Project setup

In Eryx, a project is a dedicated workspace for one audience or campaign: you define who you want (ICP), where to look (platforms), and what phrases to search for (queries). Those settings drive which leads appear in the project and how they are scored and enriched.

If terms like discovery, enrichment, and runs are new to you, read Core concepts and the Quickstart first.

Where you configure a project

When you create it

1. Sign in and open Projects.
2. Choose New project.
3. Fill in the form and save.

That sets the initial configuration: name, audience description, platforms, queries, and optional advanced fields.

After it exists

On the project page, open Edit. You get the same kind of settings dialog: change any field and choose Save changes.

Save is only enabled when there are unsaved changes and every required field is valid.

Name, niche, and description

Project name

A short label for lists and navigation (for example “B2B SaaS founders” or “Berlin coffee shops”). Required.

Niche (optional)

One line: industry, segment, or audience type. Helps both you and the AI when suggesting queries.

Description (strongly recommended)

Write in plain language:

• Who the ideal customer is: role, company size, geography, language.
• What you offer and why it matters to them.
• Exclusions: who you do not want (agencies, competitors, wrong region).
• Intent: demo, partnership, hiring — how you want to approach people.

Clearer descriptions improve relevance scoring and AI query suggestions. For Suggest with AI, you usually need at least a name, a niche, or a longer description (the UI tells you if there is not enough context).

Project status

In edit mode you can set:

Exact behaviour of run buttons may vary by release; if something is disabled, check status and on-screen hints.

Project icon

A small visual badge in project lists. It does not affect lead discovery — it is only for quick recognition.

AI model (provider and version)

The model is used for tasks such as scoring and enrichment during runs.

1. Pick a provider (for example OpenAI or Anthropic).
2. Then pick a specific model from that provider’s list.

Credit usage depends on the model: stronger models usually cost more. For broad discovery you might start with a lighter model; for difficult niches, a stronger one. Test on small batches and watch quality versus cost in Billing & usage.

Additional outreach instructions (optional)

A free-text field for tone and guardrails when preparing outbound messages: formal vs friendly voice, topics to avoid, how you like calls to action phrased.

This adds to the main system instructions; it does not replace them. Reply structure stays consistent with the product — you tune how it sounds within those rules.

Connecting email or messengers is separate — see Integrations.

Max results per query

This number is a rough cap on how many matching profiles or posts the system tries to fetch per query and per enabled platform.

• The default is typically 25.
• In production the UI usually allows 15 through 100 (the field shows min and max).

Higher values widen the net but take longer and use more credits. Lower values are faster and cheaper but may miss rare matches.

Start at the default; increase only if you need more volume and your budget allows.

Source platforms

Choose where Eryx searches for this project. You must enable at least one platform.

Typical options (labels match the in-app names):

Tip: avoid turning on every platform “just in case”. Two to four well-chosen sources usually beat a fuzzy ICP spread everywhere.

Custom queries (search phrases)

Each phrase is its own search line across your selected platforms. You need at least one non-empty query.

How to write them:

• Mix roles (“SaaS founder”), topics (“no-code automation”), geo (“Berlin startups”), and language when it matters.
• Prefer several focused queries over one huge catch-all.
• Name industry and pain instead of only generic words like “marketing”.

Suggest with AI adds ideas from your name, niche, description, platforms, and existing queries. Always review and edit — suggestions can be too broad or off-brief.

Auto-enrich leads

The Auto-enrich option means new leads from a discovery run can be enriched automatically without a separate step.

• On — less manual work, higher credit use per discovery run.
• Off — you curate leads and start enrichment & outreach from the on-page workflow (see Quickstart).

Pick based on whether you want to filter raw leads first or immediately get enriched records.

How this ties to the on-page workflow

Project settings define search and processing rules. The Workflow block on the project page is the guided path:

1. Lead discovery — run search with your platforms and queries.
2. Select leads — manually, quick presets, or “latest batch” after discovery.
3. Enrichment & outreach — for the selection (unless you rely mainly on auto-enrich).

Every execution also appears under Runs, with pipeline labels and progress.

FAQ

Save is disabled or does nothing

Check: name is set, at least one platform is on, at least one query exists, max results is inside the allowed range, and you actually changed something since the last save.

Too few leads or poor fit

Tighten your ICP description, narrow queries, drop irrelevant platforms, tune max results, and try a different AI model if needed.

I need email or Telegram to send messages

Connect channels under workspace settings — Integrations. Plans and credits — Billing & usage.

Table behaviour, CRM status, bulk actions

See Leads & CRM.

Checklist before your first discovery run

1. Clear name and audience description.
2. Relevant platforms only (not “all toggles on”).
3. Several concrete queries (optionally refined with AI).
4. AI model chosen with budget in mind.
5. Decide whether auto-enrich should be on.
6. Optional: outreach instructions filled in; integrations connected if you will send from Eryx.

Good luck with your pipeline.