Portfolio
Case studies: four AI systems Tyler Kent built
Tyler Kent led delivery of a public healthcare data platform and designed and built the four AI systems that support it. Each case study shows a kind of system Kent Applied builds for clients, on the client's own data and at a fixed price. Identifying details are withheld.
Case study 01
Tessera: daily autonomous research for a public data platform
The problem
A public data platform goes stale without a constant stream of new insights. Producing them normally means analysts spending hours a week scanning news, journals, and datasets for material, and the scanning is the first task dropped when other work gets busy.
What we built
Tessera is an autonomous agent that scans the public web every day for emerging topics in healthcare and aging, filters them against the platform's editorial priorities, and drafts possible insights, each with a claim, evidence, sources and a suggested visual. Each draft also comes with a proposed analysis: the methods, the content type (such as a chart, a data story or an article), the data in the platform's own data catalog to build it from, and where to get any data the catalog lacks. So the team starts each draft from a plan written in the platform's voice and built on data it already has.
Each draft is saved as a JSON object, a structured file that Emblema reads directly to build its visual. A person reviews every draft before anything goes public.
The result
Over its first three months, Tessera filed about 27 sourced signals a week from nearly 200 sources, and turned about 3.5 of them a week into new draft insights for the platform's content calendar, each saved in the same format with its sources and a proposed analysis attached.
scheduled scans, run unattended
every draft carries its claim, evidence and sources
each draft arrives with its methods, content type and the catalog data to use
Case study 02
Emblema: from written insight to finished infographic
The problem
Every insight worth publishing needs a visual, and every visual used to need a designer. The design queue was the bottleneck between a finished insight and publication.
What we built
Emblema turns a written insight into a branded, publication-ready infographic, with the right typography, palette, and data presentation, exported for web and print. It drafts from a Tessera insight on request or on a schedule. Analysts can also use it directly, and the tool applies the brand's typography and colors itself, so graphics don't go back and forth through design review.
The result
Visuals no longer wait in a design queue. The number of graphics can grow with the number of insights without adding designers, and every graphic is on-brand when it leaves the tool.
design-queue wait between insight and visual
typography and colors applied by the tool itself
analysts produce finished graphics themselves
Case study 03
Sentinel: continuous QA for a live public platform
The problem
A public platform with hundreds of interactive charts, plus data products and articles, can break without anyone noticing: a data refresh shifts an axis, a link dies, or a number stops matching its source. Manual spot checks catch some of it, late.
What we built
Sentinel checks the live platform against the production codebase and the published source data. Every morning it runs 46+ automated checks in 8 groups and writes a separate report for each audience, from developers to leadership. Tessera's insights and Emblema's visuals go through the same checks once they are live.
Sentinel connects to the team's ticket tracker, Jira. It opens a ticket for each finding, assigns it to its owner, and closes it when a later scan finds the problem fixed. It also points out ways to help search engines and AI assistants find and cite the platform, such as pages that should link to each other.
The result
Errors are caught within hours. The first runs caught a public page returning a server error and a broken dollar figure in a live calculator. Engineers get specific, reproducible tickets, and leadership gets a regular quality report.
automated checks, run every morning
each finding becomes a ticket that Sentinel opens, assigns and closes
dashboard, plus a separate report for each audience
Case study 04
Opus: a one-page weekly brief for leadership
The problem
Leadership needed a weekly read on progress across several products and on the decisions waiting for them. That picture was spread across a project tracker, team documentation, meeting notes, code repositories, a communications calendar and the live site. Putting it together by hand meant chasing people every week, and a long status report is the wrong format for a busy executive who reads updates on a phone between meetings.
What we built
Opus puts only what leadership needs on one page: where each product stands and which decisions are waiting on them, so leaders know at a glance. Every week it builds the brief, with expected delivery dates, from those same records plus what Tessera, Emblema and Sentinel did. Leaders can trace every figure back to its source and open any product for its full story when they want the detail.
Opus is also an app. Leaders can install it on a phone or open it at a web link on a computer, and each brief arrives on a schedule by whichever route a leader prefers: a phone notification, email, Slack or Teams.
Anything taken from a meeting or a team note is labeled as reported, so leadership can tell it apart from figures pulled from a system. Team members can also send in an update for the next brief. When a source is unavailable, Opus notes the gap and builds the brief from the other sources. It publishes only a complete brief, so leadership never opens a half-finished page.
The result
Leadership gets one page a week on progress and open decisions, on the phone in their pocket or at their desk. It is built from records the team already keeps, so nobody has to chase people for updates.
brief, built unattended and pushed to each leader's phone, email, Slack or Teams
for leadership, in an app on their phones, with each product's full story a tap away
kinds of team records read each week, plus the other three systems' output
How they fit together
Tessera
Finds the insight
Scans the public web every day and saves each finding in one structured format, with its sources, a suggested visual and a proposed analysis built on the platform's data catalog.
Emblema
Makes the visual
Reads that same file and renders the on-brand infographic, on request or on a schedule.
Sentinel
Audits it live
Checks every page, number, link and visual on the live platform every morning, then opens a ticket for each finding and assigns it to its owner.
Opus
Briefs leadership
Reads the week's records and what all three did, then writes the one-page brief: what shipped, what is in motion, what needs a decision. Leaders get it in an app on their phones or by email, Slack or Teams.
How we plan client builds
The gains add up with every tool
Opus came last. It runs on records that already existed: what Tessera, Emblema and Sentinel produce, plus the team's own tracker, documents and notes.
Client builds are planned the same way. Each tool leaves behind records the next one can use, such as a file of findings, a log of what changed or a list of open decisions. The second tool starts with data the first one already collects, so it costs less to build and is useful sooner.
That is where AI pays off over time. On its own, one tool does one job. When the tools share their work, each new one starts from what the others already know, and the savings grow with every build.
- First toolDoes one job and keeps clean records of it
- Second toolStarts from the first tool's records
- Third toolStarts from both, so it needs the least new work
Built for your organization
We build tools like these around your own workflows and data, for a fixed price plus a monthly support fee. Two tools on our menu work the same way as these systems: the Weekly Leadership Brief and the Website, Reviews and Social Desk. Custom AI Tool builds one for any other job.
Book a 20‑minute call
Tell us about the workflow. We'll tell you what a build would involve, what it costs, and how long it takes.
Prefer to skip the call? Get a written proposal
Or email tyler@kentapplied.com, or connect on LinkedIn.