Resolved, one message
A food manufacturer's DataHub v11 upgrade broke an ODBC DSN. jonjon spotted the 32-bit versus 64-bit mismatch and closed it in a single reply, five-star review attached.
jonjon.aias a Service
For OT software vendors, equipment OEMs, distributors and integrators
The best first line of support in OT. Now yours to run.
jonjon.ai is the AI support engineer Software Toolbox built to serve its own 6,000 customers in 95 countries. It answers in about a minute, closes most tickets without an engineer, and escalates the rest with the diagnosis already done. jonjon.ai as a Service puts it in front of your queue: your products, your documentation, your brand.
- tickets a month reaching human engineers, 2021 to 2023 peak vs July and August 2026
- 400 to <100tickets a month reaching human engineers, 2021 to 2023 peak vs July and August 2026
- tickets jonjon takes are closed without an engineer, September 2026
- 6 in 10tickets jonjon takes are closed without an engineer, September 2026
- to a grounded first answer in the customer's inbox
- ~1 minto a grounded first answer in the customer's inbox
- CSAT on tickets served, October 2026
- 100%CSAT on tickets served, October 2026
jj
jonjon
Ticket 48112 · Cogent DataHub v11 · ODBC DSN fails after upgrade
Diagnosis
92% confidenceThe System DSN was created in the 64-bit ODBC administrator, but DataHub v11 on this machine runs the 32-bit ODBC stack. A 32-bit driver cannot see a 64-bit DSN, so the connection fails on open.
Steps
- 1Open the 32-bit ODBC Data Source Administrator (SysWOW64\odbcad32.exe).
- 2Recreate the System DSN with the same name and credentials.
- 3Restart the DataHub service and re-test the ODBC point.
Sources
DataHub v11 help: ODBC points · FAQ 1143: 32-bit vs 64-bit DSN
This resolved itEscalate to the support team
Results
Measured on a live support operation, not a benchmark
jonjon has been in the loop on every Software Toolbox support ticket since April 2026, across ~6,000 customers on TOP Server, OmniServer, Cogent DataHub, OPC Router and N3uron. These are the numbers our customer success team reports every Monday.
Tickets a month handled by human support engineers
Software Toolbox support, monthly. Routing to the Activation Center went live in March 2026; jonjon in the loop in April 2026.
- ~4002021 to 2023 peakbefore jonjon
- ~220Early 2026 basebefore jonjon
- <100Jul 2026with jonjon
- <100Aug 2026with jonjon
- 89Sep 2026with jonjon
- 71Oct 2026 to datewith jonjon
60%
fewer tickets reaching engineers year on year: 89 in September 2026 against a historical base of about 220 a month
40%
drop in ticket volume to engineers in the first full month live, April 2026
60 to 40%
escalation rate on tickets jonjon takes, May to September 2026. It gets better every month it runs.
#1
closer on our support leaderboard: 37 tickets in October 2026, the next engineer 11
+50%
organic jonjon chats in one month: customers now go to jonjon before they open a ticket, October 2026
40 to 20%
licensing share of the queue once those tickets are routed to a self-service portal, so the deflection rate you pay for is real
Serving our largest oil and gas customer used to be almost a full-time job for one engineer. That has gone away. Their power users troubleshoot OPC UA issues in jonjon before they ever open a ticket.
Closed, no escalation
A systems integrator could not find a property shown in a driver's documentation. jonjon explained what it does and why their setup did not need it. Marked resolved by the customer.
Engineer time saved
A global chemicals company uploaded a middleware XML file believing it was a TOP Server project. Eight messages in, jonjon identified the format and the missing file, before an engineer ever picked it up.
Tickets that never happened
A national beverage company worked through SAP and OPC Router questions in jonjon during evaluation. None became tickets. The review they left was the first thing our sales team read that week.
What it is
An AI support engineer, not a chat widget
jonjon sits inside your support process. It reads the ticket, works the problem with the customer, and either closes it or hands it to your engineers with the work already done.
Grounded in your knowledge
Manuals, help files, FAQs, KCS articles, resolved tickets, even training videos converted into documentation. Every answer cites its sources and carries a confidence score. When jonjon does not have a source, it says so and offers escalation rather than guessing.
