CIM PEAK vs SkySpark: build your own analytics, or run the full loop?

November 20, 2026
Commercial building mechanical room with condenser water pipes, pumps and valves

SkySpark comes up in a lot of conversations with engineering firms and controls contractors. It is a capable analytics engine, and many firms have used it for years. The question we hear is not whether SkySpark works. It is whether building and running your own analytics is still the best use of your engineers as the portfolio grows.

This comparison is for engineering, commissioning and controls firms weighing SkySpark against CIM's PEAK Platform. It covers how each one gets from BMS data to a verified fix, what each really costs over five years, how each scales and how to move buildings across if PEAK is the better fit.

Key takeaways

  • Two different models. SkySpark is a programmable analytics engine your team builds on. CIM PEAK is a full-loop operations system that arrives with rules, workflow and fix verification built.
  • The build is the difference. On SkySpark your team tags points, writes rules and builds the workflow before the first verified fix. On PEAK, buildings are typically live in about 30 days.
  • Total cost, not licence cost. Total cost = licence + hosting + your team's hours. A self-built stack needs developer, integration and IT time that PEAK's platform licence fee already covers.
  • Scale is where it shows. Every new SkySpark site adds tagging and rule work. PEAK adds buildings without adding people.
  • Verification is built in. PEAK closes an action only when the data confirms the fix held.

What is the difference between SkySpark and CIM PEAK?

SkySpark is a programmable analytics engine: your team tags the data, writes the fault rules in its Axon language and builds the workflow around them. CIM PEAK is a full-loop operations system: it arrives with 5,000+ engineering rules, ranking, actions, follow-up and fix verification already built, so engineers start from findings.

CIM PEAK is an independent fault detection and diagnostics (FDD) platform that reads the BMS you already have, finds faults, decides what matters, assigns the work, chases it and verifies the fix in the data. SkySpark, from SkyFoundry, is a general-purpose analytics platform for building and equipment data that integrators and in-house teams program to suit their own needs.

Both can find faults. The difference is how much your team has to build first, and who carries that work every year after.

From BMS data to the first verified fix
What your team has to build before the first fix is proven
SkySpark5 build steps before the first verified fix
Connect the BMS
Tag every point
Write the fault rules
Design the ranking
Build the workflow
Build fix checks
First verified fix
CIM logoCIM PEAKRules, workflow and fix checks arrive built
Connect the BMS, read-only
AI maps the points
First verified fix
5,000+ rules, ranking, actions, follow-up and fix verification already in place. Buildings typically live in about 30 days.
Setup stepYour team builds itOutcome
Neither model is wrong. SkySpark gives a team with developers a powerful engine to build on. PEAK starts your engineers from findings, and with no developer, integration or IT roles to staff, the five-year cost is usually lower too. The AI recommends. Your engineers decide.

What is SkySpark, and who is it built for?

SkySpark is licensed analytics software from SkyFoundry. It tags data with Project Haystack, runs rules written in its Axon language and is installed by you or your host, on-premise, in your cloud or at the edge. It suits teams with developers who want to write and own their analytics.

SkySpark in brief
A programmable analytics engine, built for teams that write their own rules
Data model

Tags data with Project Haystack. Your team applies the tags.

Analytics

An engine and function library. FDD rules are written in its Axon language.

Hosting

Licensed software you install and run, on-premise, in your cloud or at the edge.

SkyFoundry describes the product on its SkySpark product page. Many SkySpark deployments in commercial buildings are delivered by value-added resellers and integrators who host the software and write the rules for their clients. That model works best where a firm has developers on staff and a rule library it has already built.

What does it take to get FDD running on each?

On SkySpark, your team connects the BMS, tags every point, writes and tests the fault rules, designs how findings are ranked and builds the workflow and fix checks. On PEAK, CIM connects read-only, AI maps the points onto an existing model and 5,000+ rules deploy automatically, so buildings are typically live in about 30 days.

Tagging is the first gate. BMS point names vary from site to site, and every point has to be mapped to a model before rules can run. On SkySpark that model is Project Haystack and the tagging is your team's work; our guide to BMS point naming, Haystack and Brick explains why it takes time. PEAK uses AI point classification to map names onto a model refined across 700+ buildings, then engineers check the result.

Rules are the second gate. A rule library has to be written, tested against real plant and tuned to stop false alarms, and that work never really finishes. PEAK arrives with that library in place, which is why the time to a first finding is weeks, not a rule-writing project. Our breakdown of how long it takes to get a building live on analytics walks through each stage.

What does each really cost over five years?

Compare total cost, not licence cost. Total cost = licence + hosting + your team's hours. With SkySpark you pay for the licence, hosting and hardware, plus the developer, integration and IT time to build and run it. With PEAK you pay one platform licence fee, with rules, hosting and verification included.

