Knowledge Agent
Altifypedia
A single source of truth trained on 600 product docs, with role-tailored answers for sellers and engineers.
/ The Problem
Altify's Test and Improve feature generates rich session data but that data was sitting unread because managers had no practical way to make sense of it at scale.
A sales manager overseeing a team of 15 sellers might have dozens of Test and Improve sessions running every week. Each session produces scores, answers, and timestamps. But extracting patterns who's struggling with which skills, which gaps are systemic vs. individual, what's improving required hours of manual review no manager had time for.
The result was that Test and Improve ran in the product but coaching conversations remained generic, and the data that could have changed that sat unused.
Pain points
- Session data existed but required manual aggregation to turn into actionable coaching recommendations.
- Managers couldn't distinguish individual performance issues from team-wide skill gaps without spending significant time in the data.
- The feedback loop between session completion and coaching conversation was too slow to drive behaviour change.
/ The Solution
I designed a single AI agent that covers the full T&I lifecycle — preparing the seller before the session by running a complete mock Test and Improve with simulated attendees, then processing the real session transcript afterwards and writing all outputs directly back into Altify.
Before the meeting, the agent reads the Relationship Map to identify every attendee, derives their mood from their Support value (Mentor, Supporter, Non-Supporter, Enemy), and role-plays each reviewer in their assigned perspective — Customer, Competitor, Leadership, or Partner. It runs through all three T&I phases — Understanding, Assessing, Improving — raising the exact questions and vulnerabilities each persona would raise. The seller walks into the real session already knowing what to expect.
After the mock Test & Improve meeting, the same agent retrieves extracts every Question, Vulnerability, and Recommendation with an importance rating, and creates PRIME actions in Altify. Nothing requires manual entry.
The entire session can be run by voice. The agent listens to the seller's speech, filters out background noise — music, other conversations, non-English speech — and only responds to what is clearly directed at it. If a turn is pure noise it waits silently. If speech is unclear it asks once to repeat. The seller stays focused on the conversation, not on typing.
/ How It Works
1
Read the plan and map the room
The agent identifies the opportunity or account under review, reads the Relationship Map to list every customer attendee, and asks the seller to confirm who is joining. Each person gets a mood derived from their Support value — Mentor, Supporter, Non-Supporter, or Enemy — which shapes how aggressively they challenge the plan.
2
Run the mock T&I by voice
The agent role-plays all reviewers simultaneously, following the real session flow. The seller can speak naturally — the agent classifies every transcription, keeping only speech clearly directed at the session and silently ignoring ads, music, or side conversations in the room. In the Understanding phase it asks the questions each persona would ask. In the Assessing phase it raises vulnerabilities in character. In the Improving phase it pushes for specific next steps.
3
Debrief before the real session
The mock closes with a ranked list of vulnerabilities and gaps by severity, the hardest questions the seller should prepare for, and a set of suggested PRIME actions ready to bring into the real meeting.
4
Post-session: transcript to Altify
After the mock Test & Improve meeting, the same agent retrieves extracts every Question, Vulnerability, and Recommendation with an importance rating, and creates PRIME actions in Altify. Nothing requires manual entry.
/ Impact
Managers went from ignoring session data to running data-driven coaching conversations every week.
0
Hours of manual analysis
3
Prioritised actions weekly
Weekly
Coaching cadence enabled
Key outcomes
- Coaching conversations shifted from general impression to evidence-based specifics.
- Systemic skill gaps identified across the team that individual sessions would have missed.
- Managers spend time on coaching, not data wrangling.
- Test and Improve feature adoption increased as teams saw visible output from their sessions.