Skip to content
Live

Nurse Next Door asks years of franchise records in plain language, with the source shown

One knowledge brain over the full history of a 400-location franchise network. Answers carry their source, or the query that found them.

Client
Nurse Next Door
Sector
Franchising, 400+ locations
System
Knowledge brain, Data extraction, Client portal
Live since
Jul 2026
Last measured
15 Oct 2026
One point for every record in the Nurse Next Door knowledge brain, settling into one field, then a question lighting the few records its answer came from.

One point for every record in the brain. Each question lights the few records its answer came from.

  1. The records were complete. Nobody could ask them anything.

    Answering a question meant requesting a report, so most questions were never asked.

  2. Only someone who could write SQL could get an answer out of the system of record.

The records existed. Nobody could ask them anything.

Nurse Next Door had years of operating history across more than 400 locations inside its franchise-management system. Every lead, every status change, every document, agreement and territory, back to the beginning. The data was complete. It was also unreadable to anyone who did not write SQL.

So a question meant a report request, and a report request meant waiting. Most questions went unasked. Head office ran on side-of-the-desk reporting. The roadmap was clear. What was missing was the capacity to execute it without adding headcount.

The brief had a constraint we agreed with. The brain had to feed the systems the client already ran, not compete with them. Their dashboards stay the source of truth. Our job was to make the history underneath those dashboards answerable by anyone on the team, in plain language, with the source shown.

A full extract, a brain on top, and a pipeline the client owns

The first job was extraction. We pulled a full structured copy of the franchise-management system into the client's own database: leads, remarks, mail, campaign email, status changes, documents and calls, plus the franchisee-operations module with its agreements, territories, renewals and terminations. Every table in the system, not a sample, and every count reconciled against the source before we called it done.

The extraction toolkit is checkpointed. Each step records where it got to, so a restart resumes instead of starting over. The toolkit and a connector were merged into the client's own repository with a handoff document. The client owns the pipeline and can rerun it without us.

On top of the extract sits the knowledge brain, deployed on infrastructure isolated to the client, with its own repository, database and deployment, under the client's own brand. The extracted documents are indexed for retrieval with a citation on every answer. If the corpus does not hold the answer, the brain says so.

Knowledge systemsDocuments and SOPs feed one index. Every answer arrives with its source attached, and a question the corpus cannot answer is routed to a person with the gap logged.DocumentsSOPsIndexAnswerA personSource attachedNo source, gap logged
Documents and the structured extract feed one index. An answer arrives with its source, or the gap is logged and routed to a person.

The part that changes the conversation is plain-language questions against the structured tables. Ask how many leads moved to a given status last year and the agent writes the SQL, runs it against the real status-change tables, and returns the number with the query it ran. The person asking does not need to know a table name. They can read the query if they want to check it.

Around the brain: a custom assistant agent for the client's team, 7 jobs on a schedule, and 10 KPIs tracked in the portal against the objectives the client set.

Ask in plain language, get the table's answer

A person at head office logs into the portal, types a question, and gets an answer with its source attached. If the answer is in a document, the citation points at the document. If the answer is in the data, the agent writes and runs a query and shows the query. If the answer is in neither, the brain says so and names who to ask.

Underneath, scheduled jobs re-read the corpus and review answers. Most of the system's activity is this automated work: 815 runs in the 30 days to 20 August. 107 of 832 runs have failed since launch. We count those because a silent failure is how a brain goes stale, and we would rather print the number than hide it.

The client's team was invited on 2026-07-21. 10 of 11 invitations were accepted. The first client login came 4 days 19 hours after the repository was created.

Who uses it, how much, and how fast it went live

The brain's own table held every record in the extract when we read it on 20 August 2026, the count at the top of this page. The records themselves stayed in the client's database for this count, and none of them is described here.

239
questions answered in the 30 days since go-live

The first answer was written on 2026-07-21. In the 30 days to 2026-08-20 the brain answered 239 questions, counted from the tenant's answer table. Because the brain was exactly 30 days old, the all-time count and the 30-day count were the same number.

