The Aleph AI Coaching Engine
How a home-services holding company got every field rep coached every day against its own sales and leadership playbooks — with no coach on payroll and nobody pressing buttons.
How coaching became a system, not a schedule.
The company already had everything a coach needs — recorded appointments, calendars, a written playbook. What it didn’t have was anyone with 40 spare hours a night to put them together.
Pocket recorders
Calendars
Transcription
Sales Playbook
Cloud storage
Rep phones
Coaching happened when schedules allowed.
One sales manager, dozens of reps, hundreds of appointments. Ride-alongs caught a fraction of a percent of the real conversations.
Coaching happens because it’s wired in.
Every recorded appointment is scored overnight against the company’s own playbook. The rep reads it with coffee.
The playbook became the rubric.
Nothing generic. The scoring model is the company’s own sales system, encoded section by section — the same language its managers already coach in.
Sales_Playbook_2026.pdf
GAM_Rubric.md
L10_Performance_Model.md
Canvass_Framework.md
You can hear the appointment going sideways the moment the price conversation starts before the value conversation finishes.
The reps who improve are the ones who see their own patterns — not the ones who get yelled at.
Every rep carries a pocket recorder. Audio syncs to the cloud on its own — the engine collects each rep’s day without anyone forwarding a file.
Each recording is matched against the rep’s calendar so every report is labeled with the client, the job, and the meeting type before scoring starts.
Scored reports render in the company’s brand and deploy to private URLs, with a push notification to the rep’s phone.
What the team wakes up to.
Nobody logs into anything. The coaching finds them — and each level of the company sees a different surface.
Discount offered before value was landed
Third appointment in two weeks with the same pattern. Worth a live ride-along on the next big bid.
Property walk skipped on a premium job
Score impact is visible in the close section. One coaching conversation likely fixes it.
Four straight days above team average
Worth a public shout-out — pattern shows the discovery work is driving it.
While the company sleeps.
One scheduled command fires the whole pipeline. Here’s one night, traced end to end.
Pull
Collected each rep’s recordings from their cloud lane — no forwarding, no uploads.
Transcribe & label
Audio became transcripts with the company glossary applied — products, people, and job types spelled right.
ⓘ glossary: company names enforcedCalendar match
Each recording matched to the rep’s calendar — client, job, and meeting type resolved before scoring.
Route & score
Sales bids to the 8-section model; GAMs, L10s, and canvass shifts to their own rubrics. Quarter-point discipline throughout.
Escalate
A recording that can’t be confidently matched to a meeting gets flagged for a human look instead of a guessed label.
Render & deploy
Reports rendered in company brand, deployed, and pushed to each rep’s phone.
“Sales Rabbit” misheard
transcription drift on a company term
Glossary entry added
the term now lands right, every night
New rep joins the company
A recorder and a calendar is all it takes — the nightly run picks them up automatically. First report the next morning.
↻ auto-included in the runPlaybook evolves
When sales leadership updates the model, the rubric is re-encoded — scoring language moves with the company, not behind it.
↻ rubric re-encodedPartner company added
Outside organizations ride the same rails, sealed to their own lane. Their reps see their reports — and nothing else.
↻ lane isolation enforcedCoaching stopped being a bottleneck.
The scarce resource — a senior coach’s attention — became software. What was occasional became structural.
Collected the day’s recordings across every rep lane
Speaker-labeled transcripts, company glossary applied
Recordings matched to calendar meetings; client + job labeled
One recording with no calendar hit — held for review instead of guessed
Reports scored on the company rubrics, quarter-point steps
Branded reports live at private URLs; push notifications sent
Their infrastructure
The pipeline, the reports, and the archive run on the company’s own accounts and domains. No third-party subscription in the loop.
Their data stays theirs
Field audio and transcripts live in the company’s own storage, sealed per lane — partners see their reps and nothing else.
The rubric is their IP
The playbook was theirs before; now it’s theirs in executable form. The encoding travels with the company.
Built like architecture
Documented, inspectable, and maintained the way you’d treat a building — not a black box that expires.
Want your field team coached like this?
One conversation. I’ll find the workflow, scope the build, and tell you honestly whether it’s a fit.