Procurement · for owners running BESS, solar and hybrid tenders
Every offer read against your RFQ.
The risk it leaves you, priced.
A tender decides who pays when something fails, whose duty it is to fix it, and what obligations come with the asset — for the next twenty years. GRID reads every bid against your RFQ and puts a price on the exposure it would leave with you.
Upload your RFQ before it goes out, or click through a finished tender with the vendors anonymised. Nothing to install. Offers already in? Start where the reading happens →
What a read gives you, on every offer
ExampleHeadline capex
€2.96M
What the offer says it costs.
Exposure left with you
€612k
Across seven dimensions, after what is recoverable.
Fifteen-year cost drift
+€940k
Against the lifetime cost your RFQ assumed.
Protections moved back
6 of 7
Guarantees your RFQ placed with the vendor, handed back.
Four figures behind the headline price.
The whole round
One workspace, from the RFQ to the contract.
Software your team logs into, working on the documents your vendors actually sent. Every stage ends with a document you can act on.
-
It writes the RFQ, and names every requirement you have left undefined
Generated from what your team settles, or built from the RFQ you already have. Either way, anything left undefined is named before a vendor sees it, with the party who decides it if you do not.
-
It reads every offer against that RFQ, and tests every guarantee in it
Works with however many vendors respond, in whatever format they send. A verdict on every requirement, an automated enforceability check — whether each guarantee would actually pay out — and every figure traced to its clause and page.
-
It prices what each offer would actually cost you
Seven dimensions of exposure, in euros. Your lender’s covenants checked against your own term sheet, and the guarantees that would fail a claim.
-
It turns the negotiation into questions, then into a contract
One set of questions per vendor, in your words, each showing the euro amount it would change if answered. At the end, a contract draft with every clause traced back to the document it came from.
How a round runs
Four stages. Start where your tender already is.
A round is one pass through your tender: the RFQ out, the bids in, the questions back, the contract. Take the stages one at a time, as far as your tender needs to go. Keep scrolling — the round runs as you go.
Tender preparation — build the RFQ, or bring yours
Before any vendor is in it: GRID turns what your team settles into the RFQ document, and prices every requirement you leave open before it goes to market.
The RFQ, and every decision it leaves open carrying a figure.
The first proposal check — the bids arrive
It reads each bid in the format it arrived and fills one table: a verdict per requirement, an automated enforceability check on every guarantee, an exposure figure per dimension, and the clause behind each. Read across the field, that is where your negotiating angles come from.
The scored table, and the negotiating angles it hands you: who to press, on what, and what it is worth if they move.
The revision round — press what is worth pressing
It writes one set of questions per vendor from the angles the first check found. Nothing reaches a vendor unless you send it — the relationship stays yours, and the round runs as many times as the negotiation needs.
The asks written, then the delta: what moved, what came off the table, what is still open.
Close-out
It drafts the contract from what was actually settled, and cuts the same round three ways for the people who sign, fund and insure it.
A contract draft, a risk allocation summary, and three packs — board, lender, insurer.
That is the round
Seen enough? Watch one run end to end, or start on your own RFQ.
A live tender with the vendors anonymised: the RFQ as it was put together, and the first proposal check on the bids that came back. You click through all of it — everything opens, nothing is editable. Or put your own RFQ through the free read below.
Rather talk it through? Oliver has run these tenders from the owner’s side.
Automated enforceability check
The offer guarantees fifteen years. The claim would die in year three.
A guarantee is worth what it would pay, and that is a different number from what it promises. GRID tests every guarantee in every offer against the seven conditions that decide whether it could be claimed at all, and prices the difference.
How a fifteen-year guarantee dies in three
The cover lasts fifteen years, and the notice period is thirty days from the day you find the problem. An owner who spots the shortfall in March and writes in June still has the guarantee, and has lost the claim. The claim was good; the clock ran out. Notice is the condition owners lose on most.
Conditions
per guarantee
The window, the notice, your own obligations, the operating envelope, the counterparty, the evidence, and who caused it.
Clocks
on every claim
Cover runs in years, notice runs in days. The shorter one decides.
Failure
is enough
Fail any one of the seven and the protection is gone, whatever the clause says — priced as what it would have covered.
Vendor exposure
We price the risk. Every vendor, every dimension, one table.
For each bid, GRID scores how much of each risk would stay with you, as a percentage, and what that comes to in euros. Read down a column to compare vendors on one risk; read across a row to see the shape of one bid.
Owner risk per dimension — lower is better
Example| Vendor | Performance | Payment | Scope | Warranty | Schedule | Enforceability | Viability | Composite |
|---|---|---|---|---|---|---|---|---|
| Vendor A | 34% | 22% | 57% | 74% | 28% | 49% | 18% | 40% |
| Vendor B | 41% | 63% | 22% | 38% | 55% | 31% | 26% | 39% |
| Vendor C | 29% | 18% | 44% | 26% | 33% | 71% | 62% | 40% |
The composite averages the dimensions assessed, at the weights your RFQ’s position sets. Forty and sixty are our lines until you declare your own.
