Bloomteq Procurement AI

Overview

Bloomteq serves two audiences from one platform. A selling team uses it to respond to the RFPs and RFIs sponsors send them: it reads the request, drafts each answer from an approved Response Library, and moves the draft through review to a signed-off export. A buying team uses it to run a sourcing event: it reads every vendor offer into one canonical bid so price, service levels, and compliance line up side by side.

What makes Bloomteq different is that nothing is taken on faith. Every value it produces, whether a drafted answer or an extracted bid figure, carries a confidence and a pointer back to the source it came from. A person stays in control of the decision, and every step is recorded.

How it works

Bloomteq is built on one extraction-and-review spine. An offer comes in, Bloomteq extracts a structured bid, a person reviews and confirms the flagged values, and the result is exported. The same spine underpins every automation in the platform.

A canonical schema is the single source of truth for what a bid looks like. Model output is normalized into that shape and validated before anything is stored, so the data an operator sees is always well formed.

Responding to RFPs and RFIs

This is the selling team's side of Bloomteq. A sponsor sends you a request, and Bloomteq carries it from the incoming document to a finished, signed-off response, grounded in answers your organization has already approved.

1. Bring in the RFP or RFIUpload the sponsor's document in whatever format it arrived. Bloomteq reads it, including scanned files, so nothing has to be retyped.
2. Understand the askFor an RFP, Bloomteq extracts the focus areas, expectations, key dates, and mandatory requirements. For an RFI, it splits the document into individual questions.
3. Draft grounded answersBloomteq drafts each answer from your Response Library, with a confidence and the exact approved sources it drew on. It invents nothing; where the Library is thin it marks the gap for an expert.
4. Review and sign offA person checks each answer, corrects what needs it, and signs off. Only a signed-off answer is authoritative.
5. Export and reuseThe finished response exports to Word, and signed-off answers flow back into the Response Library so the next response starts further ahead.

Running a sourcing event

This is the buying team's side. A tender moves through the same clear stages from first draft to signed award.

1. Capture requirementsAn operator sets out the tender: the scope, the evaluation criteria, and the weighting that will decide the award.
2. Approve to publishNamed approvers sign off in parallel. The tender cannot be published until every required approver has cleared it.
3. Publish and inviteBloomteq publishes the tender and issues each vendor a tokenized portal link. The token is the capability; no vendor account is required.
4. Collect sealed bidsVendors submit proposals through the portal. Every submission is sealed on arrival and stays sealed until the deadline passes and bids are opened.
5. NormalizeBloomteq reads each proposal, in whatever format it arrived, into one canonical bid so that offers can be compared on equal terms.
6. Score against the criteriaThe ruleset evaluates each complete bid against the published criteria and weighting, producing a transparent, repeatable score.
7. Verify and decideThe operator reviews every flagged field, confirms or corrects it, and makes the decision. Bloomteq drafts; the human decides.
8. Sign off and awardApprovers sign off on the award, Bloomteq records the decision, and a purchase order is generated for the winning vendor.

Autonomy tiers

Every capability in Bloomteq carries an autonomy tier that sets exactly how far the system may act on its own. The tier is enforced in code, not in a prompt, which is the platform's core safety guarantee.

Provenance and review

Every extracted value ships inside a provenance envelope. Alongside the value itself, Bloomteq records how confident it is and the source snippet the value was read from. No value ships bare.

Low-confidence values are flagged for a person to look at. Each flagged value moves through a simple review state as a person works through it.

The review workflow is enforced in the data layer, so a value cannot become authoritative without a person passing it through review.

Roles and access

Access is role based, and permissions are enforced on the server. Each organization is a separate tenant, and data is scoped to the tenant a person is acting in.

The audit trail

Every meaningful action is written to an append-only audit trail. Entries are hash-chained, so each one depends on the one before it. If a past record were altered, the chain would no longer verify, which makes silent tampering detectable.

The result is a complete, ordered account of what happened on a tender: what was received, what was extracted, what a person confirmed or corrected, how it scored, and who decided the award.

Vendor submission portal

When a tender is published, each vendor receives a private portal link carrying a unique token. Vendors do not need an account. They open the link, review the tender document, and submit their proposal.

Submissions are sealed the moment they arrive and remain sealed until the deadline passes and bids are opened together. This keeps the process fair and keeps early submissions from influencing later ones.

Getting started

Access is granted by your organization's administrator. Once you have an account, sign in to reach your dashboard, where tenders waiting on your decision are surfaced first.

Requesting access

If you do not have an account yet, ask your administrator to invite you, or request access from the sign-in screen. You will be given the role that matches your part in the procurement process.

Support

For help with the platform or your account, contact your organization's Bloomteq administrator. This documentation grows as further automations are sequenced into the platform.