How Inspect-It 1st reduced inspection report processing time by 75%

‍

The Challenge

For commercial and multifamily property transactions, investors use statements of probable cost (SOPCs) generated from inspection reports to forecast immediate and 5 year repair and maintenance costs of property defects.

The process of generating SOPCs is time consuming, requiring manual extraction of these defects from PDF inspection report summaries into Excel spreadsheets that contain a forecasting algorithm derived from a database of contractor pricing across various localities and project complexities.

On the residential side of the inspection industry, tools to automate the SOPC creation process are gaining traction. The tools ingest inspection report summaries, and use AI capabilities to generate cost estimates based on the extracted defects. II1st-GPM’s CEO, Martin Lenich, immediately took interest. “We’ve been preparing SOPC’s since 2016 and compared residential SOPCs that we prepared against their outputs in 2023, and they’re quite good,” Lenich shared, but for the more complex commercial and multifamily reports that made up most of II1st-GPM’s business, “we couldn’t get their output to match the complexity of our summaries and resultant spreadsheets.”

We tried different methods of automating that PDF to Excel spreadsheet, but because of the complexity of our summary report, they all kind of crash and burn - we would just have errors. We were spending more time checking errors than just raw typing them in Excel.

Lacking a suitable pre-existing solution, Lenich decided to develop a custom cost estimate automation product specifically tailored for II1st-GPM’s more complex commercial and multifamily reports. Not only would this reduce the manual burden placed on II1st-GPM’s inspectors and estimators, it would also allow the company to grow its business without expanding headcount.

Lenich explored methods to quickly and accurately extract data from the PDF report summaries that were central to an automated SOPC automation product. II1stGPM’s ideal solution would need to satisfy these requirements:

Quick implementation

With an ambitious year-end implementation timeline and a November 2023 start date, II1st-GPM needed an easily integrable solution that could quickly deliver results.

Ease of use

II1st-GPM prioritized a fully capable solution with a functional web-based user interface.

Flexible pricing for ramp period

While II1st-GPM was willing to pay for a reliable solution, they knew they would have a slow ramp given current real estate market conditions, and a flexible pricing structure cognizant of that was important.

High extraction accuracy

Extraction accuracy was crucial for the effectiveness of this step in their SOPC automation process. The team needed to be able to customize and configure extractions to guarantee high accuracy rates.

Lenich’s chance conversation with Dan Gumpright, a highly respected consultant familiar with API document extraction tools, turned into a referral to Sensible.

“Over a drink, I asked him about this,” Lenich recounted, “and he said ‘we’ve actually gone through the same step for some of our clients… we’ve evaluated 10 different extraction companies, and for what you’re describing let me do an email introduction to Sensible.’”

Over a drink I asked him about this and he said ‘we've actually gone through the same step for some of our clients… we've evaluated 10 different extraction companies, and for what you're describing, let me do an email introduction to Sensible.’

II1st-GPM sent four sample inspection report summaries to Sensible for testing, ranging from a four-plex multifamily unit to a 250,000 square feet multi-story office complex. A week later, Sensible presented II1st-GPM with a functional demo that delivered promising extraction results across all of the report summaries. “This was before we even signed up,” said Lenich, “and I mean my jaw dropped at how easy and quick it was to do these extractions with Sensible.”

During the initial demo, II1st-GPM was impressed by Sensible’s flexibility and the ease of making real-time modifications. “We were able to utilize and test [without] going through an extensive parallel effort,” Lenich explained, “I thought it might take a couple months to get something back, and then we would go through testing, but literally from day one when I saw the output I was like ‘my gosh, this works already!’”

I thought it might take a couple months to get something back, and then we would go through  testing, and literally from day one when I saw the output I was like, ‘my gosh this works already!’

The Solution

II1st-GPM uses Sensible to extract inspection report summaries  in their SOPC automation product. The product ingests a PDF inspection report summary, extracts any listed defects, reformats symbols or numbers, and maps the data to II1st-GPM’s comprehensive cost estimate Excel algorithm. II1st-GPM worked closely with the Sensible team from the outset to overcome hurdles that the residential SOPC automation products couldn’t, including decoding symbols and numbers included in the summaries, and complex mapping into subtables in Excel.

Once II1st-GPM’s product went live, Lenich’s confidence in its capabilities only grew as it easily handled a wide variety of inspection reports across property type and sizes. When faced with a handful of edge cases that caused issues with report symbol recognition and jumping between defect sections, “we spotted them right away,” said Lenich. Sensible modified its output as quickly as the issues were discovered, and following the fixes, “we’ve been running about four weeks now without a problem,” noted Lenich.

In addition to Sensible, Lenich credits II1st-GPM’s preexisting – and highly specified – output schema with the success they’ve seen. “We had a clear input, and we had a desired output,” Lenich explained, “I think having that is critical to the software developers.”

The Results

Within a month we had a solution that was 95% error free, which was a tremendous time saver... and now we’re basically error free on our extractions.

II1st-GPM’s SOPC automation product launched on schedule in January 2024. From day one, II1st-GPM saw report summary extraction and SOPC production time decrease from an average of 120 minutes per report to under 30 minutes – with an initial 95% extraction accuracy rate. After working with the Sensible team to make some tweaks, this increased to 99% accuracy within a month across all report summaries,  and has now reached 100%.

II1st-GPM quickly on-boarded their team – “we’ve trained everybody,” Lenich said, “they know how to plug the inspection report summaries into the spreadsheet, so it’s pretty error-free from that standpoint.” Two months after their initial launch, II1st-GPM has reduced SOPC generation time by 75%, allowing the team to dedicate their time to quality control and error checking. “Our estimating side can crank out a lot more estimates,” said Lenich, “and right now we’re seeing the ramp up in our business.”

“We’ve knocked an hour and a half off per report, which is a big time savings for us, and our estimating side can crank out a lot more estimates, and right now we’re seeing the ramp up in our business”

For II1st-GPM, the launch had a meaningful outcome beyond the metrics. “One of the key focus points of our company is the idea that if we can attract the best talent, we can provide the best inspections,” said Lenich, “so a lot of our technology solutions center around making life easier for our team of inspectors, report generators, and estimators.”

The success of the initial implementation led II1st-GPM to collaborate with the Sensible team to identify new features to support diversified offerings, including one feature that assisted the re-launch of their pre-listing inspection service.