Introducing

Sprocket AI

Sprocket is the AI assistant built into Clockworks. Ask a question in plain language and get answers grounded in your portfolio’s equipment, faults and performance data.

Ask your buildings anything. Get answers you can act on.

Example prompt: Help me prepare for the seasonal changeover. What should I prioritize?

Sprocket’s answer: Fix simultaneous heating and cooling first. It’s $75,461 in avoidable cost across 31 units, and it gets worse once heating starts.

  • Leaking coil valves fighting each other: AHU-3 and 4 more, max priority
  • Boiler efficiency low and boilers cycling: Heating Plant, Boiler 2
  • Malfunctioning freeze protection and economizers: 14 AHUs flagged

Next step: Stroke-test coil valves on the 5 top-priority AHUs, then tune the boilers before the first cold snap.

Example prompt: Summarize all task activities over the last 7 days.

Sprocket’s answer: All 7 units with tasks completed since June are faulting again. The recurrences cost $24,272 in the last 30 days.

  • AHU-3, heating and cooling at once: 25 of 30 days, $11,863
  • AHU-8, fresh air damper stuck open: 29 of 30 days
  • AHU-4, task closed last month: 27 of 30 days, $3,111
  • 2 operating rooms: humidity low, 30 of 30 days

Next step: Reopen the AHU-3 and AHU-8 tasks and verify the valve fix held.

Example prompt: Which critical spaces are at risk?

Sprocket’s answer: 7 of 8 critical spaces are faulting, 6 at top priority. Most have faulted every day this week.

  • Hazardous room: exhaust airflow low, 21 hrs
  • Operating room 1: air changes low, 7 faults
  • 5 OR and pharmacy spaces: humidity low, 20 to 24 hrs

Next step: Start with the hazardous room’s exhaust airflow. Then fix humidity across the five spaces.

Demo Sprocket

Complex insights in minutes.

It can take on complex research and reporting that used to take consultants months because it provides advanced asset intelligence across facilities functions.

In years past, a report like the two we just produced would have been a million dollar research project, funded by the state and requiring untold hours of researcher time.”

Chief Design Engineer, National Healthcare Organization

Generate a Supply Air Temperature Reset Report

Ask Sprocket how well supply air temperature reset is working across your air handlers. It reviewed two years of data on 23 units and built a 51-page report that shows which resets are working and which ones slipped since the last study.

51
Pages
23
Air handlers
2
Years reviewed
2025 studyOct 2025 – Sep 2026

Hover or tap an air handler to compare years.

  1. Medical office AHU 1
  2. Medical office AHU 2
  3. Medical office AHU 3
  4. Medical office AHU 4
  5. Hospital AHU 1
  6. Medical office AHU 5
  7. Medical office AHU 6
  8. Medical office AHU 7
  9. Medical office AHU 8
  10. Medical office AHU 9
  11. Medical office AHU 10
  12. Medical office AHU 11
  13. Medical office AHU 12
  14. Hospital AHU 2
  15. Hospital AHU 3
  16. Hospital AHU 4
  17. Medical office AHU 13
  18. Hospital AHU 5
  19. Hospital AHU 6
  20. Medical office AHU 14
  21. Medical office AHU 15
  22. Hospital AHU 7
  23. Hospital AHU 8
Measured reset, °F

An ASHRAE 180 report in minutes, not months.

Ask SprocketBased on ASHRAE 180 Guidelines on preventive maintenance, what tasks can be shifted to condition-based maintenance?

The result

69% of PM tasks can shift from time-based to condition-based

  • Automated

    16.5%

    Clockworks diagnostics perform the check continuously.

  • Condition-based

    25.3%

    Done only when the data says it’s needed.

  • Deferred

    27.2%

    Skipped while equipment stays healthy.

  • Manual

    31.0%

    Your team’s time goes only where it’s needed.

Results from one example Clockworks portfolio.

And that's just the start

CFO savings summaries, one pageDirector-level energy briefsRecurring faults across every siteCondition-based maintenance candidatesOffline and stale point checksYear-two reviews of past projectsWeekly service plans for every clientChiller plant readiness, plant by plant

Not just another chatbot.

Sprocket doesn’t read raw sensor data and guess. It pulls insights from Clockworks’ industry-leading fault detection and diagnostics already running on your buildings—including finding each fault, its root cause, and cost. Every answer is grounded in engineering, whether you’re sending a technician on site or briefing leadership.

Built for every role

Know what to fix before you go on site.

Ask SprocketWhat’s wrong with this unit, and what should I go check first?

Sprocket’s answer · Air handler 3

Diagnosis: Simultaneous heating and cooling because of the leaking pre-heat coil valve.

Preheat temp rise+12°F

Air heats from 65°F to 77°F while the hot water valve is commanded fully closed.

Cooling valve28% open

The cooling coil works to remove that unwanted heat to hold a 55°F supply setpoint.

ResultPaying twice

Heating and cooling energy spent fighting each other, invisible on a single BAS screen.

A ranked work list tied to energy, maintenance, cost, or comfort.

Ask SprocketWhat should my team focus on today?

Sprocket’s answer · today’s work list

  1. 1
    Raise humidity in operating rooms 1, 2, and 4

    Room humidity below minimum.

    6 to 7 of 7 days · Comfort priority 10
  2. 2
    Fix AHU-3 supply fan stuck at high speed

    Fan speed constant at high speed. No static pressure setback schedule.

    $345 avoidable in 7 days · Energy priority 10
  3. 3
    Check the pharmacy air handler for high VOCs

    Room air VOC above maximum.

    6 of 7 days · Maintenance priority 9

Targeted savings and carbon reduction.

Ask SprocketShow me which building has wasted the most energy.

