Esta página ainda não está disponível em português, por isso o texto abaixo está em inglês. O menu, os formulários e o agendamento da demo estão em português — e a demo pode ser feita em português.
Modules
13 modules. One runs today.
Every module below is listed with the state it is actually in, taken from the same registry the application reads. Nothing here is a roadmap item dressed as a feature: where a module is not switched on, the reason it is not switched on is printed under it, in the words the engineer who stopped it wrote.
| Module | State | Price |
|---|---|---|
| Control Tower | Running in production | Included in every plan |
| DeadStock AI | Beta — live for pilot tenants | ₹25,000 / month |
| SupplierGuard | Beta — live for pilot tenants | ₹25,000 / month |
| EXIMGuard | In build — not switched on | ₹30,000 / month |
| LandedCost AI | Beta — live for pilot tenants | ₹20,000 / month |
| FreightAudit AI | Beta — live for pilot tenants | ₹20,000 / month |
| StockOpt AI | Beta — live for pilot tenants | ₹20,000 / month |
| DG Compliance | In build — not switched on | ₹30,000 / month |
| DocumentGuard | Beta — live for pilot tenants | ₹15,000 / month |
| Pending BL | Beta — live for pilot tenants | ₹15,000 / month |
| OTIF Post-Mortem | In build — not switched on | ₹25,000 / month |
| SIAARU CRM | In build — not switched on | ₹20,000 / month |
| SIAARU HR | In build — not switched on | ₹15,000 / month |
Prices are per organisation and exclude GST. The eight supply-chain add-ons can be bought together — see pricing for the bundle and for what the bundle does not yet include. Modules in build are not billed and cannot be switched on by asking.
Control Tower
Running in productionIncluded in every planEvery exception that needs a decision today, with the arithmetic that found it.
What it produces
- Exception feed ranked by financial impact
- Nineteen KPIs, each computed in SQL with its formula shown
- Approvals and audit trail
- Evidence export for any figure on the page
A worked example
Days inventory outstanding — 61.4 days
average inventory value ₹4,71,20,880
÷ cost of goods sold ₹28,01,44,000
× 365
= 61.4 days · 4,182 rows · as at 31 Jul 2026
Illustrative figures. On your tenant the same page shows yours.
What is not finished
Nothing outstanding. This is the product, it runs against live tenant data, and every figure it shows carries the arithmetic that produced it.
DeadStock AI
Beta — live for pilot tenants₹25,000 / monthFind the inventory quietly holding your working capital, and say what to do about it.
What it produces
- Every material classified: healthy, excess, slow, non-moving, obsolete, stockout risk
- Blocked working capital, by material and by plant
- Open purchase-order exposure against real future demand
- A recommended action per material — stop buying, cancel, transfer, consume, substitute, liquidate
- A backtest showing how the same model would have performed on a past date
A worked example
RM-10291 — non-moving, ₹20,60,000 recoverable
stock 8,400 KG · value ₹14,20,000
last consumption 183 days ago
open PO 4,000 KG · confirmed demand next 90 days 1,200 KG
→ cancel or reduce the open PO; transfer to Plant C, where demand exists
Illustrative figures. On your tenant the same page shows yours.
What is not finished
BETA rather than AVAILABLE: engine, golden dataset, backtest, versioned run persistence, operator page and nightly scan are all built and tested, and no customer has yet run it against their own data. It moves to AVAILABLE when one has.
SupplierGuard
Beta — live for pilot tenants₹25,000 / monthSee a supplier failing before the line stops, not after.
What it produces
- OTIF, on-time and in-full separately, per supplier and trending
- Lead-time variance and confirmation reliability
- An explainable risk score — every point of it attributable
- Production exposure in rupees, per supplier
- Early-warning backtest: how many days ahead it actually called the delay
A worked example
Supplier ABC — risk HIGH, ₹42,00,000 exposed
OTIF 91% → 78% → 64% over three quarters
lead time +6.4 days against contract
12 materials affected, 3 with no alternate source
→ expedite PO 45001823; qualify a second source for RM-8821
Illustrative figures. On your tenant the same page shows yours.
