Inside Riece Keck's AI-Assisted Recruiting BD Engine

The most useful part of Riece Keck’s story is not the revenue headline. It is the decision he made before choosing a single tool: design the operating system first.

Keck had already built a recruiting company through people and effort. When the market changed, that model proved fragile. His next version started with a different question. Which parts of recruitment require judgment, and which parts only require reliable execution?

This profile is based on Keck’s appearance on The Elite Recruiter Podcast, hosted by Benjamin Mena. The episode provides an official timestamped transcript. Revenue, pipeline and performance figures remain Keck’s statements unless noted otherwise.

A recruiting business that grew fast, then lost its market

Keck started his career at Robert Half before founding his own firm at 27. The agency focused on embedded recruitment for high-growth startups. It later added permanent technology recruitment as a way to diversify.

That positioning worked during the hiring boom. Keck says he personally billed about $1.2 million in 2021, while the firm generated between $1.8 million and $2 million in total revenue. New recruiters joined as demand kept rising.

The conditions changed during 2022. Technology companies cut hiring and reduced internal recruiting teams. Embedded recruitment, which had provided predictable monthly income, contracted sharply. Keck describes the business moving from about 40 people to four.

The agency was eventually folded into a larger business through a deal Keck characterizes as financially modest. He then spent part of 2024 leading talent for an AI startup on a consulting basis. That pause gave him space to reconsider how an agency should grow.

1. 2021

   A fast-growing recruiting business

   Keck reports $1.2 million in personal billings while the firm expands during a strong technology hiring market.

2. 2022 to 2024

   Demand falls and the operating model contracts

   The embedded recruitment market weakens, the team shrinks and the agency is folded into a larger group.

3. 2025

   A recruiting desk rebuilt around systems

   Instead of rebuilding headcount first, Keck creates a signal-driven business development workflow.

4. 2026

   From agency operator to software founder

   After the reported $800,000 year, Keck steps away from agency recruitment to focus on Rune.

This context matters. The workflow did not emerge from a clean technology experiment. It came from a founder who had seen a people-heavy model grow quickly and then become difficult to sustain.

The second agency started with a system, not another recruiter

Keck’s new thesis was simple. Build the machine first. Add people only after the process works.

That idea reverses a common agency pattern. A founder wins more work, hires another consultant and expects the new person to reproduce the founder’s habits. Each desk then develops its own searches, spreadsheets, messaging and follow-up rules. Growth adds capacity, but it also adds variation.

Keck wanted every future hire to multiply the same operating system. The repetitive work would happen consistently. Recruiters would spend more time on judgment, relationships and live conversations.

This is also the useful distinction in Leonar’s guide to recruiting automation. Automation is valuable when it removes a clear constraint. It is less useful when it scales a process that nobody has defined.

Before connecting software, Keck says he drew the workflow on paper. He mapped the sources, transformations and destinations. Building the first working version then took months.

On its first day, he reports that two positive replies became four job orders and more than $100,000 in pipeline. Pipeline is not revenue, and one day is not a benchmark. The episode still offers a useful lesson: the validation event was a qualified commercial response, not the number of contacts collected.

Inside the signal-driven business development workflow

The system did not start with a purchased list and a generic sequence. It started with changes that might indicate a recruitment need.

1. 01

   Detect a hiring signal

   Collect a new job, recent departure, leadership hire or funding event in the target market.

   Automation
2. 02

   Map the company and likely buyer

   Estimate which talent leader, founder or functional manager is most likely to own the hiring problem.

   AI assisted
3. 03

   Enrich the contact

   Find and structure the contact information needed for a permitted outreach workflow.

   Automation
4. 04

   Prepare a message from context

   Use the signal, company and buyer persona to create a relevant first draft.

   AI assisted
5. 05

   Review before enrollment

   Confirm the signal, contact, hypothesis and wording before adding the person to a sequence.

   Human
6. 06

   Move the response into a real conversation

   Keep discovery, advice, negotiation and relationship building with the recruiter.

   Human

The sequence resembles the connected workflow described in our guide to using AI agents for recruitment business development. The important difference is not the model that drafts the message. It is the evidence that determines who enters the workflow.

New job postings created the simplest trigger

Keck used Apify to collect roles from a job board relevant to his technology niche. The initial run looked back 60 days. A daily process then collected newly published roles.

The job, company and link moved through n8n into Clay. An AI step estimated the likely hiring-manager titles. Apollo then helped search for matching people and contact details.