And in euros, per bid
ExampleGross
Recoverable
Net carried
Probability-weighted, and net is the figure that belongs beside the headline price.
What are you trading away?
Every offer is scored in four categories. The trade sits between them.
Every requirement in your RFQ, and every clause a vendor answers it with, lands under one of the four. That is what makes an exchange between them visible — a gain in one is almost always bought from another.
Technical
Usable energy at end of warranty, round-trip efficiency, and the degradation curve behind both.
Commercial
Capex, every line sitting outside it, fifteen years of opex, and the payment security behind them.
Delivery
COD, the scope boundaries around it, and the damages that apply if the date slips.
Warranty & service
Tenor, what starts it, the guarantees inside it, and how each one is measured.
ESG runs as an optional fifth — cell origin, end-of-life obligations, embodied carbon — scored only if you want it scored.
The weights between them are yours, and every figure shows them
You set what each category is worth, and what is hard rather than scored, before the round opens — with your organisation’s own evaluation rules in place of ours. Euros stay euros and years stay years: the categories are never added into one number.
The trade-off engine
Every gain is paid for somewhere else. It names where, and what drives it.
Direction and drivers, on your own asset. Not a price — before the market answers, no price for a trade-off exists, and the figures arrive with the bids.
Ask for
A guarantee you can actually claim
It comes out of
Capex, and the size of the field
What drives it
A claimable guarantee needs a measurement regime, a commissioning baseline and a liability the vendor reserves against. It can also narrow the field without looking extreme: a sub-50 ms response requirement entails grid-forming inverters, and a long unaugmented warranty entails an augmentation agreement. A tender narrowed to one supplier is usually caused months earlier, by a requirement nobody read as a capability.
Ask for
An earlier delivery date
It comes out of
Price, or the damages you accept
What drives it
Pulling a COD forward is priced by the vendor as expedited supply and compressed commissioning. Letting it slip is priced by the market instead — a month of first-year revenue, which is the figure a damages cap should reach.
Ask for
More energy through the asset
It comes out of
Augmentation, brought forward
What drives it
A harder cycling regime steepens the degradation curve, which brings the augmentation spend forward. The warranty may also be written against a gentler regime than the one the revenue case assumes, in which case the cover and the plan disagree.
What you get, run over run
What moved between runs, and what came off the table.
A revised bid is read against the one before it: what the vendor conceded, what they left alone, and what they quietly withdrew once you pressed elsewhere.
The revision round runs as many times as the negotiation needs, and every run scores against your RFQ — never against the run before it. Only the delta reads backwards.
What improved
What did not move
What was withdrawn
What it is worth, either way
The floor
The ordinary parts of a tender, done properly.
None of it wins a tender. All of it loses one when missing.
Comparison
Every requirement answered, and every silence named
A silence the vendor left becomes a line in your next ask. One your RFQ never raised is a note about your own document, never counted against them.
A conditional pass is its own verdict: met, but only at 25 °C ambient, or only for the first augmentation cycle. The condition travels with the cell.
Provenance
Every figure traced to its clause and page
Read from their document, taken from your RFQ, computed by a rule you can open, or a house estimate that says so.
Round-trip efficiency, system level — sourced from Technical Proposal v2, clause 7.2, page 14. Every other figure carries the same line: declared, derived or estimated.
Gates and floors
What blocks an offer, and what only flags it
Only your lender’s covenants block. Published DFI criteria flag, and a flagged offer stays fully scored — yours to carry forward.
A DSCR under 1.35 at your term sheet’s values blocks. A published DFI criterion flags, and the offer stays scored.
You know what you asked for. This is what a vendor can do with it.
Upload it before it goes out. You get back the wording that closes the gaps, clause by clause.
Where it pushes cost
The clauses driving capex up, how many sit at the ceiling of their band, and which way the document as a whole is pulling.
What it leaves to the vendor
Everything you left undefined, split two ways: what a bidder has to price, and what they can answer in their own words and define for you.
Who ends up carrying the risk
Every risk the document places, and where it lands: carried by you, transferred, shared at an interface, or mitigated by the design — and whether a clause prices it at all.
The position it encodes
Every declaration read from the clauses, and where the technical and the commercial half each land. When the two halves ask for different assets, the read says so.
Any RFQ, any stage, and nothing runs against a vendor. An account is needed to accept the terms before upload: your RFQ stays yours, and nobody here reaches it without an entry in the log.
