Lyrion has joined the Open Secure AI Alliance, alongside

NVIDIAMicrosoftPalantirIBMAdobeRead the announcementRead more
Case studies

The systems we build, inside the businesses that run them.

Four of the systems Lyrion builds, in detail: what was breaking, what we found when we looked, what we built around it, and what changed for the team running it. Every one is a custom build around one company's bottleneck, not a product off a shelf. The industries are examples, not limits.

The studies on this page are illustrative: company names, people and figures are representative, shown while client sign-offs land. We publish a named case study only with the client's written approval, and approved studies replace these as they land. In a strategy session we walk you through the real work directly.

Hartwell & Sons ConstructionConstruction · 140 people
Minutesfrom request to first response, was next-day at best
30+ hof senior estimator time returned, every week
3xquote capacity, same two estimators
3 weeksfrom first walkthrough to live
Custom system · AI quoting and operations system

Quotes that took six to nine days now leave the same day.

Construction140 peopleLive in 3 weeks

A commercial general contractor running office fit-outs and light industrial builds across the region.

Runs on
outlookgoogledrivemondayand the tools they already had

Before Lyrion

Every quote ran through the same two senior estimators. Requests landed in a shared Outlook inbox, takeoffs were retyped between spreadsheets, and sub pricing was chased by phone and buried in email threads. A quote left the building in six to nine days, and the jobs kept going to whoever answered first.

The knowledge was the deeper problem. Six decades of pricing instinct lived in two heads, and both of them were booked solid reviewing work nobody else could start. The company was not short of work. It was short of the two people every quote had to pass through.

What we found

  • Roughly 60 percent of every quote already existed in a past project, but nobody could find the right past project fast
  • Sub pricing was scattered across a dozen email threads per job, with no single sheet that held the current numbers
  • There was no record of which quotes were lost or why, so pricing never learned from its misses

The system we built

Quote intakePrice engineSub chaseEstimator reviewOperations assistant
  • Quote intake. reads each request, files the drawings, opens the job and confirms receipt to the client in minutes
  • Price engine. drafts the estimate from the price book and the closest past projects, every line traceable to its source
  • Sub chase. sends the requests, runs the reminders and lands every sub number in the estimate sheet itself
  • Estimator review. the seniors open a finished draft and correct it instead of assembling it
  • Operations assistant. carries the job after the win: submittals, change-order paperwork, the weekly client update

Rollout

One week inside their estimating room mapping how the two seniors actually price, then built directly on the Outlook, Drive and Monday setup the team already ran. Nothing was replaced. The first live quote went out inside three weeks with an estimator watching every step.

Week 1Inside the estimating room, mapping how the two seniors price
Week 2Built on their Outlook, Drive and Monday, nothing replaced
Early week 3Pilot, an estimator checking every draft line by line
End of week 3First live quote out the door, same day it arrived

What changed

The same two estimators still price every job. What changed is what reaches them: a finished draft with every line traceable to its source, instead of a blank sheet and a full inbox. Quotes answer first more often than not, sub numbers land in the sheet on their own, and sixty years of pricing instinct now lives in a system the whole company can use. The desk carries three times the quote volume without a single new hire.

Quote turnaround
Before6 to 9 days
With Lyrionsame day

“The first month it drafted a quote I would have priced the same way to within a rounding error. That is when I stopped checking every line.”

Dale HartwellDale HartwellPresident · Hartwell & Sons Construction
Meridian Talent GroupStaffing · 85 people
Minutesfrom new req to ranked shortlist, was hours of sourcing
3xsubmittals per recruiter, per day
2 daysa week of timesheet chasing, gone
4 weeksfrom first walkthrough to live
Custom system · AI recruiting operations system

Every new req starts as a ranked shortlist, not a blank search.

Staffing85 peopleLive in 4 weeks

A staffing agency placing accounting, admin and light industrial roles from three offices.

Runs on
salesforcegmailslackand the tools they already had

Before Lyrion

Recruiters were doing six jobs at once: reading reqs, sourcing, writing submittals, scheduling interviews, chasing feedback and chasing timesheets. The ATS held years of candidates nobody searched, because searching it was slower than sourcing fresh. And in staffing, the first agency to put a good candidate in front of the client usually wins.