In the loop on every ticket
Email, web form, phone-logged or API. An intent classifier routes licensing to your portal and pre-sales to your sales team, so jonjon only takes real support. The customer gets a first reply in about a minute with a button to continue the conversation, upload logs and screenshots, or escalate.
Escalates like a good engineer
Guided mode asks the clarifying questions, reads log files and screenshots, and remembers the user's OT environment between sessions. An escalation arrives with the diagnosis, what was tried and the files attached. Your engineers start at step five, not step one.
Why jonjon
Why it is the best first line of support in OT
General-purpose copilots know what a PLC is. jonjon knows why your customer's Modbus poll is hanging at 03:00, because it has closed that ticket before.
01
30 years of industrial connectivity
Built by Software Toolbox, the company behind TOP Server, OmniServer, Cogent DataHub, OPC Router and N3uron. OPC UA, Modbus, Allen-Bradley, Siemens, BACnet, DNP3, MQTT and Sparkplug B, historians and MES are home turf, not edge cases. Troubleshooting skills decode project files, parse event logs and plan migrations.
02
Trained on 250,000 real OT tickets
Not a general model with a PDF bolted on. jonjon learned on a quarter of a million resolved support cases from an operation running 97% CSAT and an NPS between 87 and 93. Then it learns your products the same way.
03
Proven on the hardest audience
Controls engineers at a Fortune 200 oil and gas producer, a global furniture manufacturer, a global chemicals company and hyperscale data centre operators use it in production. People who do not adopt new tools adopted this one.
04
Honest deflection maths
Around 40% of OT support tickets are licensing and activation. We route those to self-service first and count them separately, so the resolution rate you pay for is measured on tickets that needed an engineer.
05
Humans where it matters
Safety-sensitive equipment (high pressure, molten material, energised systems) and production-down language go straight to your engineers. Out of scope defaults to escalate. The escalate button is always on screen.
06
Enterprise governance
Never trained on your data or your customers' data. Personal information is scrubbed before a ticket reaches the model. Retention and deletion are yours to set. Single sign-on for your team.
How it works
From inbox to resolution in five steps
The same workflow that runs Software Toolbox support today, configured for your products in the Partner Console.
- 1
Intake.
Your support email, web form or ticket system. Live today with HubSpot and SAP Cloud for Customer; other systems through the API.
- 2
Triage.
An intent classifier routes licensing and activation to your portal, pre-sales to sales, and real support to jonjon. Production-down language bypasses the queue to a human.
- 3
First reply.
The ticket is scrubbed of personal data, matched against your documentation, and answered in about a minute: diagnosis, steps, sources. The customer continues on your branded jonjon.
- 4
Resolve or escalate.
The customer marks it resolved, or it closes after five business days with no escalation. Anything else lands with your engineers with the diagnosis, the attempts and the files.
- 5
Learn.
Thumbs-down, escalations and low-rated chats feed a knowledge-gaps view. Your team fixes the article, jonjon retrains. That loop took our escalation rate from 60% to 40% in five months.
The service
What jonjon.ai as a Service includes
Onboarding and ingestion. Your manuals, help files, FAQs, KCS articles, resolved tickets and videos, structured into a tenant of your own.
Eval gates before go-live. We run jonjon against your real tickets and your engineers grade the answers. Nothing ships below the bar you set.
Partner Console. Intake channels, documentation, escalation rules, sites and OT environments, ticket views by region or domain, and the metrics: resolved, escalated, hours returned to your team.
Your brand, your domain. White-labelled replies from your support address, your logo, your tone of voice.
Ticket system integration. Statuses sync both ways. Your system of record stays the system of record.
Multi-language. Your customers write in their language; jonjon answers in it.
A team that runs support for a living. Software Toolbox customer success engineers review your knowledge gaps with you every month.
Who it is for
Anyone who answers OT support tickets
OT software vendors
Drivers, connectivity, SCADA, historians, data platforms. Your docs become a first line that scales with releases, not headcount.
Machine and equipment OEMs
Dispensing, packaging, robotics, CNC. Electrical and controls tickets handled; mechanical and safety-sensitive cases kept with your people.
Distributors and channel partners
See every ticket across the brands and regions you support, filtered by domain, and route the hard ones to the right engineer.