One read, before market
Example41
clauses read
+18%
ceiling of the capex band this document sets
14
topics with no clause at all
9
clause revisions proposed, with wording
Risk this places on you
4 of 7 land with you
Three of the four you carry have no clause that prices them.
The position it encodes
The technical half
Fortified
6 clauses, without exception
The commercial half
Lean
5 clauses, without exception
“This document does not speak with one voice.” The technical half asks for an asset built to last; the commercial half prices one built to earn early. A bidder answers whichever half suits them.
What it costs
Pay as you go.
No subscription, nothing renews. You pay for a stage when you start it, and take only the stages your tender needs.
01Build the RFQ
Your requirements settled, and the RFQ generated from them.
02Read the offers
Every requirement verdicted and every exposure priced, per vendor.
03Run a revision round
The asks out, what came back, and what moved. Priced per run.
04Close it out
The contract draft, the risk allocation summary and the three packs.
A tender takes months, and what it decides you carry for twenty years. Six vendors, ninety pages each, three rounds of revisions. I have written those offers and I have read them, and it ends the same way every time: one person, holding all of it in their head, the week before they sign.
Your vendors. Your list. Always.
You invite the suppliers you want — the ones you have worked with, the ones your team trusts — or ones matched from GRID's pre-qualified network. Your list, your call, every time.
You run it. Nothing moves without you.
From your RFQ to your signature — four steps, and you are the one holding the pen at every one of them.
1 · Your RFQ
Bring your own format, build one here, or generate it from your planner position. You choose the vendors — send through the platform, or download the PDF and send it yourself.
2 · Upload the proposals
One folder per vendor, as they arrive. Every proposal is read against your requirements and the gaps priced. You will be notified when results are ready.
3 · Review with the source in hand
Every data point shows where in the document it came from; low-confidence reads are flagged for you to confirm or correct before they count.
4 · Respond
Feedback and questions per supplier, prepared for you. You edit, approve, send — nothing reaches a supplier without your review.
Before you ask
What people want to know first
So which offer do we take?
That stays with you. Each reading orders the field in its own unit, and the ordering changes between them — which is the finding. Combining euros, points and percent into one number would need a weight between them that nobody has chosen, and we will not invent it on your behalf.
We already have the RFQ written. Can we still use this?
Yes, and the free read is where to start: upload it and we read what it settles, what it leaves undefined, and what each gap would be worth. Nothing runs against a vendor.
We answer to a different lender. Whose criteria does it use?
Yours, if you have them. Your term sheet’s covenants go in at their own values and those are the only things that block. What ships in the box is a set of published DFI criteria — a starting point you can swap out, keep alongside your own, or drop. And if your organisation already has evaluation rules it uses, those go in instead of ours.
Does a flagged gate throw an offer out?
No. It stays in the field, fully scored, with the requirement and parameter named. You may well carry a flagged offer forward — to choose that, you need it scored. Only your own lender’s floor blocks.
Where do the numbers come from?
Every figure traces to a document, clause and page in the bidder’s own submission, or says that it is ours and whose. Uncertain reads are flagged before they count.
Does it talk to the vendors for us?
No — and for now that is deliberate as well as true. GRID assembles the questions from the angles the reading found; you decide what goes out, to whom, and in what words, and it leaves from your own inbox. Nothing reaches a vendor unless you send it, and the relationship stays yours. Their answers come back in here. Sending and collecting is what we automate next, and it will stay opt-in.
Does it know the regulations where we are building?
EU-level requirements are in already. Country-level ones are being added, so they arrive as your market is covered rather than as something you have to load yourself. Anything specific to your site goes into the RFQ as a requirement like any other.
How is this different from our owner’s engineer?
It does not replace them. It gives them one place where every requirement, figure and source is already assembled against your RFQ — so their time goes on judgement rather than on collating six documents. And it keeps the record afterwards, which an adviser’s report does not.
How many revision runs are there?
As many as the negotiation needs. GRID never decides that enough have been run — it can say that movement between runs is flattening, and show you the figures, but moving to close-out is your call. Runs are priced one at a time, so the pace is yours.
Start anywhere
Bring the round you are in the middle of.
Watch a finished tender, send an RFQ for a free read, or put twenty minutes in Oliver’s calendar — he has run these from the owner’s side.
The module before this one
The tender answers a position. Somebody has to take it first.
How the asset earns, what it costs to keep, what protects it and where the risk sits — settled once for twenty years rather than for one tender. The Strategic Posture Planner leaves with three documents out of one dataset: the board’s investment brief, the engineer’s brief, and the RFQ this module runs.
Where a position lands
One dot per decision. The area between the furthest four is the position your RFQ takes to market.
Round 0, if you came from the planner:
The RFQ generates from the frozen position you agreed in the planner. Edit it freely — every edit reads back against the position, so nothing drifts between what you decided and what you asked for.