Calyflow
ai workflows

Introducing Knowledge Base: give AI your team context

· 4 min read · Michal Juhas

AI tools that lack your context give generic output. You’ve probably seen it: a sourcing message that could have been written for anyone, a qualification read that misses the one thing the hiring manager actually cares about, or an outreach cadence that ignores the channel your team already knows converts.

The Calyflow Knowledge Base is the fix. It’s where you put everything the AI needs to stop guessing and start working the way your team actually works.

There are three layers. Each one narrows the context from “what every recruiter knows” to “what your agents know.”

1. Workspace Knowledge Base: your team’s institutional memory

Every team has accumulated knowledge that doesn’t live anywhere: the sourcing channels that reliably work for your market, the phrases that land in outreach, or the process you’ve figured out through trial and error.

The Workspace KB is where that knowledge lives. Think of it as a permanent briefing document your agents read before they do anything.

What belongs here:

  • Proven sourcing strategies: Which communities surface the right people and which boolean patterns work.
  • Your outreach voice: Preferred phrases, subject lines that convert, and the tone your team uses.
  • Process and SOPs: How your team handles specific situations, like a warm intro versus a cold outbound run.
  • Showstoppers: The deal-breakers your agents should never miss.
  • Email templates: The formats that have earned replies.

Without this layer, every agent starts from zero. With it, the institutional knowledge your best recruiter has in their head becomes infrastructure everyone can use.

2. Client Knowledge Base: what the AI needs for this client

Not every req is the same, and not every hiring manager is the same. The Client KB covers what’s specific to one client. This specificity is often what separates a candidate that moves forward from one that gets a vague rejection.

The AI needs to know what the hiring manager cares about, not just what the JD says.

What belongs here:

  • What this client values: The signals a hiring manager consistently responds well to.
  • Red flags they’ve called out: Things they’ve bounced candidates on before.
  • Interview process and quirks: How many rounds, who’s involved, and what the technical bar actually looks like.
  • Communication preferences: Do they want a weekly update or only when there’s news?
  • Past feedback: What they’ve said about candidates you’ve already submitted.

An agent briefed with a rich Client KB reads the job description through the lens of who’s actually making the hire.

3. Project Docs: the ground truth for this specific role

Workspace KB covers your team. Client KB covers your hiring manager. Project Docs cover this req, right now.

This is where the job-specific material lives:

  • Job description: The active JD the sourcing plan and qualification criteria are built from.
  • Intake notes: The context that didn’t make it into the JD, like the real reason the position is open.
  • Scorecard: The structured evaluation rubric that turns a candidate profile into a rank.
  • Other documents: Anything else that shapes the role, like compensation benchmarks or competitive landscape notes.

Project Docs are role-specific and finite. They’re the raw material the agents work from.

Why the layers matter together

Each layer answers a different question:

  • Workspace KB: How does our team work?
  • Client KB: What does this client actually want?
  • Project Docs: What is this role asking for?

An agent without the Workspace KB gives you generic output. An agent without the Client KB gives you accurate-but-wrong-for-this-hiring-manager output. An agent without Project Docs is guessing at the role.

When all three are in place, the AI executes with context. Sourcing plans reflect your real channels. Qualification criteria reflect what the hiring manager rewards. Outreach drafts sound like your team.

The practical upshot

The teams that get the most out of AI sourcing tools are the ones whose agents are well-briefed. A well-maintained Knowledge Base ensures your AI is always working with your best thinking.

Start with what you already know: your best recruiter’s instincts, your most effective outreach, and the one thing every hiring manager for a given client always says. Write it down once, and every agent inherits it.

That’s the point of the Knowledge Base. Not more AI, but smarter AI.

New to Calyflow? Create a free account to set up your Knowledge Base before your next search.

Related posts

Ready to run this as a workflow?

Calyflow turns this playbook into a repeatable workflow on your own AI, data, and tools.

Create free account