Total cost = licence + hosting + your team's hours
What you pay for, every year
With SkySpark
Software licence
Hosting and hardware
A developer to write the rules
An engineer to integrate the data
IT to run and secure the server
5lines on your budget, three of them people
With CIM PEAKCIM logo
Platform licence fee
Already included
5,000+ fault rules, written and maintained
Hosting, security and updates
Follow-up and fix verification
1line on your budget, no new hires

The licence is usually the smallest line in a self-built stack. The bigger lines are people: someone who can write Axon and keep the rule library current, an engineer to integrate and tag the data, and IT time to host, patch and secure the server. Those costs recur every year and grow as the portfolio grows.

PEAK's pricing model is simple to compare: engineering and service partners are licensed per data point, and direct customers per square metre of floor area. Hosting, the rule library, follow-up and fix verification sit inside that fee. Whichever way you lean, ask each option for the same number: five years of cost per building, with your own people's hours included.

Which one scales better across many buildings?

PEAK scales better across a large portfolio because the work that grows with each building, tagging points and writing rules, is already done or automated. On SkySpark, each new site adds tagging, rule tuning and developer time, so capacity grows with headcount. On PEAK, capacity grows with software, so each engineer covers more buildings.

This is the point where the two models separate most. A SkySpark practice that runs well on ten buildings has to add specialist people to run a hundred. Aero Performance Group, a Chicago commissioning firm, grew its active project portfolio by 50% without adding resources after moving its monitoring-based commissioning work onto PEAK; the Aero case study sets out how.

What happens after a fault is found?

Finding a fault is only the start. PEAK takes each fault through the full loop: an engineer confirms it matters, it is assigned to one owner, the Compliance Agent follows up until it moves and the action closes only when the data confirms the fix held. On SkySpark, each of those steps is something your team builds and runs.

The full loop, one fault16 days from detection to verified close
1
DetectA rule finds itDay 0
2
DecideAn engineer confirms itDay 1
3
AssignOne owner, one due dateDay 1
4
ChaseFollowed up until it movesDay 9
✓
VerifyClosed when the data agreesDay 16
CIM logoActions+ Create Action
AHU-03 - Simultaneous Heating and Cooling ›
Office building, level 4 plant room · Closed 16 days after detection
Closed (Recovered)P2 UrgentEnergy
1 DetectDay 0
Rule fired: heating and cooling valves both open on AHU-03 for long periods on three consecutive days.
2 DecideDay 1
Partner engineer: confirmed in the trend data, not a sensor problem. Likely a heating valve that is not closing.
3 AssignDay 1
Assigned to the site technician with a 7-day due date.
4 ChaseDay 9
Compliance Agent AI follow-up
Open for 8 days with no fix reported, and AHU-03 is still heating and cooling at once. Can you confirm the next step, or reassign?
5 VerifyDay 16
Compliance Agent AI verified close
Closing: recovered and stable. The heating valve now closes when cooling runs, and the fault has stayed clear for a full operating week.

Illustrative example, based on real Compliance Agent comments. Sites and names changed.

1 in 6closed too early

In CIM's sample, about one in six closed fault tickets were closed before the data confirmed the fix. That is why verification is built into PEAK, not left to a custom rule.

Verification is the step that most often goes missing. A work order marked complete is not the same as a fault that has cleared and stayed cleared, and in CIM's sample about one in six closed fault tickets were closed before the data confirmed the fix. Our guide to AI agents that triage faults and verify fixes explains how the agents work, and the rule that governs them: the AI recommends, your engineers decide.

How do you move a portfolio from SkySpark to PEAK?

Move building by building. Start with new buildings and contract renewals, connect PEAK read-only to the existing BMS, let AI map the points and check the first findings against what your engineers already know. Once a building is live on PEAK, rebuild only the custom dashboards that still matter and retire the SkySpark project for that site.

Nothing has to be retagged by hand before rules run: AI classification maps the point names and engineers check the result. PEAK needs no new hardware on site and no changes to the controls. CIM's engineers sit behind your team during onboarding and training, so your engineers keep the client relationship throughout.

How does each get through a client's IT review?

With SkySpark, your firm or your host runs the server, so the IT review covers your hosting, patching, access control and backups. With PEAK, CIM hosts the platform and connects to the BMS read-only, so the review focuses on that connection and the audit trail, and no client server has to be maintained.

Read-only access matters to client IT teams because it shows the platform cannot change setpoints or schedules. Our guide to getting an AI building analytics platform through IT review covers the questions security teams ask and how to answer them.

SkySpark or PEAK: which fits your firm?

Stay on SkySpark if you have developers who want to write and own the analytics, a mature rule set and clients who mainly want dashboards. Move to PEAK if you need to scale to many buildings without adding people, want engineers fixing buildings rather than writing code and need buildings live in about 30 days.

Option 1
Stay on SkySpark
Keep building and running your own analytics on the engine you know.
Choose this when

You have developers who enjoy writing Axon

Your rule set is mature and rarely changes

Clients mainly want dashboards and energy reporting

Watch out for

Hard to scale across many buildings. Every new site means more tagging, rule writing and developer time.