This process automated research that a recruiter might otherwise perform tab by tab. It did not confirm that the company wanted an agency. A job advertisement only showed that a role existed.

Leadership changes revealed new decision-makers

Keck also monitored leadership hires across a target group of about 6,000 companies. He says this check ran monthly because it was relatively expensive.

A new engineering leader or CTO can change a team’s priorities. The event creates a timely reason to learn about the company. It does not prove that the new leader has budget, headcount or interest in an external recruiter.

Recent departures suggested possible backfills

Another process looked for people who had recently moved from a target company to a new employer. If the person had left within the previous 30 days, the old employer might need a replacement.

That is still an inference. The role may have been eliminated, redistributed or filled internally. A responsible message should ask whether the change created a need. It should not pretend the recruiter knows the company’s plan.

Funding events added context, not certainty

Keck used Crunchbase data to identify funding rounds from seed stage through Series D. Funding can support hiring, but it can also finance product development, debt, acquisitions or runway.

Our recruitment business development guide makes the same distinction. A trigger creates a reason to research and call. It should remain separate from a qualified mandate in the CRM.

A qualification template for hiring signals

An agency can borrow the logic without reproducing Keck’s entire stack. Use this table before placing any contact into outreach.

Field	Question to answer	Example
Signal	What changed, and when?	A Head of Engineering left 18 days ago
Evidence	Which public source confirms it?	Company profile and the person’s new role
Relevance	Why does this matter to our niche?	We recruit engineering leaders in this market
Buyer	Who is likely to understand the issue?	CTO, VP Talent or founder
Uncertainty	What do we still not know?	Whether the role will be replaced
Opening question	How can we test the hypothesis honestly?	“Has that change affected your hiring plan?”
Human owner	Who approves the contact and message?	Named recruiter responsible for the account

This template protects two things. It keeps weak signals out of the sequence. It also prevents AI-generated text from turning a probability into a false statement.

The stack mattered less than the architecture

Keck mentioned six main products while explaining the first version of his workflow.

Apify collected public web data. n8n moved information between steps. Clay structured enrichment and AI calls. Apollo supported company and contact research. Lemlist handled sequences. Crunchbase supplied company and funding information.

The product list will age. The architecture is more durable:

1. Collect a relevant signal.
2. Preserve the evidence and source date.
3. Identify the company and likely decision-maker.
4. Enrich only the information the workflow needs.
5. Prepare a draft that reflects the uncertainty.
6. Require a named reviewer.
7. Record replies, meetings and mandates in the system of record.

That final step matters. A fragmented stack can create more work if replies, decisions and outcomes never return to the CRM. A connected recruiting agency tech stack should preserve the context that explains why a prospect was contacted and what happened next.

Modern platforms can also expose approved CRM actions to assistants through APIs and MCP. Our guide to connecting AI agents to a recruiting stack explains the difference between giving an agent narrow tools and giving it uncontrolled access.

Human review was part of the design, not an exception

Keck told Mena that he checked the leads in Lemlist before anything went out. The official transcript describes a quick review across each prepared record.

This is a crucial part of the operating model. The system reduced the cost of reaching a reviewable draft. It did not remove accountability from the recruiter.

The same boundary appeared later in the episode. Keck predicted that agents would handle more sourcing, administration and system updates. He still expected recruiters to own relationships, phone conversations and closing work.

That boundary makes commercial sense. A model can compare fields and produce text quickly. It cannot own the promise made to a client. It also cannot carry the reputational cost of a false assumption or insensitive message.

The most advanced workflow is not always the one with the fewest human steps. It is the one that places human attention where an error would be expensive.

Most-placeable candidates became another signal source

Keck applied a similar process to candidate-led business development. A strong software engineer with fintech experience could be mapped to relevant fintech companies in the right location. The system then helped identify suitable contacts and prepare the campaign.

This is more useful than sending the same anonymous candidate summary to every company in a sector. The candidate’s background should connect to a plausible business need. Our guide to marketing a most-placeable candidate explains why the final argument still needs recruiter judgment.

The compliance boundary matters too. A candidate’s identity and personal information should not be disclosed without the appropriate basis and permission. An anonymized profile must remain genuinely anonymized.

Referral partnerships added a second growth engine

The episode was not an argument for cold automation alone. Keck also built relationships with businesses that served the same clients without selling recruitment.

