A utility innovation lab that answers its own questions in a day

Client

Date

$12B+ US electric & gas utility

08/2026

The challenge

A $12B+ US electric and gas utility runs an innovation lab that has to find, evaluate and de-risk new technology for the operating business. Every question meant weeks of an analyst's time or an external scoping call, so the lab could only carry a few at once and the rest of the business queued behind them.

The solution

FifthRow put its agent research platform directly in the lab's hands: a research question in, a sourced and structured answer out. The lab's own analysts set up and modified the systems they ran, rather than briefing an external team each time.

1 day

from a new brief to a sourced assessment

147

research systems the lab's own analysts set up

Turnaround from a new brief to a first sourced assessmentOne day

Research systems the lab's own analysts set up147

Share of all research produced on systems the lab built itself40%

01 — POLICY EXPOSURE ASSESSMENT

The lab's sharpest test was a regulatory question the platform had never been pointed at, submitted deliberately off-pattern to see how far it would stretch beyond the work it was bought for.

Core Activities

Broke a new regulation down into timeline, covered sectors, principal considerations and the roles affected
Mapped which parts of the utility's customer base were exposed, and what alternatives would meet their needs
Established what comparable operators were already doing for their own customers on the same policy
Tested how much of the exposure picture could be built from public evidence alone, in parallel with the utility's own internal records
Returned a sourced report and a set of systems built against the specific questions, so follow-ups did not need a new commission

The Results

One brief produced four defined business outcomes: material to educate internal teams, a read on whether an existing product would help exposed customers, a route to giving customers visibility of their own position, and a view of what to build next. It came back inside a working day, on a question nobody had run before.

1 day

from brief submitted to first sourced assessment

02 — PROCUREMENT-GRADE TECHNOLOGY DILIGENCE

The lab's largest body of work was engineering and commercial diligence on equipment it was considering buying. Not foresight, but specification-level assessment answered from public evidence.

Core Activities

Compared competing equipment options on output, footprint, fuel flexibility and integration
Ran compliance and performance evaluations against published specifications
Modelled capacity, fuel cost and installation cost for candidate units
Assessed operating constraints, siting requirements and permitting exposure across the utility's operating states
Identified which comparable operators and which industries were already deploying each option

The Results

The lab could hold a technical conversation with a vendor on the vendor's own terms, with independent evidence already in hand. Questions that would have opened a scoping call were answered internally, in a day, with sources attached.

113

diligence questions answered in-house rather than by a consultant

03 — VENDOR LANDSCAPE MAPPING

For categories where the lab wanted to know whether new technology could change the economics of a recurring operational cost, it mapped the supplier market rather than sampling it.

Core Activities

Mapped the full vendor landscape for a category rather than the names already known
Segmented each sweep by company size, from large incumbents down to startups, so no tier was missed
Assessed the underlying technology and its cost, and built a cost-savings matrix against current practice
Benchmarked which technologies comparable operators had already deployed
Ran product-level analysis on individual offerings inside the category

The Results

The lab moved from a generic "who does this" question to a segmented vendor map with peer deployment evidence attached, plus a cost model for where the technology would and would not pay back.

market questions answered without commissioning a vendor study

04 — SCOUTING & COMPETITIVE BENCHMARKING

Around the deep diligence work, the lab used the same system for the broader questions the operating business kept asking, and answered them on the business's behalf.

Core Activities

Ran assessments against research the utility's own teams had already completed, and reviewed the two together
Scoped solutions for specific customer segments the business wanted to serve better
Assessed infrastructure options the utility was evaluating
Compiled effective policies and initiatives from comparable operators
Built product and service summaries on comparable operators, plus competitive analysis of the utility's own programmes

The Results

The lab could answer questions on behalf of other parts of the utility rather than only for itself, which is what turned it from a research consumer into an internal service.

questions answered for teams outside the lab

05 — LAB ENABLEMENT & INTERNAL ROLLOUT

The lab was not a consumer of delivered reports. Its analysts set up their own research systems, and access spread outward into the operating business.

Core Activities

Ran product training so the systems answered the questions the lab actually had
Left system authoring in the lab's hands: its own analysts created and iterated their own research systems
Had two incoming analysts creating and running their own research inside their first weeks
Extended access beyond the lab into adjacent teams, so the same system answered their questions too
Switched on scheduled report runs so repeat requests came back faster

The Results

Forty percent of all recorded research came from systems the lab set up itself rather than from anything FifthRow delivered. Access reached beyond the lab into adjacent teams, and six of its analysts became regular users running their own research.

147

research systems the lab's own analysts set up

What it proved

The lab ran five kinds of work off one system in parallel: a policy exposure assessment, procurement-grade equipment diligence, vendor landscape mapping, scouting and benchmarking for the wider utility, and its own enablement. Across the recorded platform use that came to 368 research runs. Forty percent of them came from systems the lab's own analysts set up.

What the lab was left holding was a method and a body of evidence: a segmented vendor map and cost model for a major operational category, specification-level diligence that let it meet vendors as an informed buyer, and a policy assessment turned around inside a day on a question nobody had run before. It also left with fluency it did not have before. Six of its analysts became able to create and run their own research systems, and one of them was still running research sixteen months later.

Results

MetricBenchmarkResultImprovement

Coverage on a vendor landscape question

Benchmark: The team's own prior manual pass on the same categoryResult: Full category coverage segmented by size band, with peer deployment evidence attached

From a partial list to a mapped market

Turnaround from a new brief to a first sourced assessment

Benchmark: The lab's own manual pass on the same questionResult: One day, on a question the system was never scoped for

Measured head to head

Research systems the team could author itself

Benchmark: noneResult: 147

New capability

Share of research produced on the team's own systems

Benchmark: noneResult: 40%

New capability

SPECIFIED COVERAGE

A vendor sweep is specified by size band, so the landscape that comes back runs from incumbents down to startups instead of the obvious names.

PROCUREMENT-GRADE DILIGENCE

Equipment specification, cost and permitting questions come back sourced from public evidence, so vendors are met with independent numbers already in hand.

PEER DEPLOYMENT EVIDENCE

Every assessment carries evidence of what comparable operators have already deployed, which is the viability proxy vendor material cannot supply.

SELF-AUTHORED SYSTEMS

The lab's own analysts set up and modify the research systems they run, so a new question does not need an external brief.

OFF-PATTERN RANGE

Questions the system was never scoped for still come back as sourced assessments, so the lab is not confined to the workstreams it bought.

Recent case studies

See more proof

Cutting a four-month research cycle to six weeks across five validation workstreams

Client

$25B+ US energy & home-services company

Date

08/2026

6 days

from intake brief to a ranked opportunity report

From a weekly newsletter to a self-running innovation radar

Client

$70B+ global energy utility

Date

08/2026

284,680

analyst-hours displaced, the equivalent of 71 full-time analysts

Matter More

Let's Talk