Sprocket’s answer · last 30 days

Daily avoidable energy cost - top 5 buildings (19 Sep - 18 Oct 2026)

Board-ready results in minutes.

Ask SprocketSummarize savings for our CFO.

Sprocket’s answer · savings summary

Proactive service across every customer.

Ask SprocketAcross all service clients, what work should be proposed this week?

Sprocket’s answer · this week’s proposals

  1. 1
    Client A

    Fix simultaneous heating and cooling on 5 air handlers and leaking cooling valves on 8. Propose VFDs on 7 fans and 2 pumps.

  2. 2
    Client B

    Compressor running with the system off. Add night setback across 99 zones in 5 buildings.

  3. 3
    Client C

    Propose VFDs on 2 pumps. Check chilled water supply temp below minimum.

  4. 4
    Client D

    6 fans stuck at high speed. 2 fans running while unoccupied.

The Foundation for AI-Driven Facilities

  1. Maintenance management

    “What can move to condition-based?”

  2. Asset management

    “Which assets should be replaced in next year’s capital plan?”

  3. Building automation

    “Which points are offline?”

  4. Energy & sustainability

    “Where are we wasting energy?”

  5. Risk management

    “Which critical equipment is at risk right now?”

  6. Compliance management

    “Which pharmacy spaces were out of compliance over the last 30 days?”

  7. Business management

    “Summarize savings for our CFO.”
Sprocket AIIntelligence across every facility function

Built on Trusted Building Data

Every layer is validated, from connectivity and tags to a structured information model and expert diagnostics. The result is engineering-grade insight on transparent, validated data.

  1. 4,200 buildings monitored

  2. 5.8M+ data points classified daily

  3. 50+ expert systems, running continuously

  4. Ask your questions, get actionable responses.

SPROCKET AI

3 real problems to fix. One bad sensor set aside.

AHU-3 is heating and cooling at the same time. The pre-heat valve is leaking.

TOP 3 BY PRIORITY

AHU-3Heating and cooling at once
OR-1Air changes low
AHU-4Fan on too long

A portfolio of buildings. One building is picked out and opens up floor by floor. Its messy point names are matched to plain labels by the information model, following an enthalpy-based economizer sequence: outdoor and return air enthalpy, high limit temp, low limit temp, and the outdoor air damper. Expert systems scan every floor, set aside a broken sensor, and rank the real problems: AHU-3 heating and cooling at once, then low air changes in OR-1, then a fan on too long at AHU-4. Sprocket gives the answer with links to each one.

$240M
Saved
800k
Tonnes carbon saved
70k
Completed tasks
650,000+
Pieces of equipment modeled

“It all comes down to having the data foundation. Not just data points, but expert systems diagnostic output that AI can sit on top of.”

See Clockworks AI at work.

Meet Sprocket AI Chat blog featured image

Blog

Meet Sprocket AI Chat: Ask Your Buildings Anything

Read
Condition-Based Everything, Everywhere, All At Once webinar title slide, Nick Gayeski62 MIN

Webinar

Condition-Based Everything, Everywhere, All At Once

Nick Gayeski

Watch
City skyline at night

Whitepaper

The Information Model That Powers Fault Detection and Diagnostics (FDD)

Read

Client Summit 2026

AI in Action: Applications of Clockworks AI Chat & MCP Server

Dennis Krieger and Tim Cramer

Watch

Webinar recording

Ask Your Buildings Anything: Meet Sprocket, AI Chat for Clockworks

Dennis Krieger

Watch

Sprocket AI FAQ

What is Sprocket AI?

Sprocket AI is the AI assistant inside the Clockworks Analytics fault detection and diagnostics (FDD) platform. Ask questions about your buildings in plain language, and Sprocket answers using your own Clockworks data: diagnostics, tasks, equipment and trend data. Every answer links to the diagnostics, tasks or data behind it, so you can go straight from the answer to action.

Why are Sprocket’s answers more accurate than other smart building chatbots?

Most building chatbots read raw building automation system (BAS) data and guess, because point names are inconsistent from site to site. Sprocket answers from the foundation Clockworks has already built: an information model that structures equipment, points and their relationships, and 50+ expert systems that find each fault, its root cause and its cost. Every figure links to its source, so you can check the work.

Is Sprocket AI secure?

Yes. Your building data stays in the Clockworks Microsoft Azure environment, and Sprocket reads it where it already lives. The AI model runs in Azure AI Foundry, hosted on Azure in the US, under Microsoft’s Data Protection Addendum. Sprocket sees only what you’re already permitted to see in the Clockworks portal. Data is encrypted in transit and at rest, and every interaction is monitored.

Does Sprocket AI use my data to train AI models?

No. The large language model (LLM) provider’s commercial terms prohibit it from training models on customer content. Sprocket uses pretrained models through Azure AI Foundry, and your Clockworks data is not used to train, fine-tune or modify them.

What can I ask Sprocket AI?

Ask about anything in your Clockworks data, in plain language. For example: “What should I focus on today?”, “Help me prepare for the seasonal changeover. What should I prioritize?” or “Which critical spaces are at risk?” Sprocket can also build interactive charts and complex reports, such as a supply air temperature reset study or an ASHRAE 180 maintenance analysis.

Can Sprocket AI browse the web or access other systems?

No. Guardrails keep Sprocket on building topics. It has no web browsing, no file uploads and no access to credentials or secrets, which closes common prompt-injection routes. It reads Clockworks data only through the same permission-checked access the Clockworks portal uses.

Who can use Sprocket AI?

Sprocket is available to all Clockworks users starting October 8, 2026. It works inside the Clockworks portal with your normal login and permissions. Clients can contact their Clockworks account manager for access. New to Clockworks? Request a demo to see Sprocket and the platform behind it.

Demo Sprocket