixed cost before work starts

    Scope is agreed and priced before development, and you review working software every two weeks, so surprises surface while they are still cheap to fix.

09 Questions

Common questions

Aquaculture software runs a fish, shrimp, or shellfish operation day to day. It holds batch and pond records from stocking to harvest, feeding plans and feed conversion, mortality and treatment records, water quality monitoring, harvest and grading, and cost per pond. It differs from crop farm software because the stock is alive, growing, and estimated rather than counted. It differs from a sensor dashboard because readings only matter when they are tied to the batch in the water.

Yes, and it should say so openly. The sound approach is to store the recorded events, stocking, sampling, mortality, splits, and feed, and work the standing stock out from them, shown as an estimate with its last sample date. Software that presents that estimate as a hard inventory number encourages decisions the data cannot support, and the gap only appears at harvest.

No. Existing sensors and probes usually stay in service. We scope what each device can send during discovery, connect the ones that expose their readings, and keep manual pond-side entry for the rest, so a reading lands on the same batch record either way. Replacing working hardware is rarely the right first spend.

A focused single-system build typically takes 10 to 16 weeks. A multi-site platform with sensor connections and accounting integration typically runs 4 to 6 months. We scope every project before pricing. Fixed cost, agreed before work starts. A discovery phase scopes the connections, because they drive most of the variance.

No. We often start with the part that hurts most, the batch record or the feeding and cost report, and connect or replace the rest in stages. If a full replacement is the right call, we scope it honestly, but the default is to fit what you already run rather than force a farm-wide change in the middle of a cycle.

Get a buy-or-own plan for your farm.

Tell us how many ponds or tanks you run, which species you stock, and where the record breaks down today. We will tell you how we would approach it.

  • Scope and cost agreed before work starts. No surprises. No obligation.
  • Working prototype within 3 weeks of kickoff.
  • Pay by milestone. You see progress before each invoice.
  • 60-day post-launch warranty. Bug fixes, UI tweaks, and deployment support. No retainer.
  • All conversations are NDA-protected.