Systems we have built for real teams.

The problem, what we built and what changed. Client names stay under NDA.

01 · Case studies

Real problems,
working systems.

01

AI Sales Coach

B2B technology company · Sales team

Practice tough sales conversations before the real call.

The challenge

New reps were going into live buyer calls unprepared for technical objections, and managers had little time for one-to-one roleplay.

What we built

A private voice and text practice tool with realistic buyer personas - a sceptical CTO, a cost-focused CFO - that draw on the company’s own approved sales material.

How it is used

Reps rehearse whenever they need to and get consistent feedback. Managers review short summaries instead of running every drill.

Next.jsTypeScriptModel APIsVoice interfaces
AI Sales Coach practice desk with studio microphone and conversational HUD
On demand
PracticeReps rehearse any time, without waiting for a manager
Voice + text
RoleplayBuyer personas grounded in real product material
Review
Manager roleManagers review practice instead of running every drill
02

Sales Execution Intelligence

Consumer goods brand · Field sales

Store visits that reach sales leadership the same day.

The challenge

Field reports arrived through WhatsApp groups, spreadsheets and CRM forms. By the time managers pulled them together, the information was days old.

What we built

A visit app that works offline, with photos and voice notes, plus a service that sorts each visit and compares it with the account’s history.

How it is used

Managers start the day with a short list of accounts that need attention. Reps get clear next steps with the evidence attached.

Offline-first mobileFastAPIEvent queueManager dashboards
Field execution map and planning notebook across retail district
Same day
Field reportingStore visits reach managers the day they happen
Evidence
Visit recordsPhotos and next steps attached to each visit
One view
TerritoryAccount status in one place for leadership
03

Sports Performance Analysis

Sports academy · Coaching staff

Match video turned into clear, visual feedback.

The challenge

Coaches spent hours each week scrubbing through match video by hand to note movement, court position and signs of fatigue.

What we built

Computer vision tools for badminton and basketball that track players and shots from standard cameras and produce clips and movement heatmaps.

How it is used

Coaches review rallies with players on a tablet during breaks, instead of waiting days for manual video breakdowns.

Computer visionPyTorchWebGL chartsVideo pipelines
Sports court with computer vision tracking of player movement and shot paths
Same day
Video reviewReview during the session, not after manual tagging
Automatic
Clip taggingRallies and shots tagged from standard video
Heatmaps
Court coverageMovement and shot placement shown for each player

Start something similar.Prove it on your workflow first.

A 4-Week Working Proof puts a small, real version of the system in your team’s hands. Not sure where to start? A 2-week AI Opportunity Audit maps it out.

References available under NDA

  1. Pick one workflowThe one that costs you most time.
  2. Working in 4 weeksOn your real data, with your team.
  3. Run it every dayWe keep improving it with you.