Four Years of Council Records, Searchable in Seconds: How a City Transformed Open Records Access

Government transparency is not just a legal obligation — it is a foundation of resident trust. Yet in most cities, the records that most directly demonstrate how local government works — council meeting minutes, resolutions, budget approvals, ordinances, development decisions — are technically public but practically unfindable. Published as PDFs in a library residents don't know exists, described with agenda item codes that mean nothing to a non-specialist, and unsearchable by the content that actually matters to residents. One city decided that 'publicly available' and 'publicly accessible' should mean the same thing. Here's what happened.

The City

The city serves approximately 195,000 residents and operates under a council-manager form of government. The City Council meets twice monthly in regular session, with additional special sessions and committee meetings held as needed. Over a four-year period from 2021 through 2024, the council had generated 96 regular meeting minutes, 284 council resolutions, 67 ordinances, 48 special session minutes, and 320 committee meeting minutes — a total of 815 public documents containing the full record of every significant decision affecting the city.

All 815 documents were published on the city's website in a 'City Council Records' section maintained by the City Clerk's office. Each document was listed by date, meeting type, and a brief title. The section had no search capability beyond the browser's built-in Ctrl+F text search within a single document listing page. A resident trying to find every council decision related to a specific development project, a particular city street, or a specific service provider would need to open each PDF individually and search within it — or file an Open Records request.

The City Clerk's office received an average of 340 Open Records requests per year. A review of the prior two years' requests found that 189 of the 680 total requests — 28% — were for documents already published in the City Council Records section. Residents were using the formal Open Records process to access information they could, in principle, access themselves — because the documents were effectively undiscoverable through the existing website structure.

The Problem: 'Publicly Available' Is Not the Same as 'Publicly Accessible'

The city's legal posture on its council records was that they were publicly available — and, technically, this was correct. Every document was published. Every document was downloadable. The obligation to disclose had been met.

But a resident who wanted to know: 'Has the council ever approved a variance for a property in my neighbourhood?' faced an effectively insurmountable task. They would need to know that variances are approved by resolution, know the resolution numbering format, find the resolution index, open relevant candidate resolutions (there was no way to narrow without opening), and read each one to check. There was no way to search across the text of all 284 resolutions for a property address or a specific applicant name.

The same problem appeared for journalists tracking budget decisions, neighbourhood associations following development applications, and residents trying to understand why a specific service had changed. All of them were doing work that should have been done by a search engine — and many were filing Open Records requests to get answers that were already published, because the Open Records process was actually faster and more reliable than the website's document library for finding specific information.

"I received a request asking for all resolutions related to a specific street over the past three years. I knew the information was in our published records. But it would have taken me two hours to pull it manually. It took three days to respond because we were behind on requests. The requester could have found it themselves in five minutes if the records were searchable."

— City Clerk

The Dual Goal: Transparency and Operational Efficiency

The City Clerk approached the IT department with a proposal framed around two goals. The first was public transparency — making the city's council records genuinely discoverable by any resident, journalist, or researcher without requiring legal or technical intermediation. The second was operational — reducing the Open Records request volume for already-public documents, freeing the Clerk's office to focus on complex requests that genuinely required staff involvement.

A supporting analysis calculated the operational cost. The Clerk's office employed three full-time staff, of whom one spent approximately 50% of their time on Open Records request processing. Reducing the 28% of requests that were for already-public documents by 55% — a conservative estimate — would recover approximately 14 staff hours per week. That time could be redirected to building the proactive disclosure programme the Clerk had long planned but never had the capacity to execute.

What Was Deployed and How

The deployment had two components. The first was Keyspider AI Search on the public-facing city website — specifically the City Council Records section — to make all 815 published documents searchable by content, not just by filename and date. The second was Keyspider Workplace Search for the Clerk's office internal records systems, to accelerate the document discovery phase for complex Open Records requests that required searching records not yet published on the website.

Component 1: Public Records AI Search

Keyspider's crawler was configured to index the City Council Records section, with deep PDF extraction enabled for all 815 documents. The indexing was complete within 48 hours. Each document was indexed with its metadata — date, document type, meeting body — alongside the full text of every page, enabling residents to search not just for document titles but for names, addresses, dollar amounts, ordinance numbers, contractor names, and any other specific content contained within the documents.