10 accounts were provisioned on the client's domain. 5 of them were active in the 30 days to 20 August 2026, read from the master user table. Half the team asking is a start. It is not the finish, and we do not print it as one.

The repository was created on 2026-07-17. The first client sign-in was recorded 4 days 19 hours later. That window covers provisioning, the structured extract, the index build and the invitations.

The systems keep working whether anyone asks or not. Scheduled jobs and curator passes ran 1,451 times in the 30 days to 15 October 2026.

What we measured

4 days 19 hoursfrom repository created to the first client sign-inSystem loggedby Advizr, 20 Aug 2026
How this was measured

From the moment the repository existed to the first time someone at the client signed in. Both ends are timestamps the systems wrote themselves.

Measured at Nurse Next Door knowledge brain and portal

5 of 10provisioned accounts active in the 30 days to 20 August 2026System loggedby Advizr, 20 Aug 2026
How this was measured

Accounts with activity since 2026-07-21, out of 10 provisioned on the client's domain.

239questions answered in the 30 days since go-liveSystem loggedby Advizr, 20 Aug 2026
How this was measured

Every answer row from the first answer on 2026-07-21 to 2026-08-20. All time and last 30 days are the same count.

10 of 11invitations accepted by the client's teamSystem loggedby Advizr, 20 Aug 2026
How this was measured

Invites sent 2026-07-21.

1,451automated runs in the 30 days to 15 October 2026System loggedby Advizr, 15 Oct 2026
How this was measured

Scheduled jobs and curator passes, with Advizr staff and test accounts left out.

107 of 832scheduled agent runs that failed, all timeSystem loggedby Advizr, 20 Aug 2026
How this was measured

Most runs are scheduled jobs and curator passes, not a person asking a question.

What broke, and how we know it is fixed

The brain went dark in front of the client on the night of the data load.

Why
The tail of the load pipeline re-enabled the source sync, which triggered a per-document enrichment cascade. That filled the database disk, burned all four of the resizes available in a day, and left the instance crash-looping.
The fix
A disaster-recovery instance stood up from nothing, and a guard shipped across the whole fleet so a load cannot re-arm the sync behind itself.
How we know
Live again in about three and a half hours: 341 schema migrations replayed in 5 minutes 24 seconds, a 17.6-minute reseed at 77 documents a second, and the same acceptance score afterwards. The original instance was then chosen as the keeper by counting what each one actually held rather than assuming.

After the switchback, every question came back as could not reach your brain.

Why
Search ordered by vector distance inside the same scan that joined the other tables, which blocked the index's early-stop plan. It scored every chunk in the corpus and blew the eight-second query timeout. The same query had been fine at 200,000 chunks and was broken at 368,550.
The fix
The ordering was lifted out of the joined scan so the index can stop early again.
How we know
Thirty seconds down to between one and five, with the returned rows byte-identical to the slow version.

What we did not measure

We did not measure hours saved on head-office reporting. A baseline was recorded in the portal but no after figure has been read yet, so we print neither. When it is read, it goes here with its date.

Built by

  • Andy ChenEngineer of record

What we would do again

  • If your operating history sits in a system nobody can query, we would extract it into tables first and index documents second. A status-change row answers questions a document never will.
  • If you are handing a pipeline to a client team, we would checkpoint every extraction step. A restart that begins from zero is the thing that makes a team stop running it.
  • If adoption matters more than the demo, we would count active accounts weekly from the first week. 5 of 10 is a number to work on, and we would rather print it than guess.

More stories

Every story

Run your whole company on AI.

Tell us how your business runs. If we're not the right fit, we'll say so on the call.

Your custom strategy and systems, delivered in under a month. Your named people using them every week by day 30. Or we keep working for free until both are true.

The page ends on a hand drawn chart of the sky that draws itself as it comes into view, with a small sail ship crossing its open sky from a far star to a beacon and a small planet.

© 2026 Advizr AI Inc. · Vancouver, BC, Canada

Delivered in under a month. Used by your named people every week by day 30. Or we work free until both are true.