He mentions an accounting firm focused on high-growth startups. Its team often met founders while they were building internal functions. Recruitment needs could surface during those conversations. Keck could also introduce his clients to the accounting firm when the fit was relevant.

Another source came from a recruiter who specialized in hiring for recruitment agencies. When that partner received an in-house talent request, the introduction could move to Keck.

He describes paying referral fees in some cases, including a 25% share on initial placements for one arrangement. That was his commercial practice, not universal guidance. Referral agreements can create legal, tax and disclosure obligations. Agencies should obtain advice for their jurisdiction before offering fees.

The broader method is safer and more transferable. Map the businesses that already advise your ideal client. Look for a genuine two-way benefit. Track introductions and outcomes in the client CRM.

A two-week pilot before automating the entire desk

An agency does not need six tools or 6,000 monitored companies to test the model. Start with one signal in one niche.

Use this pilot canvas:

Decision	Pilot choice
Market	One niche, location and company-size band
Signal	One observable event, such as a role reposted after 30 days
Daily volume	A reviewable number, such as five companies
Evidence rule	Source URL and date required for every record
Buyer rule	One approved set of functions by company size
Draft rule	Message must state uncertainty as a question
Review owner	One named recruiter approves every enrollment
Success metric	Qualified conversations and mandates, not emails sent
Stop condition	Pause if false positives or heavy rewrites exceed the agreed limit

Run the process manually for the first week. Record where research is repetitive and where judgment changes the outcome. Automate only the stable parts during week two.

At the end, compare five measures: research time, false-positive rate, editing time, qualified reply rate and mandates created. The test should tell you whether the workflow removed a constraint or added another dashboard to maintain.

The reported $800,000 year needs careful attribution

Keck says the rebuilt operation produced about $800,000 in its first year, without relying heavily on his previous client base. The episode does not provide audited financial statements or a channel-by-channel attribution model.

It would therefore be misleading to claim that one automation generated $800,000. His experience, reputation, sales ability, market selection, referrals and delivery all contributed to the business.

The result is still relevant as a founder’s account. It shows that the workflow operated inside a real recruiting business and produced enough confidence for Keck to keep developing the system.

He later left agency recruitment to build Rune. That makes this story different from a current agency profile. It is best read as the final operating chapter of a recruiter who turned an internal system into a new company.

The episode also contains sponsor performance claims. They are not part of this analysis because the source does not provide an independent method for checking them.

Four lessons for agencies building with AI

1. 01

   Map the workflow before buying tools

   Define the source, decision, output and accountable owner before choosing automation software.

2. 02

   Treat every signal as a hypothesis

   A job post, departure or funding event creates a reason to investigate. It does not confirm a mandate.

3. 03

   Measure review quality, not message volume

   Track false positives, editing time, qualified conversations and mandates instead of celebrating activity.

4. 04

   Keep the scarce work human

   Discovery, judgment, trust and negotiation remain the work clients associate with a strong recruiter.

Keck’s story does not prove that every agency should build its own automation stack. It shows why operating design should come before automation.

The same principle applies whether a firm uses several specialist tools or one connected platform. Start with the work you want to improve. Preserve the evidence. Put a person at the point of accountability. Then measure whether the system creates more useful conversations.

For firms assessing that model, Leonar’s workspace for recruiting agencies connects candidates, client companies, deals, outreach and recruiter-controlled AI workflows in one system. The product matters less than the operating question: does your stack help recruiters spend more time on the decisions clients pay them to make?

Questions about Riece Keck's AI recruiting workflow

How did Riece Keck use AI for recruitment business development?

He described a system that collected hiring signals, mapped likely decision-makers, enriched contact data and prepared outreach drafts. He still reviewed the leads and messages before they entered a sequence.

Which hiring signals did the system monitor?

The episode mentions new job postings, recent leadership hires, employee departures that might create backfills and company funding events. Each signal required qualification before outreach.

Which tools were in Riece Keck's recruiting stack?

He mentioned Apify, n8n, Clay, Apollo, Lemlist and Crunchbase in the first version. The workflow architecture matters more than this exact list because product features, terms and prices change.

Did AI run the entire recruiting agency?

No. The described system focused on repetitive top-of-funnel work. Riece Keck kept lead review, message approval, discovery, relationship building and closing under human control.

Is the reported $800,000 result independently verified?

No. The figure is a self-reported result from the podcast interview. Leonar has not audited the revenue, attribution model or contribution of each channel.