The search interface was configured to display results with document type, date, and a content snippet showing the passage within the document where the search terms appeared — allowing residents to assess relevance before opening the full PDF. For longer documents like meeting minutes, the snippet indicated the specific agenda item or page number where the relevant content appeared.

An AI answer summary layer was activated for common query types. A resident searching 'has the council approved any development on Oak Street?' would receive an AI-generated summary listing the relevant resolutions found, with dates and resolution numbers cited, before the full document results. This was the capability that most directly addressed the query type that had been generating Open Records requests.

Component 2: Staff Workplace Search

The Clerk's office maintained records in four separate systems: the published website documents, a shared network drive containing draft and working documents, the city's legacy document management system (housing records from 2010–2018), and email archives. For complex Open Records requests requiring a comprehensive search across all city records — not just published documents — staff had previously needed to search all four systems separately.

Workplace Search was configured to connect all four systems through a single search interface accessible only to authorised Clerk's office staff. Permission controls ensured that staff could not access records from other departments without appropriate authorisation. The search log functionality — recording every query, filter, and result set — provided the audit trail required to demonstrate good-faith search compliance under state Open Records law.

Results at 90 Days

Open Records Request Volume

At 90 days, monthly Open Records requests for documents already published online had fallen from an annualised rate of 189 per year to approximately 85 per year — a 55% reduction. Total Open Records request volume fell 16% — the 55% reduction in the 'already-public' category was partially offset by a modest increase in genuinely complex requests, consistent with residents now self-serving simple requests and investing the time saved in filing requests for information that genuinely required staff involvement.

Resident Self-Service Volume

In the first 90 days after deployment, the City Council Records search received 3,720 search sessions — an average of 1,240 per month. Analytics showed that 71% of sessions resulted in a document click-through, indicating that residents were finding what they were looking for. The most common search categories were: specific property addresses (residents researching development history or neighbouring approvals), contractor and vendor names (residents and journalists tracking city contracts), budget-related searches (tax rate decisions, budget amendments), and service-related ordinance searches (utility rate changes, programme approvals).

Staff Capacity

The reduction in Open Records requests for already-public documents freed approximately 14 staff hours per week in the Clerk's office — time that had previously been consumed by searching for, retrieving, reviewing, and preparing responses for requests that could now be self-served. This was the capacity the City Clerk had been waiting for.

Response Time for Complex Requests

For the genuinely complex requests that remained — those requiring searches across multiple record systems or involving legal review and redaction — the Workplace Search deployment reduced the document discovery phase from an average of 3.5 hours to approximately 35 minutes. Average response time for complex Open Records requests fell from 12 business days to 4.8 business days. The Clerk's office met the statutory response deadline on 97% of requests in the 90-day post-deployment period, compared to 71% in the equivalent prior period.

The Proactive Disclosure Programme

Within 60 days of the deployment, the City Clerk's office launched the proactive disclosure programme that had been planned but perpetually deferred for lack of capacity. The programme identified seven recurring document categories that generated the highest volume of repeated Open Records requests: contract awards above $50,000, development variance approvals, budget amendments, zoning map amendments, employment separation agreements, capital project bid awards, and council committee reports.

For each category, a structured publishing workflow was established: documents were published to the City Council Records section within 10 business days of final approval, tagged with standardised metadata, and automatically added to the Keyspider index through the scheduled re-crawl. Residents searching for a contractor name, an address, or a project type would now find proactively disclosed documents alongside the historical council records — without filing a request.

The Clerk's office projected that the proactive disclosure programme would reduce recurring request volume by a further 30–40% over the following 12 months as the newly published categories became discoverable and residents learned they could self-serve them. The combination of AI search for discoverability and proactive disclosure for current documents was, in the Clerk's assessment, the complete solution — not just a search improvement, but a fundamental shift in the city's relationship to public records transparency.

"We went from a records programme that responded to requests to one that anticipates them. Residents can now find four years of council decisions in seconds. That's what transparency is supposed to look like."

— City Clerk