Research Agent
User Testing Agents
Agents that write test scripts and play Seller, Sales Leader or CRO personas to pressure-test designs.
/ The Problem
Altify's help centre and documentation had grown organically over years and had become almost unusable messy, inconsistent, and far too technical for most of the people who needed it.
Sellers were raising support tickets for questions already answered in docs. Customer Success teams were spending hours digging for answers. New hires had no reliable way to self-serve. The documentation existed but nobody could find or understand what they needed.
Pain points
- Documentation scattered across multiple platforms with no unified search.
- Technical language written for engineers, not for the sellers who used the product daily.
- Support tickets and CS escalations for questions the docs should have answered.
/ The Solution
I designed an AI agent Altifypedia trained on all of Altify's product documentation. One place. Every answer. Tailored to who's asking.
The core insight was that the same question from a seller and from an engineer needed a completely different answer. A seller asking "how do I run a Test and Improve session?" needs plain, action-oriented language. A technical writer or engineer asking the same question needs the underlying data model, API calls, and configuration options.
Altifypedia handles both by detecting the user's role context and adjusting its response style accordingly without needing two separate tools.
/ How It Works
1
Knowledge ingestion
600 PDFs covering every product feature, release note, help article, and internal process doc were processed and embedded into the agent's knowledge base.
2
Role detection
When a user asks a question the agent identifies their role context seller, manager, engineer, technical writer either from their profile or from conversational cues.
3
Tailored response
Sellers receive plain-language, action-oriented answers. Engineers and technical writers receive full technical detail with relevant config options and API references.
4
Source citation
Every answer cites the source document so users can drill deeper if they want the full context building trust in the agent over time.
/ Impact
Altifypedia changed how the organisation relates to its own knowledge. Instead of searching, people ask.
600
PDFs ingested
1
Source of truth
2×
Roles served differently
Key outcomes
- Sellers can self-serve answers in seconds without raising a support ticket.
- Customer Success teams spend less time on internal knowledge lookups.
- New hires onboard faster with an always-available, jargon-free guide.
- Engineering teams get the depth they need without wading through simplified content.
- Received an AI Innovation Award for this self-initiated project.