Market research

AI-powered web research to discover products, suppliers, pricing, and regulations for your procurement needs.

Market Research helps you understand what's available on the market before you write your RFP or evaluate bids. Describe what you need to research, optionally upload reference documents, and the AI agent searches the web, reads product pages, and compiles structured findings backed by real, verifiable sources.

When to use it

Before writing an RFP — validate that your technical specifications match what's actually available on the market
Generating specifications — you know what you need but don't have a formal spec yet; the agent researches the market and drafts one
Supplier discovery — find potential vendors, compare product lines, and understand pricing ranges
Regulatory check — identify relevant standards, certifications, and compliance requirements for your procurement domain

Starting a research run

Open Market Research

Navigate to Market Research in the top navigation (next to Funding Research and Library). This workspace is organization-wide — all research runs are visible to your team. The procurement-context sidebar only appears inside an open procurement; main product sections live in the top bar.

Create a new research run

Click New Research. Fill in:

Brief — describe what you want to research in free-form text. The more context you give, the better the results. Include product types, use cases, constraints, and any specific questions.
Language — choose the language for findings and documents (English or Latvian).
Reference documents (optional) — upload existing specifications, standards, or RFPs. The agent will read these and align its research with your requirements.
Mode — choose how involved you want to be:

Mode	How it works
Autopilot	Agent researches and produces a complete report without interrupting you.
Guided	Agent pauses after initial findings so you can review and steer the research direction.

Watch the progress

The page splits into two panels:

Left side — findings and documents appear as the agent discovers them
Right side — agent activity feed showing web searches, page reads, and decisions in real time

You can see which search queries the agent is running and which pages it's reading. A typical run takes anywhere from a few minutes to roughly half an hour depending on brief complexity, the number of reference documents, and how many iterations the agent decides it needs.

Review results

When the agent finishes, you get:

Research findings — categorized discoveries with sources, confidence scores, and evidence
Research report — a polished document summarizing the market landscape, key findings, and recommendations
Technical specification (if applicable) — a draft spec based on market data

Understanding findings

Each finding includes:

Title — the claim or discovery
Description — details and context
Type — one of: Product, Specification, Supplier, Regulation, Risk, Recommendation
Confidence — HIGH, MEDIUM, or LOW
Sources — clickable URLs with extracted evidence from the original pages

Confidence levels

Level	What it means
HIGH	Multiple independent sources, including manufacturer or official body pages. Recent information.
MEDIUM	Two sources, or reseller/distributor (not manufacturer), or somewhat dated.
LOW	Single source, forum/blog, old information, or conflicting sources.

Each finding shows a confidence rationale explaining why the agent assigned that level — for example, "3 manufacturer sources, 2024-2025 data" or "single reseller listing, no manufacturer confirmation."

Navigating findings

Findings are organized into views:

Overview — summary statistics: products found, insights recorded, documents created
Products — an expandable table of product and supplier discoveries
Insights — other finding types (regulations, risks, recommendations) organized by category

Click any finding to expand it and see all sources with extracted evidence. Source URLs are clickable — verify claims directly on the original websites.

Working with research documents

The agent produces one or more editable documents:

Market Research Report — executive summary, market overview, key findings, supplier information, pricing data, regulations, and recommendations
Technical Specification (if the research warrants it) — technical requirements and qualification criteria backed by market data

Each document opens in a rich text editor with the same toolbar used for RFP composition. Changes are auto-saved.

Guided mode

In Guided mode, the agent pauses after its initial research pass. You'll see the findings so far and a text input where you can provide feedback:

"Focus more on pricing and availability"
"Investigate supplier X in more detail"
"Ignore products below capacity Y"

The agent incorporates your feedback and continues. This gives you control over the research direction without having to specify everything upfront.

Reference documents

Uploading reference documents (specs, standards, RFPs) helps the agent in two ways:

Context — it understands your specific requirements and constraints
Validation — it can compare what it finds on the market against what you already have, flagging gaps or opportunities

Uploaded documents are parsed and indexed automatically. They appear in the Overview section so you can see what context the agent used.

Statuses

A research run moves through a small set of statuses, visible in the run header and on the research list:

Status	What it means
Pending	The run has been created but the agent has not started yet. You'll only see this briefly under normal load.
Running	The agent is actively searching, reading, and recording findings. Activity feed updates in real time.
Paused	Guided mode only — the agent is waiting for your feedback before continuing.
Completed	The agent finished. Findings, documents, and Excel export are available.
Failed	The run stopped on an error. Any partial findings remain visible for inspection.

The Run again and Download Excel actions appear once the run reaches Completed.

Credit cost

Market research runs consume credits based on AI processing. Web searches and page fetching are low-cost; the main expense is the AI reasoning about what it finds. Typical observed costs are 0.5–1.0 credits per research run, with simple briefs at the lower end and broad multi-iteration runs at the upper end. The exact charge appears in the run summary after completion (for example, Completed in 21m 25s · 13 products / services · 2 insights · 0.55 credits).

See Credits and pricing for how credits are billed across the platform.

Every claim in the research output links to a real source URL. Unlike general-purpose AI chat, you can verify any finding by clicking through to the original page.

Reference document formats

Reference documents go through the same upload pipeline as RFP and bid documents — see Working with documents for the full table. In short: PDF, Word (.doc, .docx), Excel (.xls, .xlsx), PowerPoint (.ppt, .pptx), Markdown, and .edoc archives are accepted. Files must finish parsing before the agent can use them.

Limitations and things to know

Concurrent runs. Market Research currently restricts how many runs can be active at the same time per user. If you try to start a new run while one is already Pending or Running, the New Research button is disabled with the hint "A research session is already in progress". Wait for the current run to finish, or delete it, before starting another. Funding Research enforces its own limit independently of this one — running them in parallel is allowed.
Run time. A research run usually takes between a few minutes and roughly half an hour. There's no fixed estimate shown up front; the activity feed is the most reliable progress indicator.
Deleting a run. Each Market Research run has a Delete research action that removes it permanently. Use this if you need to free up the concurrency slot or clear out experimental briefs.
Reference documents must be parsed. Uploaded references run through the same parser as RFP and bid documents. Don't start the run until each reference shows Completed status.

From research to procurement

Market Research feeds naturally into the rest of the platform:

Research the market to understand what's available
Use the findings to write a well-informed RFP (manually or via Compose RFP)
Upload vendor bids and run analysis against your market-backed requirements

On this page