What is not finished
BETA rather than AVAILABLE. Contracted lead time now imports through the suppliers template, so lead-time drift — twenty of the score's 120 points — can be scored against what a supplier actually agreed. One honest limitation remains: single-sourcing is inferred from purchase history rather than read from an approved-vendor list, so a material bought from one supplier by habit reads the same as one that can only be bought from them.
EXIMGuard
In build — not switched on₹30,000 / monthCatch the document mismatch before it becomes detention.
What it produces
- Structured extraction from invoice, packing list, shipping bill, BL/AWB, COO, LC
- Cross-checks between every pair that must agree, field by field
- Discrepancies flagged for a person — never a silent correction to a trade document
- Free-time countdown with detention and demurrage exposure before it is charged
A worked example
SHP-4471 — 3 discrepancies, 2 free days left
invoice gross weight 12,480 KG · packing list 12,840 KG
BL consignee differs from the commercial invoice
last free day 22 Aug · estimated demurrage if missed ₹1,84,000
→ human review required before filing
Illustrative figures. On your tenant the same page shows yours.
What is not finished
Cross-check engine, letter-of-credit constraints and the free-time countdown are built and tested — a hand-worked golden dataset, fifteen mutants against the rules and six against the query bounds. Held back because there is no document store and no extraction pipeline, so on live data every file can only be reported as unchecked. Not released until documents can actually be read.
LandedCost AI
Beta — live for pilot tenants₹20,000 / monthWhat the imported material actually cost you, not what the invoice said.
What it produces
- Landed cost per unit, with every component itemised
- Variance against standard cost, previous shipment, supplier, route and Incoterm
- A reconciliation proving the components sum to the total
A worked example
RM-8821 — ₹247.60/KG landed, 11.4% over standard
material ₹198.00 · freight ₹21.40 · duty ₹19.80 · CHA and port ₹6.10
detention ₹2.30 — avoidable
standard cost ₹222.20 → variance ₹25.40/KG on 18,000 KG
= ₹4,57,200 above standard on this shipment
Illustrative figures. On your tenant the same page shows yours.
What is not finished
BETA rather than AVAILABLE. Per-line gross weight and volume now exist and import through the purchase-orders template, so freight allocates by weight rather than falling back to value — the precise error the module exists to prevent. The fallback chain still runs where a line has no weight, and it names the basis it used on every row, so a tenant who imports without weights gets a correct answer computed a worse way and is told so rather than left to assume.
FreightAudit AI
Beta — live for pilot tenants₹20,000 / monthCheck the freight invoice against what was actually contracted.
What it produces
- Expected charge rebuilt from the rate card, line by line
- Variance against the invoice, with the evidence for each difference
- Duplicate invoices and duplicate charges
- Potential recovery, with a confidence and a source reference
A worked example
INV-99213 — potential discrepancy ₹84,200
fuel surcharge invoiced 18.5%, rate card 14.0%
chargeable weight 2,140 KG invoiced, 1,980 KG on the AWB
one accessorial charge not present in the contract
→ evidence attached; raise with the carrier
Illustrative figures. On your tenant the same page shows yours.
What is not finished
BETA rather than AVAILABLE. Shipment weight and volume, carrier invoice numbers and contracted tariffs all import now, so weight-priced contracts audit and duplicate-invoice detection can fire. What remains is a data-quality dependency rather than a missing feature: an invoice can only be audited against a tariff somebody uploaded, and a lane with no rate card on file returns NO_RATE_CARD — correctly, and to a customer who may not know why.
StockOpt AI
Beta — live for pilot tenants₹20,000 / monthSafety stock and reorder points from your own variability, not a rule of thumb.
What it produces
- Recommended safety stock and reorder point per material, formula shown
- Demand and lead-time variability, measured from your history
- Expected service-level impact of each change
- Inventory reduction available at the service level you actually want
A worked example
RM-4410 — safety stock 2,100 KG → 1,340 KG
demand σ 318 KG/month · lead time 24 days, σ 5.1 days
service level 97.5% → z = 1.96
= 1,340 KG, holding 97.5% cover rather than the 99.4% you hold today
→ ₹1,28,000 released, service level unchanged in practice
Illustrative figures. On your tenant the same page shows yours.
