HelpGhost Mission Report 06.16.26
MISSION: Automate the capture and centralization of knowledge across the MSP.
TARGET: The knowledge layer.
OBSTACLE: Tribal knowledge.
STATUS: Objectives met. Mission ongoing.
We rarely take a break from building to talk about what we've built, but we're making an exception.
Almost a year in the trenches, shipping. Not alone, though - we've been getting feedback and insight from our MSP partners who live with these challenges day in and day out. This is where we document what we've shipped in just the last 6 months, why it matters, and where we're headed from here.
This is our Mission Report. Every single milestone below is real, dated, and part of the product available today.
Point of origin
Every MSP struggles with fragmented or scattered knowledge that's lives in historically difficult to reach or inefficient silos.
One prime example is one senior tech's head or what we call the the 'Default Search Engine' problem where the senior tech is a single point of failure, they are constantly pinged on Teams or tapped on the shoulder but if that tech decides to move on or even step out for a day there's a significant gap and rediscovery becomes the deFacto option.
Then there's the KB problem. Too much is buried in a knowledge base that nobody ever opens. A digital filing cabinet. That knowledge is technically stored, but functionally useless.
That's tribal knowledge problem, and it's the price every MSP pays. New hires take weeks to get up to speed, and the same questions get asked a hundred times. When a key tech leaves, the knowledge goes with them and it's only compounded by rising turnover in MSPs and other knowledge and support based orgs.
Our mission was and continues to be fix where knowledge lives - to automatically capture it the moment a ticket gets solved, keep it centralized, and serve it back up at the point of work. That destination's got a name - the knowledge layer - and the objective of this operation was to occupy it...
Flight path 01 - HelpGhost Scholar
Capture → Comprehension
Scholar is the knowledge engine. It's where knowledge gets created, gathered, and kept current. It owns the whole process: capturing new knowledge automatically and centralizing what's scattered across the MSP/Org.
Capture
The foundation the entire operation stands on, working in two motions.
Create at resolution. Scholar turns resolved tickets into knowledge base entries automatically, the moment a ticket closes, with no extra work from the technician. The reason MSP documentation never gets written is that it's a separate job nobody has time for. Scholar removes the separate job. Knowledge becomes a byproduct of the work itself, captured every time a problem gets solved.
Consolidate what's scattered. Most MSP knowledge isn't missing, it's spread across systems and technician brains. Scholar ingests and unifies it into one searchable layer: currently PSA data and IT Glue, Halo KB, Confluence, Hudu, and manual uploads across DOCX, PDF, and HTML. Across this mission we hardened that ingestion pipeline to hold up at real MSP scale, with more fault-tolerant IT Glue exports, direct links back to every source document, and datasource editing that updates settings in place instead of forcing a delete-and-re-add. This is the plumbing that makes "centralize" actually work in production.
Comprehension
This mission deepened how well Scholar understands the knowledge it surfaces and captures.
Image processing across every knowledge base (01.11.26)
MSP documentation is full of screenshots, diagrams, and photos. Scholar can now read them. It transcribes and understands images across IT Glue, Halo KB, Confluence, and uploaded documents (DOCX, PDF, HTML) and screenshots, turning visual knowledge that used to be invisible to search into something a technician can actually find. To date, the system has transcribed nearly 7,000+ images and counting.
Traceable, sourced knowledge (12.03.25)
Every answer Scholar surfaces links straight back to where it came from, whether that's an entry in your knowledge base or a downloadable doc. Captured knowledge a technician can audit and trust, never a black box.
Flight path 02 - HelpGhost Hero
Answers → Actions → Automation
Our assitive AI, Hero, lives inside the PSA, where your technicians already work. Over this mission it evolved from a tool that surfaced answers and knowledge into one that does the work.
Below are the updates and features releases over the last six months:
Contextual web search and inline source citations (12.03.25)
Hero can pull real-time information from the web with full context of what the user is working on and has done up to the point of search. Eery answer ships with hyperlinked sources, web results link back to the original, knowledge-base answers show the exact entry used and link to the source or a downloadable doc. No black-box answers. A tech can always see where the information came from and trust it.
The in-ticket experience leap - Hero Sidekick (01.11.26)
Hero stopped being trapped in a cramped pod or a separate tab that blocks the work. Sidekick gives technicians persistent chat history across tickets and a workspace that stays open beside them. Anyone who has fought an embedded PSA widget knows exactly why this matters.