System integrators
A first line for the plants you support, with each site's OT environment saved so the context is there before the call.
Pricing
Pay for resolved tickets, not seats
Three parts. One of them only costs you when jonjon has done an engineer's work.
Onboarding
Ingestion of your documentation, eval gates on your real tickets, ticket system integration and branding. Billed against milestones: 40% at kick-off, 40% at eval sign-off, 20% at go-live.
Platform
Your tenant, the Partner Console, hosting, model usage, retraining and the monthly knowledge-gap review with our customer success team.
Resolution meter
20 to 30% of your human cost per ticket
Charged only for tickets jonjon resolves on its own. Escalated tickets cost nothing. Resolved means closed by jonjon with no escalation within five business days: written into the contract, audited in your console.
What it could save you
Move the sliders to your own numbers. The meter is set at 25% of your cost per human-resolved ticket.
100 to 5,000
0 to 70%
$40 to $400
10 to 70%
Conservative 20%, base 30%, optimistic 40%. Software Toolbox runs at 60% after six months.
Net saving a year
after paying the meter, before onboarding and platform fees
- support tickets a month after routing
- support tickets a month after routing
- tickets jonjon resolves a year
- tickets jonjon resolves a year
- meter per resolved ticket
- meter per resolved ticket
- tickets jonjon resolves a month
- tickets jonjon resolves a month
Illustrative only. Software Toolbox's own cost to serve a human-handled ticket is $110 to $135. A connectivity vendor handling 1,200 tickets a month that we sized in 2026 modelled $383K to $703K in net savings a year across the three scenarios.
How to engage
Start with three tickets
No slideware. We demo on your tickets, pilot against your baseline, and scale when the numbers say so.
- Stage 0Free · about 10 days
Demo on your tickets
Send three representative tickets and the documents your engineers would use to solve them. We come back with jonjon answering them live, in a working session with your support team.
- Stage 1Paid pilot · 60 to 90 days
One queue, measured
One product line or one ticket queue, run against an audited baseline and the agreed definition of resolved. Eval gates before go-live. You keep the data and the metrics.
- Stage 2Scale
Every queue, your console
All queues, all regions, white-labelled and metered, with the Partner Console in your hands and a monthly review of knowledge gaps with our team.
Book a Stage 0 demo
Thirty minutes to scope it, three tickets to prove it. Reply from a Software Toolbox customer success engineer within one business day.
Questions support leaders ask
The objections, answered
Will it invent a menu path that does not exist?
Answers are grounded in your documentation and cite their sources, with a confidence score on every diagnosis. Where jonjon lacks a source it says so and offers escalation. Where product menus are thinly documented we build the paths, click by click, as part of onboarding.
What happens to a production-down ticket?
Urgency language is detected at triage and the ticket goes straight to your engineers, phone number included. jonjon never stands between a stopped line and a human.
What about our data, and our customers' data?
Never used to train jonjon or any foundation model. Personal information is scrubbed before inference. Retention and deletion are configurable per tenant. Conversations and tickets are kept separate, so your ticket system remains the system of record.
Does it replace our support engineers?
No. It takes the first line so your engineers spend their time on the 40% that needs them, and on customers. In October 2026 the Software Toolbox team walked in to an empty queue for the first time in the company's history and spent the morning on customer calls.
Our ticket notes are thin. Will it still work?
Yes. jonjon starts from your documentation, not from your tickets. Thin ticket notes are the norm in OT, which is exactly why the Stage 0 demo runs on real ones before anyone signs anything.
We compete with Software Toolbox products in some markets.
Your tenant, your documentation, your customers. jonjon runs as a horizontal service and does not recommend Software Toolbox products inside your deployment. Your customer data is never visible to our sales team.
How fast can we be live?
Stage 0 in about 10 days from receiving your tickets. Pilot timing is confirmed with the signed order and depends on the ticket system integration.
Which languages and which products?
Any product with documentation. jonjon answers in the customer's language, and its connectivity knowledge (OPC UA, Modbus, Allen-Bradley, Siemens, BACnet, DNP3, MQTT and Sparkplug B) comes along for free, since most of your customers' problems sit between your product and the device next to it.