What is not finished
BETA rather than AVAILABLE. Review policy, review period, order minimum, pack size and a nullable unit cost all import through the materials template now, so a recommendation is a shippable quantity rather than a mathematical one and an uncosted material can be told from a free one. The engine refuses a material whose review policy nobody has stated rather than assuming continuous review — so a tenant who does not fill that column gets stated refusals, which is correct and will look like the module not working until somebody reads them.
DG Compliance
In build — not switched on₹30,000 / monthDangerous-goods paperwork that a person has signed off, with the rule version recorded.
What it produces
- SDS, UN number, proper shipping name, hazard class, packing group per material
- Transport-mode and packaging requirements
- Declarations and labels
- Every determination carrying its source, rule version, jurisdiction and effective date
A worked example
CH-2210 — UN 1263, Class 3, PG II
flash point 21°C · not a marine pollutant
source: IMDG Amendment 42-24, effective 01 Jan 2026
packaging 4G/Y, limited quantity threshold 5 L
→ confirmed by a named compliance reviewer on 14 Aug 2026
Illustrative figures. On your tenant the same page shows yours.
What is not finished
COMING_SOON, and unlike the others this is not a data gap — it is safety. The check engine is built and tested — consistency checks on a stated classification, segregation across every pair including subsidiary risks, rule versioning on every verdict, thirty-three tests and twenty-one mutants, three of which attack the safety posture rather than the arithmetic. It stays COMING_SOON for three reasons. The segregation table is an unverified transcription: no competent person has checked it against the published Code, and passing tests prove only that the code applies the table it was given. There is no regulatory data source — a classification must be keyed in from a safety data sheet, because this module deliberately contains no function that infers a UN number from a description and never will. And the human-review workflow is a field on the verdict, not a screen: every verdict already demands a named reviewer, and nothing yet lets one sign.
DocumentGuard
Beta — live for pilot tenants₹15,000 / monthKnow which certificate expires before the customer's auditor does.
What it produces
- Every SDS, TDS, COA, ISO certificate, licence, insurance and contract, with its owner and expiry
- Alerts at 90, 60, 30 and 7 days, and on expiry
- Coverage gaps — the supplier or material with no document at all
A worked example
7 documents expire within 30 days
ISO 9001 · Supplier ABC · expires 11 Sep 2026
SDS · CH-2210 · expires 19 Sep 2026
insurance certificate · Carrier DEF · expired 3 days ago
→ 2 suppliers have no SDS on file at all
Illustrative figures. On your tenant the same page shows yours.
What is not finished
BETA rather than AVAILABLE. Documents import through the compliance_documents template and a daily sweep runs the expiry engine, so a finding now exists without anybody opening the page. Two things are honestly still missing: requirements — what a subject *ought* to hold — have a table and no import, so the missing-document sweep has nothing to compare against on a live tenant; and findings are recorded rather than delivered, so nothing reaches an inbox until the notification centre is wired to them.
Pending BL
Beta — live for pilot tenants₹15,000 / monthWhich bills of lading are not in hand, ranked by what it costs when the vessel lands.
What it produces
- Every sailed shipment with no bill of lading in hand, worst first
- Ranked by time to arrival, not by age — free time runs at the destination
- Issued-but-not-collected kept separate from never-issued, because the chase differs
- Value on the linked purchase orders, and how many carry none
A worked example
11 outstanding · 3 need action today
SHP-4471 · at destination 4 days · no BL issued · ₹12,00,000
SHP-4488 · arriving in 2 days · MSCU8842197 issued, not collected
SHP-4502 · ageing 21 days · vessel due 14 Jul
→ 2 shipments have no departure date and cannot be aged at all
Illustrative figures. On your tenant the same page shows yours.