The system we built

Candidate matcherSubmittal drafterInterview coordinatorTimesheet chaser
  • Candidate matcher. mines their own ATS first and returns a ranked shortlist with the reasoning attached
  • Submittal drafter. writes the submittal in the house format from the resume and the req, ready for a recruiter's pass
  • Interview coordinator. offers slots, confirms both sides, reminds, and reschedules the fallouts
  • Timesheet chaser. collects, reminds and escalates so Friday billing goes out on Friday

Rollout

Week 1Desk-side with recruiters in two of the three offices
Week 2Built on the Salesforce, Gmail and Slack they already ran
Week 3Pilot on one desk, recruiters scoring every shortlist
Week 4Live across all three offices

What changed

Recruiters open the day picking from a shortlist mined out of their own ATS, submittals go out in the house format within the hour of a req landing, and Friday billing goes out on Friday. The database they spent years filling finally earns its keep.

Time to first submittal
Before2 days
With Lyrionunder 1 hour

“The system reads every new req, pulls the candidates we already have, and drafts the submittals. My recruiters start the day picking from a shortlist instead of building one.”

Renata VossRenata VossDirector of Recruiting · Meridian Talent Group
Beacon Ridge RealtyReal estate · 60 people
Secondsto first answer, was 9 hours
24/7every inquiry answered, nights and weekends
0leads dropped without an outcome
2 weeksfrom first walkthrough to live
Custom system · AI lead qualification and follow up system

Every inquiry answered in seconds, at any hour.

Real estate60 peopleLive in 2 weeks

A residential brokerage covering three counties, listings and buyer representation.

Runs on
gmailgooglecalendartwilioand the tools they already had

Before Lyrion

Portal inquiries arrived around the clock and waited for the next working day. Agents answered the leads they liked and let the rest age; follow up usually stopped after one attempt. Nobody was lazy, the leads simply arrived faster than humans between showings could answer them.

The system we built

AnswerQualifyBook the viewingFollow up to outcome

Lyrion built one system with four jobs: answer every inquiry in seconds at any hour, qualify it in a short natural exchange, book the viewing straight into agent calendars, and keep following up until there is an outcome. The agent walks in with the whole exchange summarized on one screen.

What changed

First response to an inquiry
Before9 hours
With Lyrionunder a minute

“An inquiry at 11pm gets a real answer at 11pm. The viewing is booked before another brokerage has opened the email.”

Marcus OkaforMarcus OkaforManaging Broker · Beacon Ridge Realty
Forge & Field CreativeAgency · 38 people
Minutesper deck, was half a day of assembly each
14decks drafted before the team sits down
2 daysreturned to the senior team, every month
3 weeksfrom first walkthrough to live
Custom system · AI client reporting and workflow automation

Reporting week became a Tuesday afternoon.

Agency38 peopleLive in 3 weeks

A performance marketing agency running paid media and content for 14 retained accounts.

Runs on
slackgoogledrivehubspotand the tools they already had

Before Lyrion

The first two days of every month belonged to reporting. Account managers pulled numbers from six platforms, pasted them into decks and wrote the same commentary structure client by client, fourteen times.

Between the monthly cycles, internal status lived in people's heads, so the week also carried a meeting whose only purpose was reconstructing who was doing what.

What we found

  • Fourteen accounts at roughly half a day of assembly each, every month, done by the most senior client-facing people
  • The commentary followed the same structure every time; only the numbers and the anomalies changed
  • Anomalies were usually discovered by the client, because the person assembling the deck had no time left to read it

The system we built

Report assemblerCommentary drafterReview routerStatus board
  • Report assembler. pulls each account's numbers from the platforms into the agency's own deck template
  • Commentary drafter. writes the first pass in the agency's voice and flags every anomaly it finds in the data
  • Review router. sends each deck to its account manager, collects the edits and releases the send
  • Status board. reads the project tools and keeps a live who-is-doing-what, so the meeting stopped being a survey

Rollout

Week 1Inside a live reporting cycle as it ran
Week 2Built on their Slack, Drive and HubSpot
Early week 3Parallel run against the hand-built decks
End of week 3Live for all 14 accounts

What changed

The decks assemble themselves from the platforms and arrive drafted in the agency's voice, with the anomalies flagged before a client can find them. The status meeting stopped being a survey. The senior people got their two days back, and the clients noticed the difference before anyone told them.

Reporting time per month
Before2 full days
With Lyrionone afternoon

“Reporting week used to swallow my senior people. The decks draft themselves from the platforms now and my team spends the time on clients.”

Priya RamanPriya RamanCOO · Forge & Field Creative

Could a system like this run inside your company?

Bring us one bottleneck. We will tell you honestly whether a custom system is worth building around it.

Get your AI opportunity assessmentRead what clients say