Option 2CIM logo
Move to PEAK
Start new buildings and renewals on a platform that arrives with all the layers built.
Choose this when

You need to scale to many buildings without adding people

Your engineers should fix buildings, not write code

Buildings need to be live in about 30 days

Watch out for

Plan the move building by building. Bespoke SkySpark dashboards need rebuilding or retiring.

Where does CIM's PEAK Platform fit, and when is it not the right choice?

PEAK fits engineering firms, commissioning providers and controls contractors who want to deliver FDD, MBCx and data-driven maintenance at scale under their own name. It is not the right choice for a team whose main goal is writing its own analytics code, or for data far outside building HVAC and energy.

CIM is fully independent: it is not owned by any equipment, controls or BMS manufacturer, so the monitoring stays separate from the people being monitored. PEAK runs on 700+ commercial buildings, reads the BMS you already have and works through partners rather than competing with them. If you are weighing platforms more broadly, our comparison of FDD and building analytics platforms covers the wider field, and The Building Performance Engineer's FDD Playbook sets out how partner engineers run FDD day to day.

For engineering firms and controls contractors

See how far PEAK takes your team before anyone writes a rule

Bring a building you run on SkySpark today. We will connect read-only, map the points and show you the findings, the actions and the verified fixes your engineers would work from.

Frequently asked questions

What are the alternatives to SkySpark for engineering firms that don't want to build their own analytics rules?

The main alternatives are turnkey FDD platforms that arrive with a tested rule library, such as CIM PEAK. PEAK deploys 5,000+ engineering rules automatically, maps points with AI and includes actions, follow-up and fix verification, so engineering firms start from findings rather than writing rules. Buildings are typically live in about 30 days.

Is SkySpark an FDD platform?

SkySpark is a programmable analytics platform that can be used for fault detection and diagnostics. Its rules are written in the Axon language, so FDD on SkySpark depends on the rule library your team or integrator builds and maintains. Turnkey FDD platforms such as CIM PEAK arrive with the rules already written and tested.

How is SkySpark licensed?

SkySpark is licensed software from SkyFoundry, sized by point count and usually bought through a reseller or integrator. It is installed and run by you or your host, on-premise, in your cloud or at the edge, so hosting and maintenance sit alongside the licence. CIM PEAK is licensed per data point for partners, with hosting included.

Is CIM PEAK cheaper than SkySpark?

Over five years, PEAK usually costs less once your team's hours are counted. A self-built SkySpark stack needs developer, integration and IT time on top of the licence and hosting. PEAK's platform licence fee includes the rule library, hosting, follow-up and fix verification. Compare five years of cost per building, people included.

Does PEAK need the Haystack tags from a SkySpark deployment?

No. PEAK uses AI point classification to map a building's point names onto its own model, then engineers check the result. That means a building can move across from SkySpark without retagging by hand, and without depending on how well the original Haystack tagging was done.

How long does it take to move a building from SkySpark to PEAK?

Buildings are typically live on PEAK in about 30 days. PEAK connects read-only to the existing BMS, maps the points with AI and deploys 5,000+ rules automatically, with no new hardware on site. Most firms move building by building, starting with new sites and contract renewals.

Does PEAK need developers or coding skills?

No. PEAK's rules are already written and deploy automatically, and site-specific rules are built without code. Engineers review findings, decide the actions and track fixes. CIM's engineers support onboarding, training and escalations, so partner teams do not need a software developer to run FDD.

Guide cover
Free guide
The Building Performance Engineer's FDD Playbook

A practical playbook for the engineers who deliver building performance. Learn how to run retro-commissioning, monitoring-based commissioning and data-driven maintenance on an FDD platform: compress the RCx cycle from 12-18 months to 3-5, close the loop from alert to verified fix, and turn fixed-term projects into recurring revenue. Written for RCx and MBCx firms, engineering consultancies, and BMS and mechanical contractors.

Get the free guide
Get the free playbook
https://connect.cim.io/the-building-performance-engineers-fdd-playbook-gated
Chiller down, BAS dark: how PEAK kept a school's classrooms cool
Customer story
Chiller down, BAS dark: how PEAK kept a school's classrooms cool

A new chiller failed at a US high school on the same morning its BAS front-end went dark. PEAK flagged it first thing, and the fix landed before anyone in class noticed.

Flagged first thing Monday
No classroom disruption
Read the story →
school-chiller-failure-caught-while-bas-front-end-was-down
Paul Walsh
November 20, 2026
Share
Before you go

The Building Performance Engineer's FDD Playbook

A practical playbook for the engineers who deliver building performance. Learn how to run retro-commissioning, monitoring-based commissioning and data-driven maintenance on an FDD platform: compress the RCx cycle from 12-18 months to 3-5, close the loop from alert to verified fix, and turn fixed-term projects into recurring revenue. Written for RCx and MBCx firms, engineering consultancies, and BMS and mechanical contractors.

Guide cover

Powering property teams in these world leading companies.