How Dealglint works
Three steps, no black boxes.
1. Tell us what your agency does
Enter your website and we analyze it to suggest your services, technologies and industries. You confirm and edit everything, then choose target countries, company sizes, project budgets and keywords — including keywords to exclude. On the Agency plan you can keep separate profiles for different teams or service lines.
2. We discover and qualify relevant signals
Scheduled connectors collect public, permitted data — EU procurement notices from TED, UK notices from Find a Tender, and hiring activity from public Greenhouse and Lever job boards — plus anything your team submits by URL or CSV. Signals are normalized, deduplicated across sources and scored against every profile.
Direct
A buyer is explicitly asking for suppliers: published tenders, RFPs, service requests.
Inferred
Indirect evidence such as hiring. A job posting is not proof that a company wants to outsource — we label it accordingly.
Switching
Public statements of looking for an alternative product or provider, from authorized sources only.
3. Review opportunities and approach with context
Each opportunity shows the organization, source link, dates, country, intent type, technologies, a summary, why it matches, what's missing and a recommended next action. Save it, add notes, move it into your pipeline, and generate an editable outreach draft based on the evidence. You send it yourself — Dealglint never sends cold email or LinkedIn messages on your behalf.
The Agency Fit Score
A deterministic 0–100 score. The same inputs always give the same score, and every point is explained on the opportunity page.
| Factor | Max | How it's calculated |
|---|---|---|
| Service relevance Overlap between the services detected in the source (including procurement CPV codes) and the services in your profile, plus your keywords. | 30 | Overlap between the services detected in the source (including procurement CPV codes) and the services in your profile, plus your keywords. |
| Intent strength Direct requests (tenders, RFPs) score 15. Switching signals 10. Inferred signals such as hiring 6. | 15 | Direct requests (tenders, RFPs) score 15. Switching signals 10. Inferred signals such as hiring 6. |
| Technology fit Technologies named in the source that you work with. Not stated = partial credit and a listed limitation. | 10 | Technologies named in the source that you work with. Not stated = partial credit and a listed limitation. |
| Industry alignment Whether the buyer's industry is one you serve. Unknown = partial credit. | 10 | Whether the buyer's industry is one you serve. Unknown = partial credit. |
| Geography Opportunities outside your target countries are excluded. Unknown location = partial credit. | 10 | Opportunities outside your target countries are excluded. Unknown location = partial credit. |
| Freshness Newer signals score higher. Opportunities with a passed deadline are excluded. | 10 | Newer signals score higher. Opportunities with a passed deadline are excluded. |
| Company size Compared with your preferred sizes when the size is known. | 5 | Compared with your preferred sizes when the size is known. |
| Source reliability Official procurement data ranks highest. Your team's feedback can lower a source for your workspace. | 5 | Official procurement data ranks highest. Your team's feedback can lower a source for your workspace. |
| Budget compatibility Only when a value is stated. Values below your minimum score zero and are flagged. | 5 | Only when a value is stated. Values below your minimum score zero and are flagged. |
Signal Confidence
High: a direct request from an official source with a clear service match and known location. Medium: direct requests with a weaker match, or inferred signals with strong service and technology overlap. Low: everything else. Confidence describes the evidence, not how much you'll like the opportunity.
AI's role
AI translates and summarizes source text, and drafts outreach. It is instructed to use only facts in the source and to list what's missing. Scoring itself is rule-based, so it's explainable and consistent. If a value can't be verified, you'll see "Unknown" or "Not available".
Sources and licences
- TED (EU) — Tenders Electronic Daily, reusable under the European Commission reuse policy.
- UK Find a Tender — Open Government Licence v3.0.
- Greenhouse and Lever job boards — public job board APIs; we store a short excerpt and link to the original posting.
- Your submissions — URLs and CSV imports stay private to your organization.