Custom AI instructions, personal and org-level (02.23.26)
Teams can now shape how Hero behaves for org specific and individual needs. An individual tech can ask for more detailed, step-by-step answers or shorter ones. At the organization level, admins can load in shared context once, like SOPs, acronyms, infrastructure details, and more, so every chat across the company starts with that knowledge built in. This is per-user context that gets more valuable over time.
Faster, richer, hands-free (03.18.26)
Hero streams responses in real time with live status updates, so it feels dramatically faster and shows its work as it runs. Users can attach up to five files (images, docs, PDFs) for context, and use voice-to-text to capture notes and get answers without typing.
Private chats - a secure alternative to public AI and installable app (04.27.26)
A question no longer means pasting client data into ChatGPT. A user can start a private chat for a question with no ticket attached, with the full power of Hero hooked into the organization's knowledge, and kept inside your environment. Hero also installs as a desktop app (PWA) for anyone who wants it living in its own window instead of a browser tab.
PSA ticket note action (03.18.26)
The first move from informational tool to action tool. A user can ask Hero to draft an internal or public note, review it, and post it straight to the ticket on confirmation. Hero stopped only telling you things and started doing them.
HelpGhost Automations, powered by Bumblebee (04.01.26)
Through our integration partnership with Bumblebee, Hero can now execute real workflows like M365 user onboarding, invoice audits, service desk automations, and QBR report generation (Available as an add-on capability via the Bumblebee integration).
Quality-of-life and reliability (03.06.26, 06.10.26, ongoing)
Hero usage leaderboards so teams can see who is and isn't getting value and who could use a hand. Suggested prompts, an assigned-ticket view inside Sidekick, one-click message copy, proper table rendering, and configurable initial-chat behavior. Underneath it all, a steady stream of stability work: Paging fixes, datasource queueing improvements, and integration hardening.
The Human layer
Achievements (06.12.26)
As technicians work in Hero and Scholar, they now earn achievements, rarity tiers, and monthly season awards for the top contributors. Top Cited. Top Chatter. Top Reviewer. Top Creator.
This is all by design, it's what we're aiming for. The system is here to make your technicians as effective as possible, not to replace them. We're talking revival, not replacement. Every single milestone in this report is about freeing up working people to do the work that actually needs human touch, and the human layer is where we celebrate the people doing it.
Proof of Mission
Measured results, not theory.
Tribe MSP put their service-desk champions on HelpGhost and saw a 25% jump in throughput, with roughly 4,000 answers captured straight from solved tickets.
Equinox IT Services ran it head-to-head against other solutions. Per co-founder Steve Telford:
- Real answers, not a document pile. Techs get the exact solution for the call instead of "a stack of documents" to dig through.
- Without it, new-tech call times "would more than double."
- Beats public AI. Results came in "above and beyond" what techs got from shadow IT like ChatGPT and Copilot, and better still once their existing documentation was centralized into Scholar.
- Unanimous winner. Tested against the other tools, "HelpGhost wins hands down," across every technician.
- Day-one value, no real learning curve, and no downtime. Its biggest impact so far has been retaining existing clients by solving their issues faster.
Mission Log
12.03.25 Web search · inline source citations · richer answer formatting
01.11.26 Scholar image processing (~7,000 images) · Hero Sidekick · in-app
feedback widget + public roadmap · IT Glue export hardening
02.23.26 Custom AI instructions (personal + org-level) · Scholar datasource
editing and quality-of-life fixes
03.06.26 Hero usage leaderboards
03.18.26 Streamed responses + live status updates · file attachments ·
audio input · PSA ticket note action
04.01.26 HelpGhost Automations powered by Bumblebee (add-on)
04.27.26 Private chats (secure alternative to public AI) · installable
app (PWA)
06.10.26 Hero batch: suggested prompts · assigned-ticket view · message
copy · table rendering · stability + integration fixes
06.12.26 Achievements — the human layer
Mission status
Objectives met. The knowledge layer isn't a concept anymore. We built it. It's shipped, in production, and this is just the start...
This is HelpGhost v3. Our next mission is already under way, and it will change how MSPs handle knowledge and workflows for good.
One transmission remaining...
The knowledge engine's next evolution becomes public 06.17.26
The mission continues 👻