Portfolio

Systems we built for real businesses.

Operations automation, voice and chat agents, websites, and the AI staff that runs our own company. Most of them run every day.

Listing-operations console after a morning run: prices and statuses updated by approved rules, every change verified in the MLS, and exceptions reported, across seven Texas MLS markets

Listing-operations console, shown with fictional data. Try the demo →

Featured · Texas new-home real-estate brokerage

Listing operations across seven MLS boards

Our agents keep a home builder's listings current across seven MLS boards. The MLS systems have no API, so the agents work the MLS screens the way staff would. Every weekday they compare the builder's data with every live listing, update prices and pending or sold status by rules the brokerage approved, post open houses, and check every change against a fresh MLS pull before it reaches the client's daily report.

Try the DemoRead the Story

The demo uses fictional listings and numbers. The real system runs every weekday for the client.

Verified price changes
11,000+
Verified price changes
Listings updated
4,000+
Listings updated
Status updates
2,500+
Status updates
Added in under two weeks
3 boards
Added in under two weeks
  • Daily comparison of builder data against every live listing
  • Price, status, and open-house updates made by approved rules
  • A fresh MLS check on every change before it is reported
  • A client portal with daily and weekly reports
  • Missing or conflicting data left alone, reported, and retried

Operations automation

Daily work that used to be done by hand in browsers, spreadsheets, and inboxes.

Multi-location event-rental franchiseRunning daily

Lead routing to the right store out of 22

Every website lead is matched to the right store by ZIP code and forwarded to that franchise by email within minutes, with text alerts to store owners. A backup router takes over if the primary stops, and a separate watchdog alerts the founders if any lead sits unrouted.

What we built

  • ZIP-code matching to 22 store territories
  • Email forwarding and store-owner text alerts
  • A backup router and an unrouted-lead watchdog
Stores served
22
Stores served
From web form to store
Minutes
From web form to store
Multi-location event-rental franchiseProduced monthly

Monthly KPI reports for every store

Each month we build a report for every store from the company's booking and phone data. Owners see revenue, call performance, lead response, the quote pipeline, repeat-customer revenue, and the year-over-year comparison without assembling it by hand.

What we built

  • Booking and phone-system data pulls
  • Per-store report covering revenue, calls, leads, and quotes
  • Year-over-year and repeat-customer comparisons
Store reports a month
22
Store reports a month
Multi-location event-rental franchiseDelivered

CRM, training, and invoice migration to a new ERP

We moved the company's customer history, eLearning content, and invoices off its old rental system and training platform onto a new ERP.

What we built

  • Customer, order, and invoice mapping
  • Training courses moved into the ERP's eLearning
Staffing and recruiting firmDelivered

Market intelligence for a staffing sales team

A market-intelligence agent watches job-market signals across metro markets, scores how well each account fits, and turns the results into ranked, sales-ready opportunity reports for the sales lead.

What we built

  • Job-market and career-page monitoring by metro
  • Account fit scoring
  • Ranked opportunity reports delivered to sales
Sales-ready reports
Ranked
Sales-ready reports

Voice and chat agents

Agents that talk with buyers, brokers, and visitors, with a person behind every judgment call.

National health-benefits companyLive for shoppers

Waverley: guided health-plan advisor

Waverley talks with shoppers by voice or chat, asks qualifying questions, shows plan options on screen, quotes from the client's own rating system, and routes each person to the right enrollment path. We built its sales playbook by ranking the company's recorded advisor calls and comparing what the best advisors did with what the rest did, topic by topic.

What we built

  • Voice and chat in one conversation
  • Quotes from the client's rating system
  • Compliance review of every customer-facing line
  • A separate QA owner who tests each version before release
Ranked advisor calls analyzed
1,690
Ranked advisor calls analyzed
Sales behavior cards
76
Sales behavior cards
Offshore staffing firmLive since September 2026

Website, Pearl voice guide, and partner program

We built the public website and Pearl, a voice sales guide that talks with visitors in the browser and captures leads. An automated reviewer checks every conversation afterward, and a watchdog alerts the team if Pearl stops answering. The partner-referral program gives each partner a tracked referral link, and partner emails go out in waves that pause automatically if complaints or unsubscribes climb.

What we built

  • Public website
  • Pearl, an in-browser voice sales guide
  • An automated review of every conversation
  • Tracked partner-referral links
  • Email waves that pause on complaint or unsubscribe thresholds
Reviewed after it ends
Every call
Reviewed after it ends
Partner referral links
Tracked
Partner referral links
National health-benefits companyDelivered

Support agent for a 250-broker network

A support agent answers partner brokers' plan and process questions over Slack and phone from a large plan-document library. It refuses anything member-specific and routes those questions to the right named person, so protected health information stays with people.

What we built

  • Slack and phone channels
  • Answers drawn from the plan-document library
  • Member-specific questions routed to named staff
  • A tested question set run before each release
Partner brokers served
~250
Partner brokers served
National co-living community brandDelivered

Conference seat-filling agent

A closer-type agent followed up conference leads, answered questions, sent checkout links, and applied approved discounts, filling seats without a person dialing.

What we built

  • Lead follow-up and question answering
  • Checkout links and approved discount codes
  • Handoff to a person for complex buyers
99 AgentsLive on this site

Barbara: our own voice sales agent

Barbara answers our chat, the voice panel on our homepage, and our phone line. She asks what the visitor is trying to solve, matches it to work we have delivered, and saves every lead for a founder to follow up. A separate reviewer audits her conversations, and a person approves every change to how she works.

What we built

  • In-browser voice, chat, and phone
  • Lead capture straight into our CRM
  • Post-call review with founder-approved fixes
  • Hard limits on call length and daily spend
Talk to her on the homepage
Live
Talk to her on the homepage
Talk to Barbara →

Websites and software

Sites, portals, and migrations we build and keep running after launch.

National health-benefits companyLive since June 2026

About 90 websites moved and kept running

We moved about 90 websites and domains, including the company's affiliate sites, off its old website platform onto modern hosting, with the old site ready as a fallback during the switch. We now run the sites. Staff ask for changes in Slack, and an agent builds the change, shows a preview, and publishes only after a person approves it.

What we built

  • Migration of about 90 sites and domains
  • Redirects so old links keep working
  • A Slack website operator with preview and approval
Sites and domains moved
~90
Sites and domains moved
How staff request changes
Slack
How staff request changes
99 AgentsLive

This website

We designed, built, and run get99agents.com, including Barbara's voice panel and chat, lead capture, protected client portals, and the reporting pages our customers use.

What we built

  • Design, copy, and development
  • Voice and chat built into the homepage
  • Protected client portals and dashboards

Our own company

We run 99 Agents on the same kind of staff we build for customers.

99 AgentsIn daily use

A company run with AI staff

Two founders and more than thirty named AI agents run 99 Agents across revenue, delivery, engineering, finance, legal, and operations. Every agent has a job, a manager, and a rule for when to ask a person. Patterns we deliver to customers are used here first.

What we built

  • Named roles with decision rights
  • Shared company memory
  • A weekly operating rhythm with human sign-off
Founders and AI agents
2 + 34
Founders and AI agents
See the org chart →

Bring us the job your team does by hand.

Tell us what repeats and where it gets stuck. We will show you how we would build it and run it.

Show Us the Work