What is not finished
BETA rather than AVAILABLE: the engine runs against live tenant data today — every field it reads is already populated by the tracking pipeline, which is what lets it ship while EXIMGuard cannot. No customer has yet run it against their own shipments. It moves to AVAILABLE when one has. One honest limitation: value at risk is read from the linked purchase order, so a shipment with no PO linked is counted in the list and excluded from the total, and the page says how many.
OTIF Post-Mortem
In build — not switched on₹25,000 / monthWhy the deliveries that missed, missed — with the order numbers, not a recollection.
What it produces
- A waterfall decomposing the OTIF gap into named causes, by value
- Every missed line with its cause and the record that proves it
- The share of missed value nothing on file explains, stated plainly
- Lines that could not be assessed at all, and what is missing
A worked example
June — OTIF 87.4%, and where the 12.6 points went
supplier late · 5.1 points · ₹18,40,000 · 7 lines
promised inside lead time · 3.2 points · ₹11,60,000 · 4 lines
short shipped · 1.8 points · ₹6,50,000 · 11 lines
unexplained · 2.5 points · ₹9,00,000 · 6 lines — 20% of the miss
Illustrative figures. On your tenant the same page shows yours.
What is not finished
COMING_SOON. The engine is built and tested — 47 tests, 15 mutants, all killed — and it decomposes a period's OTIF into causes that each point at a record, refusing to name one where nothing on file supports it. What is missing is three of the six causes on live data. Stock at a past date cannot be reconstructed from the current inventory row, realised lead times are not yet wired through from supplier history, and outbound shipments are not joined to delivery lines — so PROMISE_INSIDE_LEAD_TIME, STOCK_NOT_AVAILABLE and CARRIER_LATE cannot fire against a real tenant yet. The engine handles all three and the golden dataset covers them; the queries that would feed them do not exist. Releasing now would ship a post-mortem whose unexplained share is high for a reason the customer would reasonably read as the module not working.
SIAARU CRM
In build — not switched on₹20,000 / monthA pipeline that knows whether you can actually deliver what you are about to promise.
What it produces
- Deals, contacts and activities, with the pipeline by stage and expected close
- Every deal line checked against stock, open orders and measured lead times
- A per-line verdict: from stock, from stock and orders, if ordered now, or cannot fulfil
- The shortfall and the days of slack, with the arithmetic shown
- A won deal converted to a sales order without re-keying a line
A worked example
D-2291 — Acme Industrial, ₹48,00,000 — cannot fulfil
line 3 · RM-10291 · 4,000 KG promised 12 Sep
820 available plus 0 incoming leaves 3,180 short
lead time 34 days against 21 to the promise
→ 13 days late even if ordered today; move the date or split the delivery
Illustrative figures. On your tenant the same page shows yours.
What is not finished
COMING_SOON. The fulfilment engine is built and tested — 23 tests, 17 mutants, all killed — and it is the reason to build a CRM here rather than buy one: nothing else can check a promise against your own stock and measured lead times. What is missing is the rest of a CRM. There are no screens, no import path for deals or contacts, no email or calendar capture, and no conversion of a won deal into a sales order. A pipeline nobody can enter data into answers no questions, however good the engine behind it.
SIAARU HR
In build — not switched on₹15,000 / monthLeave and skills judged against what the plant actually needs that week.
What it produces
- Employees with skills, proficiency and certification expiry, by plant
- Leave requests with an approval trail
- Skill cover per plant per week, so a leave approval can be judged against it
- Certifications lapsing before the work that needs them
A worked example
Plant B, week of 14 Sep — one CNC operator covered
3 of 4 CNC operators on approved leave
remaining operator's certification expires 11 Sep
→ approving the fourth request leaves the cell unmanned
Illustrative figures. On your tenant the same page shows yours.
What is not finished
COMING_SOON, and further back than CRM. The schema is designed around the one thing a general HR product cannot do — judge a leave request against the skills a plant needs that week — but the engine that does the judging is not written, and neither are the screens or the import path. Nothing here works yet. It is listed so the shape is on record, not because any of it can be sold.